Evidence for a Kaili-Wolio branch of the Celebic languages

THIS IS AN EARLIER DRAFT, THE PUBLISHED VERSION OF THIS PAPER CAN BE FOUND AT:

 

The Kaili–Wolio branch of the Celebic languages, Oceanic Linguistics 59(1/2), June/December 2020.

1. Introduction

In this paper, I want to present some evidence for a subbranch of the Celebic macrogroup of Austronesian languages combining languages which were assigned to two different subgroups of Celebic in previous classifications, viz. Kaili-Pamona and Wotu-Wolio (Barr et al. 1979, Mead 2003).

The Celebic macrogroup comprises most languages of the provinces of Central Sulawesi and Southeast Sulawesi, and also a number of languages in the provinces of West Sulawesi and South Sulawesi. The most prominent members are Kaili, Pamona, Tolaki and Wolio. Celebic was convincingly established as a macrogroup by David Mead (2003), but had been proposed in earlier papers by René van den Berg. Mead has proposed the following internal classification of Celebic:

1. Tomini-Tolitoli
2. Kaili-Pamona
3. Wotu-Wolio
4. Eastern Celebic
a. Saluan-Banggai
b. Southeastern Celebic
i. Bungku-Tolaki
ii. Muna-Buton

The special position of Ledo among the Kaili lects and a probable relation to Wotu and Wolio was first recognized by S.J. Esser (1934). Similar observations are repeated by Mead (2003) and Donohue (2004) for Wotu-Wolio and Kaili-Pamona as whole, and most specificly by van den Berg (2008), who arrived at the conclusion that Wolio has to be included within Kaili-Pamona.

By assigning the languages of Wotu-Wolio and Kaili-Pamona to a single branch of Celebic, we arrive at a slightly simplified classification:

1. Tominic
2. Kaili-Wolio
3. Eastern Celebic
a. Saluan-Banggai
b. Southeastern Celebic
i. Bungku-Tolaki
ii. Muna-Buton

Although it is beyond the scope of this paper, there is some evidence that Tolitoli has to be excluded from Tominic, and even from Celebic as a whole. Tominic appears to be more closely related to Eastern Celebic than to Kaili-Wolio.

The language data used in this paper comes from the following sources: Adriani 1931, Adriani and Esser 1939, Anceaux 1952, Barr et al. 1979, Esser 1934, Grimes and Grimes 1987. I have collected own additional data from native speakers of the following languages: Wolio, Loa', Wotu, Ledo, Common Kaili (Rai isolect), Pamona. The initial draft of this paper was written in 2004, and I have only recently been aware of René van den Berg's 2008 article. I am thankful to René who kindly provided me with a copy of it.

2. Languages and internal grouping

The Kaili-Wolio branch as I define it comprises the following languages:

- Wolio, Kamaru
- Loa' (Barang-Barang), Kalao
- Wotu
- Ledo
- Common Kaili (i.e. Kaili minus Ledo)
- Topoiyo
- Sedoa
- Lindu
- Kulawi
- Uma, Sarudu
- Baras
- Pamona
- (Mamuju)

Excluded are however the Badaic languages (Bada, Besoa, Napu) and Rampi, which were part of Kaili-Pamona in earlier classifications. Based on their phonological history, these languages have to be excluded from Kaili-Wolio, and even from Celebic as whole. I will dedicate a separate paper to the position of Rampi and the Badaic languages (http://ezlinguistics.blogspot.de/p/seko-badaic.html).

Kaili-Wolio can be divided into a Northern and a Southern branch, which largely coincide with Kaili-Pamona (minus Badaic) and Wotu-Wolio:

Northern: Pamona, Common Kaili, Topoiyo, Sedoa, Lindu, Kulawi, Uma, Sarudu, Baras.
Southern: Ledo, Wotu, Island subbranch (= Loa', Kalao, Wolio, Kamaru); Mamuju.

Ledo occupies a special posisition, since it appears to be a mixed language composed of a Southern and a Northern stratum. The bound morphology and basic grammatical structure is clearly Northern and basically identical to that of the other lects usually classified as "Kaili", whereas the bulk of the lexicon has characteristic Southern features. All remaining Kaili lects, called "Common Kaili" here, have a Northern lexicon and phonology.

Within Southern Kaili-Wolio, the four languages Loa', Kalao, Wolio, Kamaru form a distinct subbranch, which I will call Island Kaili-Wolio here.

The inclusion of Mamuju may appear as a surprise here, since this language is generally considered a member of the South Sulawesi subgroup. While this is basically correct, nevertheless Mamuju clearly contains a lexical stratum which is of Kaili-Wolio origin. Evidence for this will be presented below.


3. Evidence for Kaili-Wolio

3.1. Merger of *ə and *a in the ultimate syllable

The most distinguishing feature of Kaili-Wolio languages - which sets them apart from all other Celebic languages - is the complete unconditioned merger of PMP *ə and *a in the final syllable.

First we look at the reflexes where the ultimate vowel is in absolute final position or followed by any consonant but PMP *R, *j or *s.

If the vowel in the preceding penultimate syllable is not *ə, then ultimate *ə and *a are uniformly reflected as a in all Kaili-Wolio daughter languages. In the following examples, Northern Kaili-Wolio is representd by Pamona and Southern Kaili-Wolio by Loa'.

PMP
*mata
*qabaRa
*anak
*tanəq
*tazəm
*gatəl
*kudən
*maqitəm

PKW
*mata
*awaa
*anaˀ
*tanaˀ
*taja
*kakata
*kura
*maeta

Loa'
mata
awaa
ana
tana
-tada
-kakkata
kura
metta

Pamona
mata
awaa
ana
tana
-taja
-kakata
kura
maeta

The same merger of PMP *a and *ə into PKW *a is also observed if preceded by *ə in the penultimate syllable. The reflexes are not uniform though, since Southern Kaili-Wolio has preserved *a unchanged, whereas the Northern Kaili-Wolio lects have assimilated the ultimate vowel to the reflex of the preceding penultimate *ə, viz. o The reflexes of *ə in penultimate position in the two branches will be discussed below in 4.1.

PMP
*dəngan
*dəpa
*təngaq
*əpat
*ənəm
*səjəm
*pəcəq
("rope")
("hungry")

PKW
*rənga
*-ndəpa
*təngaˀ
*əpaˀ
*əna
*səya
*pəsa
*kaləra
*-əraˀ

Loa'
ranga
-nrapa
tanga
apa
ana
sea
pasa
klara
-ara

Pamona
rongo
-dopo
tongo
opo
ono
sojo
poso
kaloro
-oro

< PCelebic *kalərən
< PCelebic *ərəp

Most likely, Proto Kaili-Wolio final *a (< *a/*ə) did not directly assimilate to *o, but rather to *ə at an earlier stage of (pre-)Proto Northern Kaili-Wolio. Thus PKW *əpaˀ first became *əpəˀ in pre-PNKW, with subsequent unconditioned change of *ə to PNKW *o, resulting in *opoˀ. This unusual vocalism is the most characteristic feature of the Northern branch of Kaili-Wolio.
Final *s, *j and *R had a coloring effect on the preceding vowel. In the case of final *s and *j, ultimate *a and *ə are uniformly reflected as e.

PMP
*panas
*pang-alas
*təngəs
*bəRkəs
*palaj
*laləj
*pusəj

PKW
*-pane
*pangale
*tənge
*boke
*pale
*lale
*puse

-pane (Loa', Wotu, Ledo, Common Kaili, Kulawi, Pamona)
pangale (Common Kaili, Kulawi, Wotu)
tange (Wolio, Wotu); tonge (Pamona)
boke (Wolio, Loa, Rai); boe (Wotu)
pale (Uma, Ledo, Common Kaili, Kulawi, Pamona)
lale (Wolio, Loa', Wotu, Ledo, Common Kaili, Kulawi); yale (Pamona)
puse (Wolio, Loa', Wotu, Ledo, Common Kaili, Pamona)

In the case of final *j, this is quite common in other Celebic languages too, since *j became *y in Proto Celebic (*palay, *laləy, *pusəy). Vowel coloring before *s is unique however among Celebic languages, and also among all other languages on Sulawesi.

With final *R, we have to distinguish again between roots which have *ə in the penultimate syllable and the all remaining cases. In the latter case, ultimate *ə and *a are generally reflected as o.

PMP
*luaR
*dipaR
*dataR
*ulaR
*bahaR
**lisəR

PKW
*-lua°
*dipa°
*rata°
*ula°
*pe-wa°
*lisa°

luo (Loa', Wotu, Ledo, Common Kaili, Kulawi)
dipo (Kulawi, Uma)
rato (Pamona)
ulo (Wolio, Loa', Wotu)
pewo (Ledo, Lindu, Kulawi)
liso (Loa'), liho (Kulawi)

Notes:
1. Ledo amd Northern Kaili-Wolio lects have ule meaning "snake", but this might well be a reflex of *quləj. Note that "caterpillar" in Wolio, Loa', Wotu is ulo-ulo, equating it to a small snake.
2. **lisəR (measure noun for small round objects) is reconstructed based on corresponding forms in South Sulawesi languages which point at Proto South Sulawesi *lisər.

This coloring effect of *R changing pre-Proto Kaili-Wolio *a (as merged from *a and *ə) to o is not found elsewhere on Sulawesi. In several other Celebic languages, *R merges with *y thus leading to vowel fronting. It is therefore also an important piece of evidence for the Kaili-Wolio subgroup. Yet, I do not reconstruct *o for the Proto Kaili-Wolio vowel, but rather a distinct final *a°, the reason for which will become clear if we now turn to roots with *ə in the penultimate syllable.

PMP
*dəngəR

*bəsaR

**əngəR

**kədəR

PKW
*(n)rənga°

*bəsa°

*ənga°

*kakəra°

rango (Wolio, Loa', Wotu)
ronge (Lindu), donge (Pamona)
-mbaso (Ledo)
-bose (Common Kaili), -bohe (Kulawi, Uma)
ango (Wolio, Loa', Wotu)
onge (Ledo, Common Kaili, Kulawi, Uma)
kakaro (Wolio), m-aro (Wotu)
-kangkore (Kulawi), -kakore (Pamona), -kore (Uma)

Notes:
1. *əngəR can be reconstructed for Proto Celebic (Proto Muna-Buton, Proto Saluan Banggai *ngooR, Proto-Tominic *enge).
2. *kədəR is based on Proto-Tominic *koroG, Proto Northern South-Sulawesi *kedder.

As with un-colored *a, South Kaili-Wolio languages have the same reflex for *a° as in all other correspondence sets, whereas Northern Kaili-Wolio languages have a fronted reflex of *a° following penultimate *ə, viz e. This shows that in Proto-Kaili-Wolio, final *a° still had a different quality from *o, and could still undergo some kind of assimilation to penultimate *ə in pre-Proto Northern Kaili-Wolio.

To sum up, we have three important pieces of evidence for the Kaili-Wolio subgroup based on reflexes of *a and *ə in the ultimate syllable:

1. PMP ultimate *a and *ə have merged to PKW *a;
2. PKW *a is raised and fronted to *e before earlier final PMP *s;
3. PKW *a undergoes raising and subsequent backing before earlier final PMP *R, although a separate development is found in Northern Kaili-Wolio in roots of the type CəCa°.

The merger of ultimate *a and *ə also occurred - indepedently - in several other parts of the Austronesian world. The most prominent example are the Malayic languages where this merger occurred in most lects although a minority still maintains the distinction. In Maluku, the merger is found in the following subgroups: Timor-Southwest Maluku (including the Babar languages), Tanimbar-Kei, Geser (including Kowiai), Bobot-Masiwang, and probably others as well. Still, it is unique in the Celebic context and on Sulawesi as a whole.


3.2. Further phonological evidence

The high vowels *u and *i remain unchanged either in final position of if followed by any consonant, including *R, *y and *s. This also includes the final sequence *-uy, which becomes *-u in Proto Kaili-Wolio.

PMP
*hapuy
*babuy

PKW
*apu
*bawu

apu (Loa', Wotu, Ledo, Common Kaili, Kulawi, Uma, Pamona)
bawu (Wolio, Ledo), bahu (Loa'), bau (Wotu)
wawu (Common Kaili, Kulawi, Uma)

This reflex is unusual but not unique though among Celebic languages, since it found also in Tukang Besi, Balantak and Saluanic.

Note that Kaili-Wolio languages do not at all take part in the raising of *i before final *q, which is found in all other Celebic languages

Reflexes of consonants in Kaili-Wolio languages provide no special evidence for Kaili-Wolio as a subgroup.

Kaili-Wolio has a tendency for contracting earlier vowel sequences into a single vowel, unlike most other Celebic languages, which tend to preserve these sequences (and even tend to create secondary like-vowel sequences).

PMP
*wahiR
*tuhud
*kau
*bahaq

PCel
*weeR
*tuur
*koo
*baaq

PKW
*uwe
*-tuˀ
*-ko
*-waˀ

In this context we can also cite the reduced reflex of PMP *qaləjaw, PCelebic *qələyo, which became *əyo in Proto Kaili-Wolio.

PMP
*qaləjaw

PKW
*əyo

eo (Wolio, Wotu, Ledo, Common Kaili, Kulawi, Uma, Pamona)
ajo (Loa')

In all other Celebic languages, *qələyo has a trisyllabic reflex.


3.3. Exclusively shared lexcial innovations

This is my tentative proposal for a list of ESI's in support of the Kaili-Wolio subgroup:

PKW
*iniˀ
*inosaˀ
*pangəne
*kapeoˀ
*kəro
*-rate/*-ndate
*tambe
*(mb)oreˀ

"sweat"
"breath"
"earlier"
"space under house"
"body, self"
"long"
"below"
"play"

iniˀ (Uma), ini (Wolio, Ledo, Common Kaili, Kulawi, Pamona)
inohaˀ (Uma), inosa (Ledo, Pamona), nosa (Loa'), inoha (Kulawi)
pangane (Loa', Ledo), ngone (Uma)
kapeo (Wolio, Loa', Ledo, Kulawi), kepeoˀ (Uma)
karo (Wolio, Loa', Ledo), aro (Wotu), koro (Common Kaili, Pamona)
-rate (Wolio, Loa', Pamona), -nrate (Wotu), -ndate (Ledo, Common Kaili)
tambe (Wolio, Loa', Wotu, Commom Kaili)
oreˀ (Uma), -ore/-mbore (Uma), mbore (Wotu, Loa'), more (Ledo)

This list is not very impressive, which is due to the fact that many shared innovations also appear in languages were strongly influenced by some Kaili-Wolio language, e.g. Badaic languages, Petapa or Rampi.


4. Further sound changes

4.1 Reflexes of PMP *ə

In 3.1. above, we have seen that *ə always merged with *a in the final syllable. PMP/PCel *ə was retained in PKW in penultimate position if the vowel in the ultimate syllable was *a/*ə and apparently also in case it was *i or *e (< *ay). This PKW *ə became *a in Proto Southern Kaili-Wolio and *o in Proto Northern Kaili-Wolio. Examples for the template *CəCa(C) are given already in 3.1. above. There are only few etyma with inherited ultimate *i or *e, but several internal cognate sets support the reconstruction of penultimate *ə, adding *o as ultimate vowel which can follow PKW *ə.

PMP
*qəli

*zəlay

---

---

---

---

PKW
*əli

*jəle

*ləre

*bəriˀ

*kəro

*təmo

ali (Wolio, Loa, Wotu, Ledo)
oli (Common Kaili, Kulawi, Pamona, Uma)
dale (Ledo), anrale (Ledo)
jole (Common Kaili, Pamona)
lare (Wotu, Ledo)
lore (Lindu, Kulawi, Pamona), role (Uma)
bari (Wolio, Loa')
woriˀ (Uma)
karo (Wolio, Loa', Ledo), aro (Wotu)
koro (Common Kaili, Lindu, Pamona)
tamo (Wolio, Loa', Wotu, Ledo)
tomo (Common Kaili, Kulawi, Pamona, Uma)

This reflex of *ə as a in Ledo, Wolio and Wotu was considered by van den Berg (2008) to be the result of external influence, viz. from Buginese. However, as we can see here and in 3.1. above, the reflex of *ə as a in Southern Kaili-Wolio is completely regular in the stated contexts.

If the ultimate vowel is *u, there are several cases where PMP *ə is uniformly reflected as o in all Kaili-Pamona languages; therefore I reconstruct PKW *o here. In all these cases, the root is disyllabic.

PMP
*bəsuR
*təbu
*qətut
*ləsung

PKW
*(m)bosu
*towu
*otuˀ
*no(n)su

mbosu (Wolio, Loa', Wotu), bosu (Ledo, Pamona), bohu (Kulawi, Uma)
towu (Wolio, Ledo, Common Kaili, Pamona, Uma), tohu (Loa'), tou (Wotu)
otuˀ (Uma), otu (Wolio, Loa', Wotu, Ledo, Common Kaili, Kulawi, Pamona)
nosu (Wolio), noncu (Kulawi, Pamona), nonju (Ledo, Common Kaili)

And yet, penultimate *ə in PKW could also cooccur with ultimate *u. Note that all three forms inherited from PMP are trisyllabic. PKW *ngəluˀ has external cognates on Sumba island (Kambera ngilu, Mamboru ngàlu).

PMP
*ta-təlu

*qapəju

*qitəluR

"wind"

PKW
*tatəlu

*apəyu

*Vntəlu

*ngəluˀ

tatalu (Ledo), talu (Wolio, Wotu)
totolu (Lindu), tolu (Common Kaili, Kulawi, Uma)
ampayu (Ledo), piu (Wolio)
apoju (Pamona), poju (Uma), ampou (Lindu), pou (Kulawi)
ntalu (Ledo), ontolu (Wolio)
ntolu (Common Kaili, Kulawi, Uma), toyu (Pamona)
ngalu (Wolio, Loa')
ngoluˀ (Uma), ngolu (Kulawi), ngoyu (Pamona)

4.2 Lenition of *b in Northern Kaili-Wolio

Medial *b was subject to lenition to *w already at the Proto Kaili-Wolio stage.

PMP
*qabaRa
*quban
*tubuq

PKW
*awaa
*uwa
*tuwuˀ

awaa (Wolio, Pamona)
uwa (Wolio, Ledo, Common Kaili, Kulawi, Pamona)
tuwuˀ (Uma), tuwu (Ledo, Common Kaili, Kulawi, Pamona), tuu (Wotu)

In initial position, Southern Kaili-Wolio languages have retained PKW *b unchanged (with a very few exceptions). Northern Kaili-Wolio languages however show a split: while for some etyma, *b was retained as a stop in all Northern Kaili-Wolio languages, in most cases however it was lenited to *w presumably in Proto Kaili-Wolio already.

PMP
*bulud
*balu
*baqəRu
*bəsuR
*bituqən

*bibiR
*bulan
*babuy
*bulu
*buku
*babaw
*buaq
*batu
**burokoˀ
**baluak

PKW
*buluˀ
*balu
*baəˀu
*(m)bosu
*betueˀ

*biwi
*bula
*bawu
*bulu
*buku
*bawo
*buaˀ
*batu
*burokoˀ
*buluaˀ

PNKW
*buluˀ
*balu
*boˀu
*bosu
*betueˀ

*wiwi
*wula
*wawu
*wulu
*wuku
*wawo
*wuaˀ
*watu
*wurokoˀ
*wuluaˀ

buluˀ (Uma), bulu (Common Kaili, Kulawi), buyu (Pamona)
balu (Common Kaili, Kulawi, Uma)
boˀu (Pamona, Uma), bou (Lindu, Kulawi)
bosu (Common Kaili, Pamona), bohu (Kulawi, Uma)
betueˀ (Uma), betue (Common Kaili, Kulawi, Pamona)

wiwi (Common Kaili, Kulawi, Uma, Pamona)
wula (Common Kaili, Kulawi, Uma), wuya (Pamona)
wawu (Common Kaili, Kulawi, Uma, Pamona)
wulu (Common Kaili, Kulawi, Uma)
wuku (Common Kaili, Kulawi, Uma, Pamona)
wawo (Common Kaili, Pamona)
wuaˀ (Uma), wua (Common Kaili, Kulawi, Pamona)
watu (Common Kaili, Kulawi, Uma, Pamona)
wurokoˀ (Uma), wuroko (Pamona)
wulua (Lindu, Kulawi), wuyua (Pamona)

The conditioning for this split remains opaque. An irregular split of *b in initial position is observed in most other subgroups on Sulawesi.

4.3 Proto Kaili-Wolio final glottal stop: evidence from Uma

Almost all Kaili-Wolio languages are “vocalic”, i.e. they do not allow final consonants, nor medial clusters other than ‑NC‑. The only two exceptions to this are Uma and Loa'. Uma has word-final glottal stops, e.g. ataˀ‘roof’, muliˀ ‘return’, while Loa' has final glottal stop and ng, as well as geminate consonants and medial clusters of the shape ‑ˀC‑. In the case of Loa', words with a "non-Kaili-Wolio-like" shape are clearly external loans, whereas the directly inherited main stratum of Loa' is fully vocalic. The glottal stop in Uma however occurs with etyma which have a characteristic Kaili-Wolio or Northern Kaili-Wolio vocalism, e.g. ataˀ with a as reflex for ultimate *ə or opoˀ with the characteristic assimilation of ultimate *a to o.

The final glottal stop therefore must be a retention from Kaili-Wolio, which was only preserved in Uma, probably as a contact-induced conservatism, since final glottal stop is also found in neighboring Bada'. The glottal stop in Proto Kaili-Wolio developed from earlier final stops such as *q, *k, *t, *p, *d, *b.

PMP
*qatəp
*əpat
*langit
*tuhud
*anak
*udahik
*manuk
*tanəq
*tubuq

PKW
*ataˀ
*əpaˀ
*langiˀ
*tuˀ
*anaˀ
*ureˀ
*manuˀ
*tanaˀ
*tuwuˀ

Uma
ataˀ
opoˀ
langiˀ
ko|tuˀ
anaˀ
m|ureˀ
manuˀ
tanaˀ
tuwuˀ

The immediate parent language of Proto Kaili-Wolio, viz. Proto Celebic, still had the (almost) full range of final consonants as inherited from PMP. For other branches of Celebic such as Bungku-Tolaki or Muna-Buton, which are fully vocalic now, we have conclusive indirect evidence that their respective proto-languages Proto Bungku-Tolaki and Proto Muna-Buton still retained final consonants. We therefore cannot not exclude that also Proto Kaili-Wolio still had more stop contrasts in final position, or even retained final nasals that were lost in all daughter languages. The only indirect evidence in Kaili-Wolio comes from the "thematic" consonants that appear between verb roots and suffixes such as *-i or *aka. While some forms retain the "correct" final consonant (e.g. Kulawi topohi, Wolio tapasi < *təpasi, cf. Proto South Sulawesi *təppas; PKW *insani cf. Proto South Sulawesi *insan), we find others which have a "wrong" thematic consonant, e.g. Kulawi pelihi < PMP *piliq, Ledo patesi, Kulawi patehi < PMP *patay.

Therefore I hesitate to draw a conclusion about the Proto Kaili-Wolio final consonant inventory at the current stage of research, which would then allow for internal or top-down reconstructions beyond the bottom-up reconstruction of PKW *ˀ.


5. Early contacts between Kaili-Wolio and Rampi

The earliest stratum of Kaili-Wolio loans into neighboring languages is found in Rampi, which shares some lexcial innovations with the Kaili-Wolio languages and has closely preserved the original shape of Proto Kaili-Wolio etyma.

Rampi'
ləka

ləri

wəriˀ

PKW
*ləka

*ləre

*bəriˀ

-laka (Wolio, Loa'), -laa (Wotu)
-loko (Lindu, Kulawi), -yoko (Pamona)
lare (Wotu, Ledo)
lore (Lindu, Kulawi, Pamona), role (Uma)
bari (Wolio, Loa')
woriˀ (Uma)

"fly"

"hinterland"

"many"

Note that Rampi has preserved *ə in the penultimate vowel without the assimilation of the following vowel as in the Northern Kaili-Wolio languages. The Rampi form ləka in fact confirms our reconstruction of the Proto Kaili-Wolio vocalic pattern *CəCa, even though it changed into CoCo in Northern and into CaCa in Southern Kaili-Wolio languages.

Another "time-capsule" form in Rampi is bəi "husked rice". This is actually an indirect reflex of PMP *bəRas, which must have entered Rampi via PKW *bəe, still preserved as bae in Wolio, Loa' and Wotu (no reflex in Northern languages). The raising of final *e to Rampi i is common and also found in all reflexes of the final PMP diphthong *-ay.

This evidence shows that Rampi speakers were in contact with Kaili-Wolio speakers at a very early stage, either at the Proto Kaili-Wolio stage, or shortly after the breakup before the daughter branches developed their own characteristic features.


6. The languages of the Southern branch

A remarkable feature of the Southern branch of Kaili-Wolio lies in the fact that all its members underwent some obvious strong influence from neighboring languages: Ledo shows convergence with the Common Kaili lects, Wotu is under heavy pressure from Buginese, Wolio has converged with Munic and Butonic languages, while Loa' contains serveral strata of external influences, the latest being from Selayar.

And yet the unity of the Southern branch is still clear from phonological and some lexical evidence. The most characteristic feature is the reflex of Proto Kaili-Wolio *ə as a in all daughter languages. Another uniform sound change is the shift of PKW *j (< PMP *z) to d, although the same shift occurred independently in Lindu, Kulawi and Sedoa.

Phonological innovations include the additional *w in the following two etyma:

PMP
*uRat
*wahiR

PKW
*uaˀ
*ue

PSKW
*uwa
*uwe

uwa (Wolio, Ledo), uha (Loa')
uwe (Wolio, Wotu, Ledo), uhe (loa')

Lexical innovations include:

PSKW
*berei
*kawa

"spouse"
"arrive, find"

berei (Wotu, Ledo), brei (Loa')
kawa (Wolio, Ledo), kaha (Loa'), awa (Wotu)

Northern languages use inherited *rongo (< *rənga "companion") and *rata here.

6.1. The position of Ledo

Ledo can be best described as a Common Kaili lect with a predominantly Southern lexcial stratum. The basic morphology and syntax of Ledo is more or less identical to that of i.e. Rai or Tara.

The mixed character of Ledo is easily visible in the reflexes of PKW *ə.

PMP
*qəli
*zəlay
*ta-təlu
*dəpa
*təngaq
*pəcəq
("heavy")
("hinterland")
("body/self")
*a-ənəm
*səjəm
**əngəR
**kalərən
("kunai grass")

PKW
*əli
*jəle
*tatəlu
*ndəpa
*təngaˀ
*pəsaˀ
*təmo
*ləre
*kəro
*aəna
*səya
*ənga°
*kaləra
*jənaˀ

PSKW
*ali
*dale
*tatalu
*ndapa
*tanga
*pasa
*tamo
*lare
*karo
*ana
*sea
*ango
*kalara
*dana

Ledo
ali
dale
tatalu
-ndapa
tanga
pasa
tamo
lare
karo
aono
soyo
onge
kaloro
jono

PNKW
*oli
*jole
*tatolu
*ndopo
*tongoˀ
*posoˀ
*tomo
*lore
*koro
*aono
*soyo
*onge
*kaloro
*jonoˀ

In its reflexes of PKW *j and initial *b, Ledo predominantly shows reflexes of the Southern stratum.

PMP
*zəlay
*zilaq
*tazəm
*quzan
*zalan
**zuəl
("kunai grass")

PKW
*jəle
*jilaˀ
*taja
*uja
*jala
*jua
*jənaˀ

PSKW
*dale
*dila
*tada
*uda
*dala
*(n)dua
*dana

Ledo
dale
dila
-tada
uda
dala
-dua
jono

PNKW
*jole
*jilaˀ
*taja
*uja
*jala
*jua
*jonoˀ

PMP
*bibiR
*bulan
*babuy
*bulu
*buku
*babaw
**baluak
*buaq
*batu
**burokoˀ

PKW
*biwi
*bula
*bawu
*bulu
*buku
*bawo
*buluaˀ
*buaˀ
*batu
*burokoˀ

PSKW
*biwi
*bula
*bawu
*bulu
*buku
*bawo
*bulua
*---
*batu
*buroko

Ledo
biwi
bula
bawu
bulu
buku
bawo
bulua
wua
watu
wuroko

PNKW
*wiwi
*wula
*wawu
*wulu
*wuku
*wawo
*wuluaˀ
*wuaˀ
*watu
*wurokoˀ

There are a couple of Southern-Northern lexical isoglosses involving retentions from a higher level, where Ledo goes with the Southern languages.

"eat"
"yesterday"
"know"

PSKW
*kande
*(r)i awi
*sani

Ledo
kande
riawi
sani

PNKW
*koniˀ
*i wengi
*insani

*kande is found in Badaic, Seko and in the South Sulawesi languages. PSKW *awi derives from PMP *Rabiˀi, PNKW *wengi from PMP *bəRngi. PSKW *sani is shortened from PKW *insani, which is related to Badaic inca and Proto South Sulawesi *insan plus a suffix *-i.

With most important functors however, Ledo goes along with its Common Kaili neighbors.

"have + noun"
"still"

PSKW
*ko-
*-po

Ledo
no-
*-pa

PNKW
*no-
*-pa

Historically, we can assume that Southern Kaili-Wolio speakers came to the Palu area and basically gave up their original speech, replacing it with a local Northern Kaili-Wolio lect, but retaining a good part of the original Southern Kaili-Wolio lexicon. After this original formation process as a mixed language, Ledo became the prestige language of the area and in turn exericed some influence on the neighboring Common Kaili lects.


6.2. The emergence of Island Kaili-Wolio

Wolio and Loa' display a few common features which clearly must have emerged due to a short but significant period of common development from Proto Southern-Kaili-Wolio.

The most important feature is the development of a Buton-type system of person marking, which does not distinguish between transitive and intransitive verbs. Here, every subject is expressed by pronominal prefix, e.g. Wolio ku-pesua, Loa' ku-pəsua "I enter". This is contrast with Wotu minsua=u which does not employ a pronominal prefix. The construction itself probably was not unknown to Proto Southern-Kaili-Wolio, since in the Northern language Pamona, ku-pesua appears as a dependent construction for intransitive verbs. Its use however as the only main construction for intransitive verbs must be due to convergence of the common ancestor of Wolio and Loa' to the indigenous languages of Buton island.

An important functor innovation of Island Kaili-Wolio is Wolio opea, Loa' apea, presumably from PKW *apa + third person clitic *-ia.

The common period of Island Kaili-Wolio did not last too long, and speakers of Loa' sailed off to Selayar before the lexical impact of Buton languages on the Kaili-Wolio "intruder" started. Wolio subsequently underwent a very strong lexical influence from both Munic and Butonic languages, i.e. pebaho "bathe" is a loan from Butonic, whereas seli "dig" comes from a Munic source. After the initial convergence phase, Wolio in turn exercised a strong influence on all languages of the Muna-Buton area, as it was the dominant language of the Wolio court in Baubau.

On Selayar, the Loa' language came under strong influence from Selayarese. The influx of loans from Selayarese (and other languages as well) radically altered the phonological structure of Loa', which now allows closed medial syllables and final stops.


6.3. The Southern Kaili-Wolio stratum of Mamuju

Mamuju is usually classified as a language of the South Sulawesi subgroup, and within this grouping, it appears most closely related to Mandar, Yet, in the Mamuju wordlist presented by Grimes & Grimes (1987), there are six entries which betray a stratum of non-South Sulawesi origin. Three of these have a clearly Southern Kaili-Wolio vocalism:
.

PMP
*ulaR
*dəngəR
**kədəR

PSKW
*ulo
*rango
*karo

Mamuju
ulo
ranngo
karro

PNSSul
*ular
*rəngngi
*kedder

The other three entries contain two inherited numerals, which nevertheless clearly must be re-borrowed, since they were replaced by innovative forms in all South Sulawesi languages.

PMP
*wa-walu
*sa-siwa
**sema

PSKW
*uwalu
*sasio
*sema

Mamuju
uwalu
sasio
sema

PNSSul
*karuaˀa
*kamesaˀa/kaseraˀa
*minna/*nai

There are several Kaili-Wolio languages spoken not too far from Mamuju, e.g. Topoiyo, Baras, Sarudu'. These belong the Northern branch however and can therefore be excluded as the source at least for the first three etyma. So where does this Kaili-Wolio stratum in Mamuju originate from? At the current stage, we can only speculate that Southern Kaili-Wolio speakers migrated to Mamuju at the same time when they entered the Palu Bay as pre-Ledo speakers.


7. The prehistory of Proto-Kaili extension

The present-day distribution of Kaili-Wolio languages is characterized by a contigious speech area for the Northern branch (although disrupted in the interior by the Badaic speech area), and widely dispersed "islands" in other speech areas in the case of the Southern branch. Before discussing the origins of the Kaili-Wolio subgroup as a whole, I will first outline two respective scenarios for the expansion of the Northern branch and the Southern branch.

7.1. Northern Kaili-Wolio

As a working assumption, I propose that the homeland of the Northern Kaili-Wolio languages lay in the current Pamona speech area, most probably in the area between Tentena and Poso. From there, PNKW speakers moved westwards, close along the coast to the area between Palu and Parigi. Moving to the interior, they encountered Badaic speakers, interacting with them at the western edge of the Badaic area. It is for this reason that Kulawi and Uma underwent some influence from Badaic, Kulawi mostly in its lexicon, Uma even more strongly in lexicon and basic structure (including e.g. loss of the Proto Northern Kaili-Wolio tense system). From the coastal area, Common Kaili spread to the west and southwest, and to the north. The current range of the Pamona speech area is the result of a southward and eastward expansion from a core area which must have been quite identical to what I propose for the Northern Kaili-Wolio homeland.

7.2. Southern Kaili-Wolio

The most likely source of origin of Southern Kaili-Wolio dispersion is the Wotu area and its northern surroundings, long before the area became pressured by Luwu' Buginese and Pamona. Early neighbors may have included Rampi', Limola and Bungku-Tolaki languages. Wotu myths claim that several kingdoms on Sulawesi were founded by Wotu emigrants, including Wolio (Bau-Bau on Buton island) and Palu (seat of the Ledo kingdom). The Buginese myth of Sawerigading (part of the La Galigo epic) actually may represent a Buginese adaption of earlier Wotu-based Southern Kaili-Wolio traditions. The similar pattern of language convergence found in the Palu and Wolio area indicates a parallel historical background, most propably based on a single expansion movement out of Wotu as mirrored in the Sawerigading myth. Southern Kaili-Wolio speakers must have been superior in terms of weaponry, which is compatible with the early history of Luwu, and certainly took advantage of a low degree of political organization beyond the village level in those areas into which they penetrated.

7.3. The common Kaili-Wolio "homeland"

The actual homeland of Proto Kaili-Wolio before its breakup into the two main branches must have been located somewhere between Poso and Wotu. Based my observations on early contacts between (pre-)Rampi and Proto-Kaili-Wolio, a likely candidate is the area around Lake Poso. From there, Proto Northern Kaili-Wolio expanded northwards, while Southern Kaili-Wolio moved southwards to the Bone Gulf.

References

Adriani, N. 1931. Spraakkunst der Bare'e-Taal. (Verhandelingen van het Koninklijk Bataviaasch Genootschap van Kunsten en Wetenschappen, LXX.) Bandoeng: A. C. Nix.

Adriani, N. and Esser, S. J. 1939. Koelawische taalstudien. (Bibliotheca Celebica, I, II, III.) Bandoeng: A. C. Nix.

Anceaux, Johannes Cornelis. 1952. The Wolio Language: Outline of Grammatical Description and Texts. Rijksuniversiteit te Leiden. 11. 's-Gravenhage.

Barr, Donald F. and Barr, Sharon G. and Salombe, C. 1979. Languages of Central Sulawesi: checklist, preliminary classification, language maps, wordlists. Ujung Pandang: Hasanuddin University.

Bowden, John and Nikolaus P. Himmelmann, 2004. Papers in Austronesian subgrouping and dialectology. Canberra: Pacific Linguistic 563.

Donohue, Mark, 2004. The pretenders to the Muna-Buton group. In Bowden and Himmelmann (eds.), 2004: 21-36.

Esser, S.J., 1934. Handleiding voor de beoefening der Ledo-taal. Bandoeng: Nix & Co.

Grimes, Charles E. and Grimes, Barbara D. 1987. Languages of South Sulawesi. Canberra: Pacific Linguistics: Series D, 78.

Lander, Yury A. and Ogloblin, Alexander K. (eds.) 2008. Language and Text in the Austronesian World: Studies in honor of Ülo Sirk, 89-113. München: Lincom.

Lynch, John (ed.) 2003. Issues in Austronesian historical phonology. Canberra: Pacific Linguistics 550.

Mead, David. 2003. Evidence for a Celebic supergroup. In Lynch, John (ed.), 2003: 114-141.

van den Berg, René. 2008. Notes on the historical phonology and classification of Wolio. In Lander, Yury A. and Ogloblin, Alexander K. (eds.), 2008: 89-113.

Woensdregt, Jac. 1929. Rampi'sche verhalen venaald en van aanteekeningen voorzien. Tijdschrift Bataviaasch Genootschap 69. 254-319.