An Oceanic Origin for Aiwoo, the Language of the Reef Islands?

Ross, Malcolm, Naess, Ashild, Oceanic Linguistics

Whether the languages of the Reefs-Santa Cruz (RSC) group have a Papuan or an Austronesian origin has long been in dispute. Various background issues are treated in the introductory section. In section 2 we examine the lexicon of the RSC and Utupua-Vanikoro languages and show that there are regular sound correspondences among these languages, and that RSC languages display regular reflexes of Proto-Oceanic etyma and are therefore Austronesian. We also show that together the RSC and Utupua-Vanikoro languages form an Oceanic subgroup, which we label "Temotu," and that the Temotu group is probably a first-order subgroup within the Oceanic family. In section 3, we examine a variety of constructions and morphemes in Aiwoo, the language of the Reef Islands, to see whether they have plausible Oceanic sources. The answer in most cases is that they do. This is important, as several of these constructions have in the past been given as evidence that the RSC languages have a Papuan origin. We conclude that the RSC languages are Austronesian and that there is no need to posit a Papuan element to explain their origin.

1. INTRODUCTION. The languages that we are concerned with in this paper are located in the Temotu Province of the Solomon Islands. Santa Cruz Island is roughly 390 km from Makira in the main Solomons archipelago to the west and about 270 km from the northernmost Torres Islands of Vanuatu to the south (see map 1). (1) The languages of the area are listed in (1) together with alternate names and abbreviations. (2) Their approximate locations are shown on map 2. The language or languages of Santa Cruz Island (also known as Nedo or Deni) form a chain of dialects. We refer here simply to two for which Tryon and Hackman (1983) supply lexical data, Malo (here labeled Natugu) and Nanggu (here Nagu). Omitted from (1), but included in map 2, are the three Polynesian languages Vaeakau-Taumako (= Pileni), Anuta, and Tikopia.

(1) * Reefs-Santa Cruz group

--Aiwoo (= Reefs) [AIW]

--Natugu (= Malo, Lodai, Nedo) [NAT] (3)

--Nagu [NAG]

* Utupua

--Nebao (= Aba) [NEB]

--Asuboa[ASU]

--Tanibili [TNB]

* Vanikoro

--Buma (= Teanu) [BUM]

--Vano (= Vana) [VNO]

--Tanema (= Tanima, Tetau) [TNM]

There are two Temotu historical topics that we will not consider here. One is contact between Polynesian and Temotu languages (see Naess and Hovdhaugen 2007). Vaeakau-Taumako certainly, and perhaps Anuta and Tikopia as well, have left their mark at least in the form of borrowings on Temotu languages. The second topic is the amazing diversity of the Temotu languages. This is less surprising when one considers the distances between them--the Reef Islands are some 70 km north of Santa Cruz, for example--but distance does not explain the diversity Francois (2006) finds among the languages of Vanikoro alone.

Instead, our focus is on the genealogy of the languages in (1) and especially upon Aiwoo. The UV languages have been taken to be Oceanic (Tryon and Hackman 1983, Tryon 1994). The genealogy of the RSC group has been questioned ever since linguists have had any knowledge of them (i.e., since Codrington 1885), but our starting point here will be two oft cited papers read at the Second International Conference on Austronesian Linguistics in January 1978 and published as Lincoln (1978) and Wurm (1978). The authors take opposing views. Lincoln (1978:929) begins his paper: "The main theme of this paper is that the Reefs-Santa Cruz (RSC languages could be classified as Austronesian--or more specifically as Oceanic languages--free from the influence of other language families in the Pacific." Wurm (1978:971), on the other hand, suggests that the RSC languages are descendants of "a non-Austronesian language or languages and that they have incompletely taken over an Austronesian language:" He goes on to suggest that there have been several different Austronesian inputs.

[ILLUSTRATION OMITTED]

We quote Lincoln's opening sentence because others have read him as saying that RSC languages show signs of Papuan contact (Tryon 1994:613; Dunn, Reesink, and Terrill 2002:41), but in the sentence cited above Lincoln specifically eschews Papuan influence on the RSC languages. His position is more radical than his colleagues have acknowledged, and might have received more attention if its radicalism had been recognized. He was not the first radical to take up the cudgels against Wurm's view of RSC as Papuan, however. To conclude his paper, Lincoln quotes the closing sentence of La Fontinelle (1974) who, after presenting a brief phonology and grammar sketch of the Neo dialect of Santa Cruz, notes how similar it is to languages in Vanuatu and New Caledonia and asserts: "... et, en conclusion de cette premiere etude du Neo, nous optons, au contraire, pour le rattachement de ce dialecte au groupe des langues melanesiennes." (4) Our position is similar to La Fontinelle's and Lincoln's: the RSC languages--and UV--are Oceanic, and it is unnecessary to posit a Papuan element in their histories.

[ILLUSTRATION OMITTED]

Wurm argues that the RSC languages are Papuan but have undergone contact-induced change through the presence of Oceanic languages. He argues this on two grounds: (i) there is a large proportion of Oceanic vocabulary in the RSC lexica, but regular sound correspondences with other Oceanic languages cannot be established, so the Oceanic words must be loans; (ii) there are grammatical features in the RSC languages that he finds atypical of Oceanic but present in East Papuan languages. These features are principally a variety of gender or noun classifying devices and the agglutinative verb complexes with their many components.

In a review of possible RSC alternative histories, Clark (1999) shows that Natugu and Nagu reflect Oceanic lexical items with signs of a regular phonological history and reflexes of Oceanic bound morphemes, and gives reasons for believing that there are no statistically significant grounds for positing the presence of Papuan lexicon in Natugu. He remains agnostic on the question of whether Oceanic-Papuan contact forms part of the history of the RSC languages.

More is known today than was known in 1978, both about the Temotu languages and the history of Oceanic and about the typology of language contact situations. A source for Temotu as a whole is Tryon and Hackman's (1983) wordlist collection. For UV we also have Tryon's short grammar sketches (Tryon 1994, 2002). For Santa Cruz we have only the data cited by Wurm (1976, 1978), Lincoln (1978), and Tryon (1994), all of it collected by Wurm, together with Natugu pronoun paradigms and some verbal morphology from Brenda Boerger (pers. comm.). For Aiwoo, however, we have a goodly quantity of material collected in the field and analyzed by the second author of this paper (Naess 2005, 2006a, 2007a, 2007b) and by Frostad (2006). (5)

Thanks to these materials and to an improved knowledge of Oceanic lexicon and POc phonology, it has been possible to recognize Temotu reflexes of POc etyma and to establish sound correspondences for the Temotu languages, including RSC (2.2-2.3). This answers one of Wurm's objections to classifying them as Oceanic. We have been able to go further, however, and to show that they constitute a single group, that this group probably has no members outside the Temotu area, and that the group is probably a primary subgroup of Oceanic (24). This is counterevidence to a recent archaeological claim that speakers of the Southeast Solomonic languages have their origins in RSC (Sheppard and Walter 2006).

Wurm's arguments for Papuan contact in Temotu linguistic history are partly typological. Post-1978 contact studies provide some clarity on the significance of typology for diagnosing contact. (6) It is bound forms that reflect the genealogy of a language. Syntax changes relatively easily as a consequence of contact. A good deal of Aiwoo (and for that matter Utupua) bound morphology is Oceanic, pointing to an Oceanic pedigree. But with an increased knowledge of Oceanic typology? and Naess's typologically informed description of Aiwoo grammar, some of Aiwoo's alleged peculiarities turn out to be less peculiar than they seemed to Wurm. Its noun-classifying devices (3.2) and verb complexes (3.3) can be plausibly derived from early Oceanic grammatical constructions. We take it that what is true of the history of Aiwoo is almost certainly also true of the history of the Santa Cruz languages, but this, of course, remains to be demonstrated.

1.1 GENEALOGICAL ISSUES. Lurking behind the topic of the previous paragraph is a larger issue, namely, what does it mean to say that a language is Oceanic or that it is Papuan? We take the comparative linguist's conventional position: it makes sense to say that a language is Oceanic if a group speaks a language that either (a) has been passed from generation to generation since Proto-Oceanic (POc) times, or (b) has been acquired by a community's shift to a language of which (a) is true. This implies no rejection of contact induced change. Generational continuity does not exclude the bilingualism that engenders contact induced change, and--much more rarely--shifting speakers may retain features from their former language.

What does it mean to say that a language is Papuan? "Papuan" is a residual category, a label for all non-Austronesian and non-Australian languages spoken in a region centered on New Guinea but stretching from Timor to the Solomons (except for languages that have entered the region since European contact). Recent work suggests that Papuan languages belong to more that twenty genealogical groups (Ross coos), and if we want to say that a language is Papuan, we accordingly need to name the group to which it belongs. Wurm suggests that the alleged Papuan elements in Temotu languages are East Papuan (although he trawls further afield at times), but the integrity of East Papuan is also controversial, as it seems to consist of eight genealogical groups, none obviously related to any of the others (Ross 2001, Dunn, Reesink, and Terrill 2002).

Perhaps the indefiniteness of the term "Papuan" has contributed to an unhappy practice among Oceanic linguists of attributing to unnamed "Papuan" sources features of Oceanic languages that cannot be readily attributed to their ancestry. Capell (1943) was an early adopter of this practice, regarding the Oceanic languages of New Guinea as "pidgins" and dividing them up according to their alleged Papuan elements. Most recently Blust (2005) has done "a surprising thing;" as Pawley (2006:243) puts it: he argues for the presence of Papuan speakers in Vanuatu and New Caledonia in order to account for allegedly non-Oceanic typological features in the languages of the region. This proposal flies in the face of relevant archaeological evidence, and Pawley (2006) has argued in some detail against Blust's position, which is based on biological and cultural, as well as linguistic, arguments. Pawley argues that there have never been Papuan speakers in southeast Melanesia (which he defines as the islands of the Temotu Province, Vanuatu, New Caledonia, the Loyalties, and the Fiji group). The essence of Pawley's linguistic argument is that the typological features that Blust regards as Papuan can be accounted for as Oceanic. One of them is the presence of serial verb constructions, which are reconstructible in POc (Ross 2004a). (8)

The Temotu area lies on the northern fringe of Pawley's southeast Melanesia. One of Blust's assumptions is that the RSC languages are Papuan, and implicitly that the Temotu area served as a jumping-off point for Papuan speakers to reach the rest of southeast Melanesia. If there are no Temotu Papuan languages, Blust's scenario becomes yet more unlikely. If one looks at a map of the Pacific, of course, the distance between Santa Cruz and the easternmost Papuan language, Savosavo, looks small. But it is 680 km in a straight line from Santa Cruz to Savo, 390 of them across open ocean.

The archaeology indicates that the Reef and Santa Cruz islands had been settled by Lapita people around 3200 BP and that this was the first human settlement there (Green 2003). This makes it one of the very earliest Lapita settlements outside the Bismarcks. One would expect the whole of northwest Melanesia from the Bismarcks to the eastern end of the Solomons archipelago to have been settled before this, but on present evidence this does not seem to have occurred. A At the earliest the New Georgia group in the western Solomons shows signs of settlement 300-400 years after the Reefs and Santa Cruz (Felgate to appear). The apparent lack of early Lapita settlements in the Solomons has caused Sheppard and Walter (2006) to propose that Lapita settlers somehow leapfrogged the Solomons to arrive in the Temotu area. It is not our brief to engage in archaeological argument, but if the Temotu languages form a primary subgroup, then this suggests that the arrival of their ancestors was separate from the arrivals of either the Southeast or Northwest Solomonic groups in the areas they currently occupy.

Given the early settlement of the Temotu group and their isolation, it is reasonable to suggest that the Temotu languages, including RSC, are Oceanic languages that owe their oddities to developments that have occurred independently of the rest of Oceania.

1.2 ORTHOGRAPHY. The phonemes b, d, g are prenasalized throughout UV, on Santa Cruz, and in the northern dialects of Aiwoo. In the practical orthographies of Aiwoo and Santa Cruz, /n/ and /[eta]/ are written as ny and ng, respectively, but we have used n and [eta] in this project to facilitate searching in our database. In Aiwoo the pairs /n/~/n/,/t/ ~/s/, and /d/~/j/ are neutralized before /i/ and are here written n, s, and j, respectively; they are distinct elsewhere.

RSC vowel phonemes and their representation in the practical orthographies used by Wurm for Santa Cruz and Naess for Aiwoo are shown in (2). Parenthesized vowels occur only in Santa Cruz languages.

(2) i      (u)[u]  u
    e      (e)[e]  o[o]
    d [ae] (o)[b]  a[D,V,C]
            a

Wurm (1978:969) uses [a.bar] instead of a and gives its phonetic value as [C,D]. He gives the phonetic value of e as [[LAMBDA]].

The languages of Vanikoro have a five-vowel system, /i e a o u/ (Alexandre Francois, pers. comm.). This is probably also true of the languages of Utupua, but we do not know this for certain and thus retain the phonetic distinctions in Tryon and Hackman's wordlists, using the RSC orthography to represent them.

2. TEMOTU HISTORICAL PHONOLOGY Some initial sense of the degree of relationship among the Temotu languages and with other Oceanic languages can be gained by an examination of the POc and Temotu numerals from 1 to 9 in table 1. Once one recognizes that the disyllabic form of each numeral from 2 to 9 has been truncated by the loss of its second syllable and that in most cases the resulting monosyllable is preceded by another syllable that was once a prefix, then some sense of systematicity emerges in table 1. The monosyllables display regular sound correspondences (cf. table 2). Furthermore, the numerals from 1 to 4 allow us to see which languages use additive (Aiwoo, Natugu, Nebao, and Tanibili) or subtractive (Nagu) forms--shown in italics--for the numerals 6 to 9. Asuboa, Buma, Vano, and Tanema have full decimal systems.

Closer investigation reveals more Oceanic features. In Aiwoo the numbers 6-9 are formed with a morpheme polV-, perhaps cognate with Bali-Vitu and Bola polo-. The prefixed syllables probably reflect earlier numeral classifiers. The general POc numeral classifier was * puaq 'fiuit', which is perhaps reflected as the first syllable of Aiwoo u-va '4'. Asuboa tV-, si-, and su- are conditioned reflexes of a PIM * tV , presumably reflecting the POc classifier * tau 'person' (Lynch, Ross, and Crowley 2002:73-74), as are the initial syllables of Buma and Vano 2-9 and Tanema 3 and 4. A number of initial syllables in Aiwoo, Nagu, Nebao, and Tanema reflect a PTM * 1V of unrecognized origin.

Some of the forms here can also serve as a warning. Nebao varo '8' and wa-hia '9' look suspiciously like reflexes of POc * walu '8' and * siwa '9', but in the context of the number systems of the individual languages we see that they belong to an additive system and mean '(five) plus three' and '(five) plus four'.

The forms in table 1 also provide a foretaste of the unexplained in Temotu historical phonology. Truncation of disyllabic roots is normal in RSC languages, but unexpected in UV

2.1 SOUND CORRESPONDENCES. Establishing the sound correspondences of the Temotu languages is a major step in establishing that they are Oceanic. Consonant correspondences are given in table 2. More lexical data are urgently needed, as some reflexes are attested by only one etymon, and POc * j and * y are omitted for want of data. Most reflexes are better supported, however. There are several facts about the phonological histories of Temotu languages that cannot be readily encapsulated in a table of sound correspondences. These are described in 2.2 and 2.3.

We attempt only a rough sketch of phonological history here, partly because the data do not allow us to fill in many details and partly because our focus is on Aiwoo. We shall not deal with the histories of vowels, other than to say that they are rather complex, involving deletion, epenthetic insertion, and harmony.

2.2 CANONIC FORMS. The canonic forms of Temotu languages differ widely, even though there is reasonable evidence that they constitute a subgroup and are descended from a single interstage language, Proto-Temotu (PTM) (see 2.4). The development of these differences is discussed in this section.

PTM must have retained at least POc final * -p, * -m, * -s, * -r, * -R, and * -k, as they are reflected in Vanikoro languages, followed by an added vowel, as shown in (3) through (8). An added vowel suggests that * -q was retained until relatively recendy. (9) POc * -m and traces of other POc final consonants are also present in Utupua languages. Generally, however, Utupua languages have lost POc final consonants. Utupua cognates, where they are known, are shown in parentheses below Reflexes of final consonants and added vowels are shown in bold.

(3) * -p: (10)

a. POc * maqurip 'alive' > Bum maluo VNo milipie TNM magiliva 'life' (TNB migipio)

In Buma maluo 'life', -o appears to reflect the added vowel after a now lost reflex of final * -p.

(4) * -m:

a. POc * rodrom 'be dark, be night' > Bum nedemo VNO nedume TNM ladome 'night'

b. POc * lalom 'inside (N)'> Bum ne-lema 'inside (ADV)'

c. POc * rabum 'grass' > Bum abo VNO abume TNM abome (NEB abeme ASU lcbumc TNB ubomc)

In (4c) Utupua languages also retain final * -m. This is the only instance in our data where all three Utupua languages reflect a POc final consonant.

(5) * -s:

a. POc * manipis 'thin' > Bum meneviro VNo menievire TNM mepiri

b. POc * panas 'hot' > TNm panara (but Bum VNo papa) (TNB kei pano)

In (5b) Buma and Vano do not reflect * -s, but Tanema does: we have no explanation for this.

(6) * -r:

a. POc * laur 'sea, seawards' > VNo laure 'reef'

b. POc * [b.sup.(w)] arapu 'long' > * bapur > PUV * biaure > BUM biouro VNo beure TNM va-beura (ASU a-bawa TNB kei-beu)

PUV * biaure 'long' reflects POc * r as a final consonant (the medial reflex would be PUV * 1): by inference it reflects a metathesized and consonant-final form, * bapur, of POc * [b.sup.(w)]arapu.

(7) * -R:

a. POc * gauR 'bamboo' > Bum jo-koro VNo je-wire TNm o-kaure

b. POc * waiR 'water' > Bum ero VNO wire TNm n-ira (NEB n-ie, ASU n-io, TNB no-wio)

c. POc * gio(r,R) 'spear, arrow' > VNO ore TNB VNo ora 'bow'

d. POc * pusuR 'bow' > Bum visone

Buma reflects * -R as -n- rather than as expected -r- in (7d).

(8) * -k:

a. POc * manuk 'bird' > BUM menuko VNO menuka TNM manuxa (TNB minia)

b. POc * namuk 'mosquito' > BUM muko VNO TNM muka (ASU muso TNB nanuyo)

The -a of TNB minia 'bird' appears to reflect an added vowel after a now lost reflex of * -k. The listed reflexes of POc * namuk 'mosquito' appear to reflect a form that has lost its first syllable, PUV * muke. Vanikoro languages also reflect an added vowel in items reflecting final * -q, but the reflex of * q itself has vanished:

(9) * -q:

a. POc * meRaq 'red' > BUM molae

b. POc * tanoq 'earth' > BUM tande VNO lenoe

c. POc * buaq 'areca nut' > BUM buioe TNM boie

A few alienable nouns in Vanikoro and Utupua languages have an accreted initial nV-, as in (7b), reflecting the POc determiner * na, but this accretion seems quite unpredictable. It seems evident that PTM retained the POc determiner in its NP construction, but it has been lost as a separate morpheme in all the modern Temotu languages.

RSC languages have gone a step …

Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: An Oceanic Origin for Aiwoo, the Language of the Reef Islands?. Contributors: Ross, Malcolm - Author, Naess, Ashild - Author. Journal title: Oceanic Linguistics. Volume: 46. Issue: 2 Publication date: December 2007. Page number: 456+. © 2008 University of Hawaii Press. COPYRIGHT 2007 Gale Group.

This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.