To Archive Index

The following is a translation from the German of the conclusion to Hans Schmidt's doctoral dissertation, "Rotuma: Sprache und Geschichte" (Rotuma: Language and History), submitted to the University of Hamburg in 1999.

Conclusion

Let me summarize my findings and put them into a historical order.

In the Bismarck-Archipelago and the Solomon Islands there are traces of human settlement over 30,000 years old. Suddenly, at least in archaeological dimensions, a new culture spread into the South Pacific between 1,600 and 1,000 BC. It is characterized by conspicuously artful pottery designs and different vessel forms, usually accompanied by further traits such as the use of earth ovens, shell tools, obsidian, one-piece fish-hooks, adzes and scrapers made of stone and shell; their dwelling places were large and close to the coast, often on offshore islands (Pawley & Ross 1993:446).

Contrary to earlier glottochronological estimates, the Proto-Oceanic language community split up around 1,600 BC. This more precise dating has become possible now that we can safely assume that the spread of Oceanic languages paralleled the spread of a distinctive archaeological tradition labeled Lapita in the central Pacific as well as elsewhere.

Around 1,200 BC a slightly modified form of this so called Lapita culture appears in Vanuatu and New Caledonia, i.e. Remote Oceania, which had not been settled before. Two hundred years later we find another variant, Early Eastern Lapita, having developed in Fiji, Tonga, and Samoa, which in turn gave rise to the materiel culture of Fiji and Polynesia—with such a smooth transition that one need not assume another wave of immigration (Green 1981:148). It can be deduced that the first settlers in Fiji spoke a language which later differentiated into the contemporary languages of Fiji, Polynesia, and Rotuma (Pawley 1996:111).

The central Pacific was probably discovered from North or Central Vanuatu since the languages of North/Central Vanuatu are the closest relatives of the Central Pacific ones. Both comprise the Eastern or Far Eastern subgroup. Ocean currents and winds also suggest that voyages of discovery to Fiji originated in Vanuatu more probably than anywhere else, e.g. the Southeast Solomons or New Caledonia. Lynch (personal communication) assumes that the first settlers came from Ambae. They spread so rapidly over Fiji and West Polynesia that archaeologists cannot identify one center of the Central Pacific culture (Green 1981:148). Because it is home to the greatest internal variety, linguistic comparisons point to Fiji as the center of dispersion from which the speakers of the proto-Central Pacific dialects spread out. The Fijian dialects have, for example, kept most oppositions between Proto-Central-Pacific phonemes that were largely lost in Rotuma or Polynesia.

Fiji was settled more than 3,000 years ago, but the first colonists did not speak a unified language like Proto-Polynesian, for example; Proto-Central Pacific is believed to have been a chain of dialects stretching from Fiji to at least two other island groups. If we keep in mind that Proto-Central Pacific presumably evolved within a few hundred years out of Proto-Oceanic, it is conceivable that this period of time was too short to allow more than a few common innovations, especially as the language area covered a vast region (Pawley 1996:111).

The settlement area was too far extended and contacts between island groups too limited to create and retain a single unified language, although Green claimed the opposite:

For a considerable time after much of West Polynesia was initially settled, a reasonably well unified chain of Central Pacific dialects was maintained throughout the whole Fijian and West Polynesian area (1981:147).

The small number of innovations found so far to define the Central-Pacific subgroup confirms archaeological findings proving that Fiji and Western Polynesia were settled almost simultaneously. This reflects a short period of common development. Innovations did occur, but they just spread to neighboring dialects (compare the features shared between Rotuma and Western Fiji or Rotuma and Vanualevu versus Eastern Fiji with Polynesia). Therefore Pawley & Ross (1995:52) labeled the Central-Pacific subgroup as an “innovation-linked subgroup” and not an “innovation-defined subgroup.”

To shed light on the origin of Rotuman it is necessary to study the development of the Fijian dialects in detail because they were at the center of the Central-Pacific dialect chain. Judging from the distribution of a couple of innovations, Geraghty (1983) concluded that a first split among the Central-Pacific dialects occurred between far eastern or “Tokalau-Fijian” and central/western dialects. The boundary ran through the Koro sea and East Vanua Levu. Pawley (1996) speculated that one of the reasons for the first split was that the settlers moved their attention to the fertile land and the food-rich lagoon at their new location. Contacts with their home islands in the west and their neighbors became sporadic.

An Eastern dialect (“Pre-Polynesian”), which developed into Proto-Polynesian, split off from the Tokalau-Fijian dialect chain about 1000 BC. Green argued that

a linguistically unified Pre-Polynesian language community was distributed over much if not all of the West Polynesian area, and perhaps even extended outside it, well before the break-up of Polynesian began. The development of a distinct Polynesian branch of Central Pacific did not take place in isolation on a single island group of West Polynesia (Green 1981:147).

However, Ross (1995a:103 fn. 39) was convinced that a Tokalau-Fijian dialect on just one of the island groups north-east of Fiji developed into Proto-Polynesian, and that centuries later it displaced other Central Pacific dialects in Western Polynesia and the Lau Islands. Based on phonological, grammatical, and lexical innovations in Polynesian languages, Pawley & Ross (1993:446) estimated the length of common development apart from other Oceanic languages to be 1,000 years. Archaeologists have established a corresponding period between the age of the Lapita sites in Western Polynesia and the settlement of East Polynesia.

Around 500 BC the culture of the Lau Islands started to adopt more Fijian rather than Polynesian traits (Best 1984:653f), and Geraghty found “support for the contention that the languages of Tokalau (Far Eastern) Fiji were originally genetically Polynesian rather than Fijian, but that they have been assimilated into the Fijian communalect continuum over the past one or two thousand years” (1983:379-382; also 1995a:11f). It likely happened during the rise of a regional Fijian power base, because increased contact among the Fijian dialects before 0 AD is obvious from the many phonological, morphological, and lexical innovations that spread over the whole Fijian Archipelago. These shared innovations define the Fijian Subgroup. Therefore this phase of the development of the Fijian dialect chain is labeled “Proto-Fijian.” The “stay-at-home” halves of the dialect chain of Western Fiji and Tokalau-Fiji merged again into a pan-Fijian dialect network. Still the pan-Fijian diffusions did not erase all earlier differences between the former central/western and far eastern parts of the Central-Pacific dialect chain in Fiji (Pawley 1996:111, Ross 1995:93).

About 2000 years ago there was another split into eastern and western dialects. New innovations with a more restricted distribution spread and changed the dialect geography of Fiji again. “Eventually, a sharp linguistic boundary developed between a well-marked Western Fijian subgroup and the chain of dialects extending over the rest of Fiji, which form a loose-knit Eastern Fijian subgroup” (Pawley 1996:111). The proto-Fijian period started with the spread of the pan-Fijian innovations and ended with the divergence of Western Fiji from the rest of the re-united Fijian dialect chain.

“The appearance of the well-marked Western Fijian subgroup can be explained partly in terms of geography and settlement patterns. Archaeological evidence indicates that Viti Levu had a sizeable inland population by 2000 years ago (Green, personal communication). Before this inland population built up, it is hard to see how the present Eastern-Western linguistic boundary could have formed. When people left the coastal regions and established sizeable inland populations up the main river valleys of Viti Levu the central dividing ranges of Viti Levu became a major barrier to communication, isolating the people of western Viti Levu from the rest of Fiji except for coastal contacts” (Pawley 1996:111).

So where does Rotuman come from? Since Rotuman shares about the same number of innovations with the Fijian dialects of Vanua Levu (especially the north coast) as with those of Western Fiji, Geraghty concluded:

Since it appears that Vanualevu was heavily influenced in recent prehistory by languages of the coastal southeast Vitilevu prestige area . . . I would tentatively propose that Rotuman derived from Vanua Levu at a time when that area was more similar to Western Fijian (1996:90).

This means that the first settlers of Rotuma came from the western half of the Central-Pacific dialect chain and settled Rotuma before Northwest Vanua Levu was dominated by East Fiji. Their dialect was the predecessor of Rotuman.

When did Rotuman leave the Central-Pacific dialect chain? The earliest archaeological specimens from Rotuma are approximately 1,700 years old. Pawley (1979:40) first thought that Rotuman had split from the Central-Pacific dialects in Western Fiji earlier than Polynesian had split from the Tokalau-Fijian dialects because it shares most innovations with all other regions of Fiji. However, the isoglosses in sections 5.5 and 6.4 of my dissertation clearly show a special connection between Rotuma and Northwest Vanua Levu. In a more recent paper Pawley reconsidered his position:

The separation of Rotuman must have preceded the Proto Fijian period. To be more exact, it must have happened before the spread across Fiji of those particular innovations that are not found in Rotuman. For example, it is characteristic of the whole Fijian group, with sporadic exceptions in Vanua Levu, that the PCP rising vowel clusters *ae and *ao reduce to a single vowel, which may be a, e, or o, depending on stress placement in the original form. As Rotuman preserves the original clusters, the spread of this innovation across Fijian can be dated after the divergence of Rotuman (1996:111).

The present boundary between Western and Eastern Fiji has been pushed further to the West compared to the borderline between Tokalau-Fijian and central/western Central-Pacific dialects; it runs now through Vitilevu and Kadavu. So the Fijian dialects in northwest Vanualevu that are probably the closest relatives of Rotuman do not belong to the western half of the Fijian dialect chain any longer, but the eastern one. They were strongly influenced by the so-called East Viti Levu prestige area—the home of the Bauan dialect that was adopted as the standard variety of Fijian (SF).

Although “the historical relationships between the Fijian dialects themselves and with regard to the Polynesian languages cannot adequately be shown in a family tree” (Pawley 1979:13), I try and present a synthesis of the claims of Geraghty (1983) and Pawley (1996:95) in the following diagram. If one accepts that Rotuman arose from the central/western half of the Central-Pacific dialect chain, then it is obvious that many splits and fusions have occurred since among the “stay-at-home” dialects in Fiji, obliterating or blurring many former similarities with Rotuman.

Differentiation of the Central-Pacific Languages

The relationships may be confusing: Central-Pacific languages are spoken in Rotuma, Fiji, and Polynesia; they grew out of the Proto-Central-Pacific dialects. We speak of Fijian dialects only after Proto-Fijian developed such different varieties. There are Eastern Fijian dialects today like those in Northwestern Vanua Levu which belonged to the (Western- or) Central-Pacific dialect chain (as opposed to the Proto-Tokalau-Fijian one).

The overlap of the dialect boundaries is shown on the following map.

Dialect Boundaries in Fiji (Pawley 1979:39)

Isoglosses overlap the traditionally accepted branching (Fijian, Polynesian, Rotuman). On the one hand, the Polynesian Language family probably shared some innovations with several Eastern Fijian dialects, mainly those of the Lau Islands and parts of East Vanua Levu, and not others. On the other hand, Rotuman shows considerable similarity in obvious innovations with western Fijian languages as opposed to Polynesia. The isoglosses link Rotuman especially to contemporary dialects of West Vanua Levu. By contrast, Rotuman displays very few similarities with Polynesian languages that are innovations and not borrowings (Pawley 1996).

Rotuma’s separate way

The first settlers on Rotuma may have kept contact with their homeland for some time (Green 1981:149), but then a longer period of isolation or separate development ensued. I assume that the striking changes in Rotuman occurred in this phase. Ross (personal communication) and Irwin agree:

“Islands began to diverge faster in isolation, from the time effective communication between them slowed or ended, rather than when contact between them began” (Irwin 1992:174).

Compared to Central-Pacific languages that had not lost touch with their neighboring dialects and languages, the sound changes in Rotuman were much further reaching and more numerous during this period of isolation than in later centuries (when contact with Polynesia and perhaps with Fiji was more intensive).

The trigger for some of the changes in the language may have been pulled when the predecessor of Rotuman had not yet differentiated itself from the other Central-Pacific dialects in Western Vanua Levu, but most peculiarities of Rotuman—the far-reaching sound changes, the creation of short forms of all content words, lexical and semantic changes, syntactic changes like the postposition of the article—became common usage after the split-off from other dialects. The idiosyncratic development during the long isolation is the reason that “The Rotuman language is totally unintelligible to speakers of Fijian and Polynesian languages, to which it is genetically most closely related” (Geraghty 1984:34). The changes were not triggered by Polynesian influence. The short form creation, for example, may be viewed as a parallel to the independent creation of a definitive accent in some Western Polynesian languages. The tendency to shift stress from the penultimate to the last syllable to express emphasis or definiteness seems to have been present in the common predecessor language, even though one cannot find traces of it in Fijian. The accent shift in Western Polynesian languages is probably not due to Tongan influence (compare Geraghty 1984:34 and Tsukamoto 1994:54). So Rotuman short form creation preceded the first intensive contact with Polynesians.

Lynch’s observation that the possessive systems of the Fijian dialect of Gone Dau and Rotuman resemble each other very closely confirms the theory that Rotuman split off from a Central-Pacific dialect on northwest Vanua Levu. Further evidence can only be found by studying the Fijian dialects of that area, which has either not been done or not yet been published. The distribution of reflexes of *R “is consistent with Geraghty’s hypothesis of a Proto-Central-Pacific dialect chain, with Polynesian most closely connected to Lau and eastern Vanua Levu” (Pawley 1996:98).

As far as they can be pushed into a strict framework of time, the events which may be linked to the history of the Rotuman language are listed in the following table:

Dates in the History of the Rotuman Language

Polynesian influence on Rotuman

Rotuman, especially its lexicon, shows great influence of Western Polynesian languages, which is the result of centuries of intensive contact.

What was the nature of the Polynesian influence on Rotuman? The Rotumans borrowed many technical terms for objects of material culture and terms used in political and social organization, from fishing to warfare–just like other non-Polynesian languages in Melanesia that have borrowed many terms from the same areas.

When and from where did these Polynesians come to Rotuma? A group of immigrants from Samoa and the Tongan invasion from Niuafo'ou are remembered in Rotuman folklore. Gardiner (1898) reckoned from Rotuman genealogies that the leader of the Tongan invaders from Niuafo'ou had lived twelve generations before the turn of the 20th century. The neighboring islands were subjected to Tongan maritime dominance at about the same time.

The Tongans were the great travelers of the western Pacific. They certainly had two-way contact with Kiribati, over 2,000 km away, and probably also with Ponape, at a distance of 4,000 km” (Geraghty 1989:380).

Woodford assumed that the Tongans had stopped at Rotuma en route: “We know that Tongan piratical expeditions were also in the habit of visiting the Ellice [Tuvalu] group. They probably used Rotuma as a resting-place and point of departure for further voyages” (1916:28).

I apply the same rationale to the travels of other Polynesians who left traces of nuclear Polynesian languages in Rotuman. These immigrants or visitors can no longer be identified in genealogies and have been relegated to mythical founder heroes. Can the marked impact of these non-Tongic languages on Rotuman be dated more precisely than some time between the split-up of the Polynesian language family (into Tongic and Nuclear-Polynesian) and the rise of Tongan predominance in the Central Pacific? Yes, if we identify these Polynesians as the ones who also settled Tuvalu and the Polynesian outliers, because their languages belong to the Samoic-Outliers-Group of the Nuclear-Polynesian languages. The outliers and Tuvalu were settled in the second millenium AD from Western Polynesia (Kirch 1984b:237, Garanger 1972:134), long after the separation of Tongic and East-Polynesian languages from the rest of the Polynesian group (which can be subsumed under the heading of Samoic-Outliers-languages).

Linguistically, the Outliers have their closest ties with Samoa and with islands west of there, including East Futuna and Tuvalu [...] However, at a more general level, it may be worth noting that the primary language link is with Samoa, which may suggest an earlier phase of a wider West Polynesian contact sphere prior to the situation at the time of contact, when Tonga appeared to dominate foreign affairs and to influence the language of some neighbors and sometimes even to overlay an early influence by Samoan (Irwin 1992:188).

Another piece of evidence is the existence of Polynesian loanwords in Kiribati, the Micronesian island group north of Tuvalu. Their shape led Harrison (1994:336f) and Geraghty (1994b:243) to believe that they originated for the most part from Nuclear-Polynesian languages like Samoan or Tuvaluan.

The most obvious sources of Polynesian borrowings into Gilbertese are the geographically adjacent Samoic-Outlier languages or their immediate antecedents. And, indeed, in all but a very small number of cases the likely items are found in Samoic languages and appear in Gilbertese in a form suggestive of the phonology of Samoic languages. There are, however, a small number of non-Samoic items in Gilbertese (Harrison 1994:336).

With the exception of Tuvaluan, there is little evidence pointing to particular Samoic-Outlier languages as sources for Gilbertese items” (Harrison 1994:337).

Several words can only have been borrowed from Samoan because they reflect PPN *k as Ø. This is typical for a language without phonemic glottal stop when it borrowed from another language with glottal stop. “Glottal stop is the Samoan reflex of PPN *k, e.g. KRB -ura red < SAM 'ula, KRB ie sail < SAM 'ie” (Geraghty 1994b:243).

The Samoan sound change of PNP *k to glottal stop was already completed during Cook’s time (Hovdhaugen 1986:316). According to Maude (1963) Kiribati has been in contact with Polynesians for 600 years. Centuries later the Tongans also reached the outliers and Kiribati.

What has the influence of Nuclear-Polynesian languages on Kiribati and the outliers to do with Rotuma? Well, Rotuma is situated right along the road from Western Polynesia to Kiribati, Tuvalu, and to the outliers in Melanesia and Micronesia. In a computer simulation of drift voyaging, Ward, Webb & Levison (1973) showed 'Uvea, Futuna, Rotuma, and Tuvalu to be the most likely points of origin for the settlement of the Polynesian outliers. During most of the year, ocean currents and winds in the Central Pacific go in the direction from Southeast to Northwest. As the Polynesian outliers lie to the west of Polynesia, it is quite likely that vessels drifted there by accident. In the last centuries many more voyages from east to west, whether accidental or on purpose, were documented than from west to east (Dening 1963:129). This scenario contrasts with the first planned settlement of triangle Polynesia.

We may assume that the present population [on the Polynesian outlier Takuu] is a remnant of a tribe which had come from Polynesia, probably from Samoa, using the islands Rotuma and Tikopia as stepping stones (Parkinson 1907:518).

The navigators of the Nuclear-Polynesians used Rotuma as a land mark, if not as a supply station for the rest of the long journey that lay ahead through hundreds of miles of open ocean. There were immigrations from various islands to the outliers, perhaps even planned and targeted trips back and forth. Hollyman (1959:362) used the example of the outlier West 'Uvea to show how a small group of immigrants founded a Polynesian colony by marrying Melanesian women and largely adapting to the local customs while retaining their Polynesian language.

After the colonization of the outliers and the rise of Tongan maritime predominance, Rotuma still had contact with speakers of Nuclear-Polynesian languages. Just think of the trade links with Vaitupu (Tuvalu) and the loanwords from Futuna, 'Uvea, and Samoa.

After its “discovery,” Rotuma became a popular destination for European and American ships. Many sailors jumped ship on the island, many Rotumans sought work on the whaling ships. On board they communicated with their colleagues from other islands and the white officers in a kind of Pidgin English. These mariners were highly respected after their return to the island, and fragments of Pacific Pidgin English found their way into the Rotuman lexicon.

After the annexation of Rotuma in 1881 everyone who wanted to visit or leave the island had to do this via Fiji. The whole trade went through Fiji. Consequently, English and Fijian became the two languages that exerted the main outside influence on Rotuman during the past two centuries. In the following diagram I show when each language had what kind of influence on Rotuman.

The Influence of Other Languages on Rotuman

(the stronger the influence, the wider the column)

I envisage the prehistoric development of Rotuman in several phases:

1. Rotuma was first settledby people who spoke one of the Central-Pacific dialects; its closest related varieties were still spoken in northwest Vanua Levu;

2. many peculiarities or idiosyncrasies of the language developed during a period of relative isolation caused by its geographical remoteness and a negligence in sustaining outside contacts;

3. first wave of Polynesian borrowings, caused by immigration or multiple visits of speakers of Nuclear-Polynesian languages who presumably also settled the Polynesian outliers (from around the 13th/14th century onwards);

4. second wave of Polynesian borrowings, caused by the conquest and temporary occupation of the island by Tongans (from around the 16th/17th century onwards).

See the table above for an approximate dating of the phases.

My work has by and large confirmed the hypotheses of Pawley and Geraghty that Rotuman belongs to the Central-Pacific language group and that its closest relatives are Fijian dialects in northwest Vanua Levu. After the departure of the people who settled Rotuma, these areas became subjects of the expansion of the militarily superior chiefdoms in east Viti Levu (Bau, Rewa etc.) and consequently their dialects today resemble the East-Fijian ones more closely than the West-Central-Pacific ones they had belonged to earlier.

The striking similarity of the Rotuman lexicon with Polynesian languages must be explained as the result of massive borrowing. After a long period of isolation, Rotumans were eager to learn new techniques and to adopt new ideas and fashions from Polynesians who started to visit the island about 750 years ago. My work has demonstrated the large extent of the Polynesian, especially Tongan, influence on language and culture of the Rotumans.

References