The western Pacific is divided by botanists into two floristic regions, northwest Melanesia and southeast Melanesia. NW Melanesia includes New Guinea, the Bismarck Archipelago (New Britain, New Ireland and the Admiralties) and the Solomons archipelago including Buka and Bougainville but excluding the Reef and Santa Cruz Islands. SE Melanesia includes the Reef and Santa Cruz Islands, Vanuatu, New Caledonia and the Loyalties, and Fiji. The boundary between NW and SE Melanesia thus corresponds with the boundary between Near and Remote Oceania (vol.2, ch.2, §2).
Within NW Melanesia there is a gradual reduction in the number of genera and species as one moves east- and southeastward through the islands, but there are no major new genera. In SE Melanesia, on the other hand, we encounter genera that are not represented or hardly represented in the islands of NW Melanesia, for example the genus Agathis, to which the kauri species of SE Melanesia belong.
In this chapter, indeed in this volume, we are mainly concerned with NW Island Melanesia (i.e. NW Melanesia other than New Guinea), and particularly with the Bismarck Archipelago, as this is where the POc homeland was located (vol.2, ch.2).
There are a number of ways in which one might classify plants as a basis for dividing this book into chapters. Ideally the classification should be one which has some basis in POc terminologies, since the reconstruction of these terminologies is what we are about. Some obvious ways of classifying Oceanic plants are:
Although use forms the basis of an important work on the plants of the Solomon Islands (Henderson & Hancock 1988), so many named plants have multiple uses that this criterion is not a suitable basis for classification. Plants may provide food (starch, greens, fruit or nuts), construction materials (housing timber, canoe components, cordage), firewood, caulking, adornment, perfume, medicines, fish poison, parcelling, baskets and mats and more. The extent of multiple usages is highlighted by Thaman (1994), who analyses the uses of 140 plant species commonly found in the habitats of the Oceanic-speaking Pacific.1 He finds that the 140 species have a mean of 7.3 uses each, the coconut palm having the maximum of 125 uses and just two species having no uses. The most frequently reported use was medicinal, with 113 species out of 140 being used medicinally. Sixty species provided construction timber (including 54 out of 62 tree species). Thirty-four species were used in canoe- or boat-building (including 30 tree species). Fifty-one species were used as firewood (including 43 tree species). Thirty species provided dyes. Twenty-eight were used in making fishing equipment. And so on. The picture presented by Powell (1976) for New Guinea is similar.
There is a vast literature on the medicinal uses of plants in Oceanic societies, but no attempt is made to review it in the pages of this book. Many plants have many different medicinal uses, often varying from place to place, and, conversely, similar maladies are treated with preparations from many different plants in different places. This means that there is no real likelihood of reconstructing the medicinal uses to which early Oceanic speakers put individual plants. In chapters 5-8 medicinal uses of plants are mentioned for those Island NW Melanesian and niVanuatu societies for which a substantial body of ethnobotanical information is available (sources are listed inch 2, §5.2), but this is intended as a sample, not a survey.
The second possibility is ethnobotanical classification. In chapter 3 Evans reconstructs five major first-order categories of plant distinguished by POc speakers: *kayu ‘tree, shrub’, *waroc ‘vine’, *pali[s,j]i ‘grass’, *taliŋa ‘mushroom’ and *limut or *lumut ‘moss, algae’. A classification on this basis is certainly possible, but, as Evans points out, there are residual categories like palms (ch.3, §4.3), ‘clumps’, bamboos, ferns and pandanus (ch.3, §4.8) which may or may not have been counted as *kayu by POc speakers. Thus there were quite possibly a number of fairly small first-order categories in addition to the five for which Evans is able to reconstruct labels above. Candidates are *qauR, generic for bamboos (ch. 13, §3.1), and *padran, generic for pandanus species as well as denoting the coastal pandanus, Pandanus tectorius (ch.11, §2.5.1).
The largest problem with using ethnobotanical classification as a basis for categorising the reconstructions in this book, however, is practical: *kayu is a huge category, in need of subdivisions for which Evans finds no linguistic warrant.
The third and fourth classificatory approaches form the basis of the chapter divisions from chapter 5 onward. Vegetation habitats, described in §3, §4 and §5 are the basis for chapters on wild plants, while plants that are generally tended or cultivated occur in separate chapters which have as a partial basis the ingredients of a typical Oceanic meal, described in §6.
This mixed classification reflects the primary POc land use division between *qutan ‘bushland, hinterland; inland’, where wild plants grow, and *quma ‘garden, plantation’, where plants are cultivated.
Wild plants are then divided primarily by vegetation habitat: coastal strand vegetation, mangrove swamp, primary lowland rain forest, and secondary lowland rain forest and grassland. This has the virtue of placing plants in their environments and in relationship to one another within the environment. Oceanic languages have names for types of location (even though the reconstruction of POc names for them is somewhat problematic, partly because the best described Oceanic environments are rather different from one another: see vol.2, ch.3, §1), and we can assume that this arrangement is at least not radically at odds with POc speakers’ conceptualisations of their environment,
Food plants are divided first into the two major categories of the POc menu: *kanaŋ ‘starchy food, staples’ and *tamaji ‘additional ingredients to accompany starchy food’. The latter is included in three chapters, on green vegetables, nut and fruit trees, and the coconut palm. The last is treated in a separate chaplter because of the complexity of its associated terminology. A further chapter treats plants that are cultivated, but not primarily for food.
The two chapters following this one, chapters 3 and 4, are by Bethwyn Evans, who treats the POc primary ethnobotanical classification and POc terms for parts of plants. The remaining chapters follow the outline just provided:
The placing of plants in the first four of these chapters is at times a little arbitrary, as the same plant may occur in more than one habitat type. Crossreferences are provided to take account of this.
The agronomic boundary between bush and garden is fuzzy in Oceanic food production and was almost certainly just as fuzzy in POc agriculture (§4). The distinction between trees that are tended or transplanted and those which aren’t is somewhat clearer, and the presence of a chapter on nut and fruit trees is a response to this fact, as most of these are trees that are at least tended in some Oceanic societies and likely to have been tended by POc speakers. This chapter is placed after the chapters on cultivated plants in recognition of the fact these trees form part of the overall agroforestry systems of NW Island Melanesian societies.
Five natural vegetation habitats are recognised by botanists in NW Island Melanesia. The types and their descriptions are drawn mainly from Mueller-Dombois & Fosberg (1998: 50–72).2 Four of these are:
The fifth is the montane rain forest of New Britain, New Ireland and Bougainville, but this probably played little role, if any, in the lives of POc speakers and is not discussed further in this volume.
In addition there are three anthropogenic vegetation habitats:
These vegetation habitats are briefly described below, each with a listing of the species of which it is typically composed.
Coastal strand vegetation falls into three zones, a herbaceous zone, beach scrub, and littoral forest.
On sandy beaches and beach ridges,3 vegetation begins at the high-water mark on the seaward slope with a herbaceous cover of creeping plants such as Ipomoea pes-caprae, Canavalia rosea (syn. Canavalia maritima) (no reconstruction) and Wedelia biflora, as well as grasses and sedges, including Thuarea involuta.
Next to or overlapping the herbaceous zone comes bush scrub with shrubs such as Pemphis acidula and Scaevola taccada and low-growing bushy-crowned trees like Hibiscus tiliaceus Thespesia populnea and Tournefortia argentea often densely tangled by climbers like Flagellaria indica. There is a ground layer of ferns, grasses, gingers and herbs including Crinum asiaticum.
On its landward side beach scrub borders, gradually or abruptly, on littoral or coastal strand forest. This is often dominated by evergreen broadleaftrees like Barringtonia asiatica, Terminalia catappa, Heritiera littoralis and Calophyllum inophyllum (and in the Solomons archipelago Cerbera manghas) or the screwpine Pandanus tectorius, (on coral Pandanus dubius) or sometimes Casuarina equisetifolia. On beaches where the ridge formation has been eroded, the littoral forest borders immediately on the beach, and Barringtonia asiatica predominates. On coral islands where mixed littoral forest covers the whole island, the forest may include Diospyros, Myristica and Rhus species. In the Solomons the lower storey includes Diospyros species, Ficus austrina, Hibiscus tiliaceus, Kleinhovia hospita, Morinda citrifolia and Premna corymbosa (Paijmans 1976:29-20, Henderson & Hancock 1988:321, Mueller-Dombois & Fosberg 1998: 50, 59, 70).
As Barrau (1955: 17) notes, in many areas the littoral forest has been replaced by coconut palms.
Mangrove forests occur on stretches of coastline sheltered from wave action or along estuaries or even on protected coral reefs, and especially in areas where there is regular rainfall which washes salt out of the soil. The mangrove forest may extend from seawater salinity on the seaward side to almost freshwater conditions on the landward side, where there is usually an abrupt transition to freshwater swamp forest. Mangrove forests are located in the intertidal zone, so the seaward margin undergoes much deeper twice-daily inundation than the landward margin. On the seaward margin species include A vicennia marina, Sonneratia caseolaris, Sonneratia alba and occasionally Ceriops tagal. Further landward Rhizophora and then Bruguiera species take over and the canopy assumes a forest stature. Towards its landward border the mangrove forest becomes more diverse, forming a canopy up to 25 m in height, and in the Bismarcks includes Lumnitzera littorea, Xylocarpus granatum, Excoecaria agallocha, Camptostemon schultzii, Heritiera littoralis, Intsia bijuga and Inocarpus fagifer. Intsia bijuga and Inocarpus fagifer are also common lowland rain forest and swamp forest trees. The more landward mangrove forest is more open and has an undergrowth of shrubs and low-stature trees including Dolichandrone spathacea and Myristica hollrungii. The Nypa fruticans palm also grows on the landward side of estuarine swamps (Paijmans 1976: 31-34, Mueller-Dombois & Fosberg 1998:50–51).
In the Solomons Lumnitzera littorea, Ceriops tagal and Dolichandrone spathacea occur towards the landward border of the mangrove forest. Inland of these are Sonneratia species and Xylocarpus granatum and as the ground becomes less saline Calophyllum inophyllum, Fagraea racemosa, Heritiera littoralis, Intsia bijuga and Pandanus species, with an understorey of ferns and the shrub Acanthus ebracteatus (Henderson & Hancock 1988: 319).
Freshwater swamp forests occur on larger islands in areas where the water table, often brackish, is near the surface and during high-rainfall seasons sometimes above it. These locations are typically flat areas near the coast with poorly drained soils. In NW Island Melanesia, all large tracts of freshwater swamp forest are found in Bougainville and the Solomons. They are few and small in the Bismarcks: there are patches on the north coast of New Britain, two deltas in the south of New Ireland, and a small amount (about 7%) on Manus Island and Los Negros (Paijmans 1976:37-48, Mueller-Dombois & Fosberg 1998:51-52,60,71, Freyne & Bell 1982). Since the homeland of POc speakers was quite clearly in the Bismarcks, no chapter in the present volume is devoted to freshwater swamp forest. The trees characteristic of this habitat are in any case all found elsewhere, usually in lowland rain forests.
The most widely distributed natural vegetation habitat in NW Island Melanesia is lowland tropical rain forest. Its most important species belong to the genera Calophyllum, Campnosperma, Canarium, Cryptocarya, Dracontomelon, Ficus, Intsia, Octomeles, Pometia, Pterocymbium and Terminalia, along with Eucalyptus deglupta.
In Bougainville Pometia pinnata and Vitex cofassus form a forest with a tall canopy up to 35 m, and other genera include Alstonia, Celtis, Cryptocarpus, Dysoxylum Elaeocarpus and Sterculia. The understorey in Bougainville includes trees of the genera Diospyros, Garcinia, Gnetum, Myristica, Syzygium and palms of the genera Areca, Caryota and Licuala, as well as bamboos, tree ferns (Cyathea) and Pandanus (Paijmans 1976:64-65, Mueller-Dombois & Fosberg 1998: 60–61).
Lowland tropical rain forests are more species-rich than other kinds of vegetation but nonetheless grow poorer as one moves southeastward. In the Solomons there are twelve species of big (canopy) trees: Calophyllum kajewskii, Calophyllum pseudovitiense, Campnosperma brevipetiolatum, Dillenia salomonensis, Elaeocarpus sphaericus, Endospermum medullosum, Gmelina moluccana, Maranthes corymbosa, Parinari salomonensis, Pometia pinnata, Schizomeria serrata and Terminalia calamansanai. Where forests have been broken by cyclones or man, colonising species include Canarium species and Vitex cofassus. Lower tree and shrub layers consist of Barringtonia and Boerlagiodendron species, Leea indica and Areca catechu. The herb layer is patchy, but where gaps occur in the canopy, species of Calamus, bamboos and gingers predominate (Henderson & Hancock 1988: 320)
In the few locations in NW Island Melanesia where there is a marked dry season Garuga floribunda occurs alongside Terminalia and Ficus species (Paijmans 1976: 52).
Much of the vegetation in NW Island Melanesia today is anthropogenic. It falls into three categories: garden vegetation, secondary lowland rain forest and grasslands.
Lapita villages were apparently typically located on or immediately behind the beach, as many modern villages in the Bismarcks still are. When a contemporary village is on the beach, however, the gardens often lie a distance away in the rainforest, sometimes in the submontane foothills. Villagers in rain forest areas of NW Island Melanesia typically follow a system which has been labelled ‘bush fallowing rotation’. The forest is cleared, the garden is planted and harvested for about a year, then left fallow for 10–15 years. During the first two years of the fallow period, some crops continue to be harvested from the garden. Typically a family is entitled to use a number of garden plots which have been cleared by earlier generations and which are at different stages of fallowing rotation. Children are taught which plots the family may use, which places are good for taro, which for yams, and so on, across the whole fallow area used by the family as well as in current gardens (Kwa’ioloa & Burt 2001: 30). If a family has enough garden plots, then there is no need to create new plots by clearing primary forest. If not, then either plots will be used more frequently (not an ideal choice) or a new plot will be created by clearing forest. If the fallow period is long enough and rainfall is sufficient, then primary regrowth occurs and the garden plot more or less merges back into the primary forest. If the land is reused too early, then eventually secondary forest replaces primary regrowth. In drier areas it is replaced by grassland (Barrau 1955: 17, 31, 1962: 45-47).
Garden vegetation today includes planted trees and sometimes wild trees that were present when the forest was cleared and are tended in situ (chapter 11), as well as the various staples (chapter 9) and plants grown as green vegetables (chapter 10). Trees vary from place to place but include breadfruit (Artocarpus altilis), mango (Mangifera indica), Malay apple (Syzygium malaccense), golden apples (Spondias cytherea), dragon plums (Dracontomelon dao and Dracontomelon vitiense), canarium almonds (Canarium indicum and Canarium salomonense), Barringtonia edulis, Citrus (mostly recent introductions), guava (another recent introduction) and coconut palms (chapter 12).
Henderson & Hancock (1988: 323) give an account of the regrowth sequence in the Solomons. Secondary rain forest trees tend to be more light-demanding species, which include Acalypha grandis, Alphitonia incana, Hibiscus tiliaceus, Macaranga species, Melochia umbellata, Pipturus argenteus and Schleinitzia novo-guineensis, alongside planted species of Musa and Heliconia. As the fallow progresses, they lose their dominance to species typical of older regrowth such as Falcataria moluccana, Cananga odorata, Ficus species, Kleinhovia hospita, Rhus taitensis and Trichospermum psilocladum, along with the breadfruit and the mango and treeferns like Cyathea brackenridgei and Cyathea lunulata. Gingers occur in the shrub layer, along with the palms Areca catechu and Caryota rumphiana. Finally, some of the large tree species of primary forest return, especially Pometia pinnata and Vitex cofassus.
Mueller-Dombois & Fosberg (1998: 63) describe forest regrowth in Bougainville, but their stages are less clearly delineated than Henderson & Hancock’s. Regrowth vegetation includes wild varieties of cultivated plants, along with Heliconia indica, gingers, Caryota palms, grasses (Imperata cylindrica, Paspalum, Pennisetum macrostachyum), Kleinhovia hospita, Hibiscus tiliaceus, and species of Macaranga and Ficus (Mueller-Dombois & Fosberg 1998: 63). Important secondary growth is dominated by species of Glochidion, Macaranga and Mallotus, along with species of Trema, Alphitonia, Casuarina, Trichospermum and Hibiscus, and the species Leucaena leucocephala, Kleinhovia hospita, Paraserianthes falcataria, Melanolepis multiglandulosa and Burckella obovata. There are also primary forest trees that have not been removed: Canarium and Barringtonia species, breadfruit and Pangium edule. In older secondary forest Cananga, Endospermum, Canarium, Euodia, Laportea and Sterculia occur (Paijmans 1976: 59).
Grasslands are quite rare in most of NW Island Melanesia, but occupy extensive areas of north Guadalcanal. They are dominated by kangaroo grass (Themeda australis), sword grass (Imperata cylindrica) and Pennisetum polystachion. They are usually maintained by regular burning. In less frequently burned areas Saccharum spontaneum and Miscanthus floridulus also occur. Phragmites karka and Cyperus species appear in more poorly drained areas (Henderson & Hancock 1988:318-319, Mueller-Dombois & Fosberg 1988: 56-57).
The division of vegetation habitats into natural and anthropogenic reminds us that human beings have brought substantial changes to their NW Melanesian island habitats. Recent scholarly work on the agriculture of the region has emphasised that there is no clearcut agronomic boundary between garden and bush. As noted above, gardens sometimes include fruit- or nut-bearing trees that survive from the earlier primary forest, or forest tree species that have been planted in the garden. Sometimes fruit or nut trees are planted near the village as well. Some species are tended in their natural forest habitat.
Kennedy & Clarke (2004) argue that there is no sensible line to be drawn between crops grown in gardens and crops acquired from the bush, because they together constitute an integrated system of resource management. They examine the literature on sago, canarium, pandanus, breadfruit and bananas, and show that all five have long been tended and transplanted in ways that have fundamentally altered the landscape. They list other plants with similarly long-term relationships with human beings: Gnetum gnemon, Inocarpus fagifer, Pangium edule and Pometia pinnata. We can add to this list at least Barringtonia edulis, Barringtonia novae-hiberniae, Barringtonia procera, Burckella obovata, Dracontomelon species, Spondias cytherea, Syzygium malaccense, Syzygium aqueum, Terminalia catappa and Terminalia kaernbachii (A. Walter 1994).
Breadfruit (Artocarpus altilis) and canarium almond (Canarium indicum and Canarium salomonense) trees are usually individually owned and tended where they have grown in the forest, either naturally or from planted suckers. Forest growth is cleared away from around the base of the breadfruit tree, and it may be fenced. Breadfruit seedlings are often also transplanted to village areas and older plants may be propagated by planting - depending on location - seeds, root cuttings or suckers.4 In the Solomons most canarium trees are planted near villages, but some grow wild. Terminalia catappa and Terminalia kaernbachii are also often planted, whilst other Terminalia species with edible kernels (Terminalia copelandii, Terminalia impediens) are harvested from the forest. Several Syzygium species have edible fruit (Syzygium malaccense, Syzygium aqueum) and are either tended within the forest or planted. Barringtonia edulis is commonly planted as a village fruit tree in the Solomons (Paijmans 1976: 123-124, Evans 1999: 1).
The importance of tree crops to Pacific Islanders was first stressed by Barrau (1955, 1963) and reinforced by Yen (1974a). They pointed out that fruits and nuts that were reported simply to be gathered were in fact tended in varying degrees as part of the agricultural system. This has been reaffirmed by other scholars, e.g. Thaman (1989), Flavelle (1991), A. Walter (1994), McEldowney (1995), Hviding & Bayliss-Smith (2000: 23), Chowning (2001: 77-78), Walter & Sam (2002: 76–77) and McClatchey et al. (2006a). For example, on Baluan Island (Admiralties) McEldowney found 23 tree crops, 17 of which were fruit- or nut-bearing. All occurred in three different types of location: near villages, in orchards, and scattered across gardened land. Kennedy & Clarke (2004) point out that the situation is sometimes even more complicated, as the status of a crop can range from wild through tended to cultivated, i.e. planted and cloned, and wild varieties are often used alongside domesticated plants, i.e. propagated vegetatively by cuttings or by planting suckers.
Before the introduction of steel implements tree crops are known to have had greater nutritional significance on the islands off the north coasts of New Guinea and New Britain and on the Mussau Islands than they have today, and it is likely that this is true at least of small islands throughout NW Island Melanesia. For example on Karkar Island, very important food sources included coconut, canarium almonds, breadfruit (Artocarpus altilis), Tahitian chestnut (Inocarpus fagifer), Indian chestnut (Terminalia catappa ), dragon plum (Dracontomelon dao) and the fruit of Pouteria maclayana. Arable agriculture has increased in significance since the introduction of metal tools 80 or so years ago (Bourke 1996, Allen et al. 1994), but from my own observation it is clear that these nuts and fruits retain traditional significance.
Because agriculture is centred on what are conventionally called ‘gardens’ in English, the distinction between the English terms ‘horticulture’ and ‘agriculture’ is largely irrelevant in a Pacific context (Brookfield 2001:6, Kennedy & Clarke 2004). However, recent literature calls the use of either term into question because of the absence of a clear boundary between garden and forest in Pacific agriculture. Pinning down exactly whether a particular crop should be regarded as wild or cultivated in a particular location is extremely difficult. In consequence Kennedy & Clarke (2004) adopt the term ‘agroforestry’ from Clarke & Thaman (1993).5 Whilst the reason for this decision is clear, however, there is still reason to retain ‘horticulture’ for activities in food gardens, as Meredith Osmond does in chapter 5 of volume 1, and to use ‘arboriculture’ for tree-tending activities in the forest.
The extent of agriculture in the Bismarck Archipelago before the arrival of POc speakers, i.e. Lapita people, is unknown, but there is reason to believe that the tending of bush trees and the planting of trees at convenient locations had been practised there long before the Lapita period. Before the arrival of Austronesian speakers, NW Island Melanesia was populated by groups who may or may not have cultivated a staple crop. However, as early as 20,000 years ago Allen (1993) and Gosden (1995) detect a gradual change in pattern in New Ireland, whereby people had started to move scarce resources to the locations where they lived instead of themselves moving from resource to resource. A species of Phalanger (possum) was imported into New Ireland, canarium trees were transplanted there from elsewhere, and obsidian (for making blades) was imported from the Willaumez Peninsula of New Britain. There was thus a move away from mobile hunting and gathering in the direction of what Spriggs (1997a: 61) calls ‘wild food production’ and others label ‘foraging sedentism’ . Bringing resources to people (rather than people moving from resource to resource) allowed a community a degree of sedentariness not available to hunters and gatherers. In pre-Lapita NW Island Melanesia, foraging sedentism evidently included activities like tending trees and transplanting seedlings.
LeBlanc (2002) argues that for a population to adopt agriculture, it is necessary for it already to be sedentary. If NW Island Melanesian populations encountered by arriving Austronesian speakers in New Britain, New Ireland, Manus, Buka and Bougainville did not already practise agriculture, they were almost certainly sedentary and would probably have shifted fairly rapidly to agriculture. The language map of New Britain, New Ireland and the Solomons offers circumstantial support for a hypothesis that these populations either did not have agriculture or were agriculturally inferior to the new arrivals. There are today tiny scattered groups each speaking its own Papuan language, surrounded by numerous Oceanic languages, a number of which show signs of Papuan influence in their phonology or grammar.6 The implication of this linguistic geography is that there were once far more Papuan languages in NW Island Melanesia, but as their populations adopted agricultural practices from their Austronesian-speaking neighbours they also adopted their languages.
This does not mean that these populations simply shifted mode of subsistence. Foraging sedentism also seems to be a necessary precursor of agriculture (Cohen 2002), and we have no reason to suppose that populations acquiring agriculture simply abandoned their previous practices. Rather it is a reasonable inference that they retained their hunting and tree-tending practices and combined these with newly acquired cultivation practices. It is thus possible that the hunting and tree-tending practices of early Lapita culture reflect at least in part an inheritance from pre-Lapita inhabitants of the Bismacks. What is certain is that about 1.3 kg of Canarium indicum and about 130 grams of Terminalia have been found in Lapita archaeological contexts in the Arawe Islands (off the southwest coast of New Britain) (Matthews & Gosden 1997). This is of course not direct evidence of Lapita tree-tending, but it is direct evidence that canarium nuts in particular were an important item of consumption among POc speakers. Kirch (1989) found circumstantial evidence for arboriculture in a Lapita assemblage on Mussau Island which included shells of Aleurites moluccana, Burckella, Canarium, coconut, Spondias cytherea, and Terminalia, pericarp of Inocarpus fagifer and seeds of Pometia pinnata.
We cannot take it for granted that agriculture in the POc period was the same as it is today, but there is reasonable evidence that it was not very different. Presumably the area of forest that had been turned into gardens was less than it is today, and the areas of anthropogenic vegetation habitats were therefore also less extensive. But linguistic evidence (§5) indicates that bush fallowing was already well established in NW Island Melanesia by the time POc broke up into daughter-languages.
POc had terms denoting the contrast between land with natural primary vegetation and land cleared for gardening:
In a few languages the reflex of *qutan means ‘garden’ (Mussau utana, Banoni ɣitana, Tolo uta, Nese na-ut), a seemingly odd reversal of meaning - odd until one recognises that a secondary sense of *qutan was ‘inland’ (vol.2, ch.8, §2.2.1), itself a natural outcome of the fact that the bush is always inland in relation to a coastal village. Since gardens were also inland from a coastal village, and sometimes quite a distance inland, the inference that *qutan is a metonymous term for ‘garden’ entails only a small jump.
Contrasting with *quma along the bush fallowing cycle was POc *talu(n) ‘old garden, fallow land, land returning to secondary growth’ (vol.1, ch.5, §3.1; vol.2, ch.3, §5.1).
The term *qutan appears to have denoted uncultivated land in general, whether it was *talu(n), land returning to secondary growth after cultivation, or bushland that had never been brought under cultivation. The term for the latter appears to have been POc *[waRu]waRu- ‘appears’ because there are phonological problems in the data. On one hand Misima, Proto Malaita-Makira and PPn all reflect the reconstructed form regularly, assuming that final *-o of PPn *wao reflects assimilation. On the other hand PMic medial *-l- reflects POc *-l-, not *-R-. This irregularity is unexplained, but cannot be dismissed too lightly, as Proto Malaita-Makira *[walu]walu could also reflect a form with POc *-l-. I reconstruct POc *[waRu]waRu because this explains a larger range of reflexes, but further data might lead to a revision.7
| POc | *[waRu]waRu | ‘primary forest’ (French-Wright 1983: *wao) 8 | |
| PT | Misima | walu-walu | ‘the bush’ |
| Proto Malaita-Makira | *[walu]walu | ‘the world; uncultivated bush’ | |
| SES | Kwaio | kʷalu | ‘unusable bush’ |
| SES | Sa’a | walu(malau) | ‘the world, all the islands’ |
| SES | Arosi | waru-waru | ‘the inhabited world generally, all the known islands’ |
| PPn | *wao | ‘forest’ | |
| Pn | Tongan | vao | ‘forest, bushland, scrub, land in its natural uncultivated state’ |
| Pn | Tahitian | vao | ‘wilds, wilderness’ |
| Pn | Māori | wao | ‘forest’ |
| SES | Gela | ao | ‘forest, land never brought under cultivation’ |
| PMic | *walu | ‘vegetation, forest’(Bender et al. 2003) | |
| Mic | Ponapean | wāl | ‘jungle, forest’ |
| Mic | Mortlockese | wali-wel | ‘forest’ |
| Mic | Chuukese | wənɨ-wən | ‘vegetation, bush’ |
| Mic | Puluwatese | wāl | ‘forest, jungle’ |
| Mic | Carolinian | walɨ-wal | ‘forest, jungle’ |
| Mic | Carolinian | wal(lap) | ‘dense: forest, thickly forested area’ |
| Fij | Rotuman | vao | ‘forest, large number of trees or big plants growing together’ |
In volumes 1 and 2 a number of other POc terms for landscape features were reconstructed which correspond very roughly to the vegetation habitats listed in §3. The correspondence is rough because the POc terms denote landscape features, not vegetation habitats in the botanical sense. Such terms are POc *(b,bʷ)iker ‘beach, esp. sandy beach’ (vol.2, ch.3, §3.1)9 and the three terms reconstructed in vol.2, ch.3, §5.2, denoting swamps. As we note there, we have no principled way of determining which te:rm(s) denoted mangrove swamps and which freshwater swamps.
A minor piece of evidence that POc speakers recognised a division between wild and cultivated varieties of plants is that wild plant names are sometimes fully reduplicated forms of a cultivated variety or a similar cultivated plant species (§7.2).
People in traditional Oceanic-speaking villages ate one cooked meal a day, usually after the day’s work. The meal typically consisted of starchy staples, made more appetising by the addition of coconut milk (ch.12, §4.2), leafy vegetables (ch.10) and sometimes some meat or fish. Today the meal is most often boiled. Traditionally, boiling would have been common in communities which had clay pots. Will McClatchey (pers. comm.) points out that it was also possible in communities without clay pots: a stone was heated in the fire and dropped into a container of water. Such containers might be (watertight) leaf baskets, a wooden bowl, a ground out rock or a large shell. Since boiling with such utensils can still be observed in parts of NW Island Melanesia, there is good reason to think that it occurred traditionally. Food was presumably also roasted in the fire or baked over it or - on special occasions - steamed in an earth oven (for food preparation methods, see vol.1, ch.6).
The lexicons of Oceanic languages usually distinguish two main categories of ingredient:
Barrau (1955: 44) remarks on the existence of this division in New Caledonia, and translates the terms for the categories as ‘food’ and ‘condiments’.
Dictionary definitions do not always mention all three components of the ‘additional ingredients’ category, and it is difficult to know whether the definitions are sometimes deficient or whether the composition of the additional ingredients varies from one place to another. Among Takia speakers living inland on Karkar Island, for example, the additional ingredients often lack a protein food if no one has been hunting, as domestic animals (pigs and chickens) are slaughtered only for feasts.
The encoding of the two categories of ingredient in the lexicon varies from language to language, as is shown by the terms below from widely dispersed languages. Oceanic languages have quite complex terminologies to do with eating, and the terms listed below are limited to generic terms for eating and food and to terms which presuppose the distinction between staples and additions as categories of meal ingredient. I have omitted, for example, terms for ‘meat’ where these have the wider meaning ‘flesh’.
| VERBS | -ɣa | ‘eat’ (POc *kani) |
| -rɔm | ‘eat mixed food including meat’ | |
| V&N | tmul | ‘mix various kinds offood together and cook them’ (V); ‘a mixed stew’ (N) |
| NOUNS | nos | ‘food (generic), but not including meat or greens’ |
| rɣu | ‘meat food’ |
| VERBS | _-a | ‘eat’ (POc *kani) |
| -a-kʷayakʷaya | ‘eat meat alone’ (kʷayakʷaya- ‘white’) | |
| -kuda-taʔula | ‘eat starchy vegetables and meat together’10 | |
| v&N | [-]onanaga | ‘crave for meat’ (v); ‘craving for meat’ (N) |
| NOUNS | aʔa | ‘food (generic), cultivated crops’ |
| kevakeva | ‘meat and fish, protein food’ |
| VERBS | ian | ‘eat’ (VT) (POc *kani) |
| gama | ‘mix meat with starchy food’ | |
| bite, bui | ‘crave for meat’ | |
| NOUNS | ni-an | ‘food’ (ni- NOMINALISER) |
| g⟨in⟩ama | ‘meat mixed with starchy food’ (⟨in⟩ NOMINALISER) |
| VERBS | ani, ŋani | ‘eat’ (VT) (POc *kani) |
| NOUNS | namnam | ‘food’ |
| gemnai | ‘eat (s.t.) as an accompaniment to starchy food’ | |
| balbal | ‘starchy food, root vegetables’ | |
| gemgem | ‘meat, meat animals’ |
| VERBS | an | ‘eat’ (VT) (POc *kani) |
| wəŋan | ‘eat’ (VI) (POc *paŋan) | |
| naŋin | ‘eat starch and meat together’ | |
| bet | ‘eat meat alone’ | |
| odo | ‘eat greens alone’ | |
| NOUNS | ni-an | ‘food’ (ni- NOMINALISER) |
| VERBS | ŋau | ‘eat’ (vT) (POc *ŋau ‘gnaw, chew’) |
| maŋa | ‘eat’ (vi) (POc *paŋan) | |
| ŋau-koŋari | ‘eat one thing without relish’ (koŋari ‘empty’) | |
| ʔonari | ‘eat only fish’ | |
| ŋau-bʷara-bʷara | ‘eat one thing with relish’ (bʷara-bʷara ‘fern sp.’) | |
| mamu | ‘eat two kinds of food together’ | |
| ŋau-kokona | ‘eat only greens’ (kokona ‘smooth, slippery’) | |
| NOUNS | hinaŋa | ‘starchy foods’ |
| VERBS | ɣiñ | ‘eat’ (vT) (POc *kani) |
| haŋ | ‘eat’ (vi) (POc *paŋan) | |
| topʷ-haŋ | ‘eat starch without additions’ (topʷ ‘just, merely, only’) | |
| leɣleɣ | ‘eat meat or fish without starch’ | |
| N& v | aθepyañ | ‘eat meat or fish with taro’ (vi, vT); ‘meat or fish eaten with taro’ (N) |
| NOUNS | in-haŋ | ‘food, meal’ (POc *paŋan) |
| ni-tai-ɣiñ | ‘food (generic)’ (lit. ART-thing-eat) (POc *kani) | |
| ni-tai-haŋ | ‘food (generic)’ (lit. ART-thing-eat) (POc *paŋan) |
| VERBS | -eni | ‘eat’ (vT) (POc *kani) |
| -vaŋ | ‘eat’ (vi) (POc *paŋan) | |
| -elat | ‘eat meat or fish’ | |
| -aŋot | ‘hungry for meat’ (lit. ‘itch’) | |
| -etki | ‘eat meat or fish with starchy food’ 11 | |
| NOUNS | n-vaŋ | ‘food’ (POc *paŋan) |
| potninvaŋ | ‘staples, including yam and banana’ (< poti-n n-vaŋ ‘base of food’) |
| VERBS | kɨmɨs | ‘eat’ |
| kʌfa | ‘eat’ | |
| kiyɔyɔi | ‘eat one kind of food without side dish’ | |
| kokɒɛk | ‘eat main dish without side dish’ | |
| V&N | moŋo | ‘eat’ (v); ‘food’ (N) (PMic *mʷaŋau ‘eat (staple food)’ (v); ‘staple food’ (N)) |
| ɛnʌt | ‘eat a side dish of (s.t.)’ (v); ‘side dish’ (N) | |
| NOUNS | kiyɔyɔ | ‘one kind of food eaten without side dish; naked’ |
| ɛnte | ‘side-dish’ |
| VERBS | kani | ‘eat’ (vT) (POc *kani) |
| tovi | ‘crave for meat or fish’ (vT) | |
| NOUNS | maŋiti | ‘food, something to eat; staples (root crops, bananas, breadfruit)’ |
| ilava | ‘savoury food, preferably fish or meat, eaten with staples’ | |
| vīdāgʷana | ‘protein food, especially seafood, and leafy vegetables eaten with staples’ |
| VERBS | *kai | ‘eat’ (POc *kani) |
| *samu | ‘eat only protein food’ | |
| NOUNS | *kina, *kiki | ‘food eaten as a relish with other food’ 12 |
Aside from POc *kani and *paŋan,13 which are widely reflected, the sets of semantically similar forms above contain no cognate sets that justify POc reconstructions. Other cognate sets are decidedly localised but similar meanings recur:
| 1 | (V) | ‘eat starchy food; eat (generic)’ | (all) |
| (N) | ‘starchy vegetables’, also used hypemymously for ‘food (generic)’ | Mapos Buang, Sursurunga, Arosi, Wayan | |
| 2 | (V) | ‘eat food consisting of starch and additional ingredients (coconut milk, leafy vegetables, fish or meat)’ | Mapos Buang, Ramoaaina, Arosi, Anejom |
| (V) | ‘mix and/or cook food consisting of starch and additional ingredients’ | Mapos Buang, Iduna, Patpatar | |
| (N) | ‘food consisting of starch and additional ingredients’ | Patpatar | |
| 3 | (V) | ‘eat (s.t.) as an addition to starch’ | Sursurunga, Anejom, Sye, Kosraean |
| (N) | ‘fish or meat and leafy vegetables as an addition to starch’ | Anejom, Kosraean, Wayan, PPn | |
| 4 | (V) | ‘eat fish or meat without starch’ | Iduna, Ramoaaina, Arosi, Anejom, Sye, Kosraean, PPn |
| (V) | ‘crave fish or meat’ | Iduna, Patpatar, Sye, Wayan | |
| (N) | ‘fish and meat as food’ | Mapos Buang, Iduna, Sursurunga | |
| 5 | (V) | ‘eat greens alone’ | Ramoaaina, Arosi |
The meanings above are categorised into those referring to (1) starchy vegetables and hypemymously to food in general; (2) food consisting of starch and additional ingredients (coconut milk, leafy vegetables, fish or meat); (3) the additions; (4) fish and meat without starch; (5) leafy vegetables.
In the languages above, only two verbs mean something like ‘eat starchy foods’. They are Arosi ŋau-koŋari ‘eat one thing without relish’ and Anejom topʷ-haŋ ‘eat starch without additions’. Their literal meanings are instructive: Arosi ‘eat empty’, Anejom ‘just eat’. They imply that the basic meaning of the verb ‘eat’ is ‘eat starchy foods’, and this appears to be true in other Oceanic languages too. All the verbs meaning ‘eat’ except Arosi ŋau (< POc *ŋau ‘gnaw’) are reflexes of PMP *kaen / POc *kani ‘eat’, and all apparently refer primarily to eating starchy foods (although few dictionaries mention the fact). They have the generic meaning ‘eat’ by hypernymy. This inference is supported by the facts that (i) other verbs of eating never have this meaning; (ii) the form for ‘starchy foods’ in many Oceanic languages is or reflects a nominalisation of the reflex of this etymon; and (iii) reflexes in New Caledonian languages, e.g. Voh-Kone cani, Xaracuu kɛ̃, still mean ‘eat carbohydrates, eat tubers’.
A verb with the meaning ‘crave fish or meat’ is often found in Oceanic languages, reflecting the fact that in Oceanic speaking communities the main meal is often eaten without fish or meat.
Fewer languages appear to have terms referring to eating leafy vegetables alone. Ramoaaina odo and Arosi ŋau-kokona are the only instances found in a survey of Oceanic dictionary sources. In Ross (1996d) I distinguished two categories of meal ingredient, starchy staples and leafy greens. With more dictionary data, it has become clear that the primary categorisation of meal ingredients is into starch and additional ingredients, and that leafy vegetables are a subcategory of ‘additional ingredients’.
The ubiquitousness of verbs and nouns which presuppose the ‘starchy food’ (§6.1) and ‘additional ingredients’ (§6.2) categories suggests strongly that these categories were already present in POc.
Staple food sources are easily grown starchy foods of vegetable origin that are high in food energy. Dictionary definitions of terms for ‘staple food’ or ‘starchy food’ in Oceanic languages are rarely exhaustive, sometimes referring to yams, sometimes to taro, sometimes to root crops in general. This is, of course, often due to variations in the staples consumed from one area to another: taro is grown in wetter regions, yams in drier areas, and breadfruit on atolls. For New Guinea Bourke (in preparation) lists Colocasia taro, yams (Dioscorea esculenta and Dioscorea alata), banana and sago as the most important staple foods before the arrival of the sweet potato. An early botanical account suggests that taro, some yam species, bananas and breadfruit probably formed the traditional staple foods of Pacific islanders (Guppy 1906: 412-415). Pawley & Sayaba (2003) define Wayan maŋiti as including the staples bananas and breadfruit as well as root crops, and knowledge of Oceanic eating habits justifies assuming definitions of this kind of breadth for the corresponding terms in many Oceanic languages. The sago palm is included with other staples in chapter 9 because it serves as a staple in some Oceanic societies and as a famine food in others.
Two POc forms are reconstructed with the probable meaning ‘starchy food’ and hypernymously ‘food in general’: *kanaŋ and *kuta.
POc *kanaŋ and its variant *kanan, both ‘staple food, food in general’, reflect PMP *kanan ‘dish, plate, meal’. This was a nominalisation of PMP *kaen ‘eat’, or *kan in the context of certain affixes. PMP had several nominalising affixes, among them *-en and *-an. PMP *-en was reflected in POc only in a few fossilised contexts like *kanoŋ ‘flesh, meat, coconut flesh’ (ch.12, §4.2), from PMP *kan-en ‘something to be eaten, food’. PMP *-an ‘locative nominaliser’, on the other hand, became POc *-an/*-aŋ, a productive nominalising affix with wider functions than in PMP. Because of this productivity, as the nominalising forms and strategies of Oceanic daughter-languages changed in various ways (vol.1, ch.2, §3.2.1), so lexical items formed by nominalisation sometimes retained their old forms and sometimes changed in accordance with the changes in nominalising morphology. Forms which reflect such changes are listed below under ‘cf. also’.14
| PMP | *kan-an | ‘dish, plate, meal’ | |
| POc | *kanaŋ, *kanan | ‘staple food; food in general’ | |
| PAdm | *kanana | ‘food’ (Blust 1996b) | |
| Adm | Wuvulu | anana | ‘food’ (Blust 1996b) |
| Adm | Mondropolon | kanna | ‘food’ (Blust 1996b) |
| Adm | Drehet | kana | ‘food’ |
| NNG | Takia | anaŋ | ‘food’ |
| NNG | Takia | anŋ-anaŋ | ‘lesser yam, Dioscorea esculenta’ |
| NNG | Lukep | kana- | ‘share of food, provisions’ |
| NNG | Matukar | anan | ‘lesser yam, Dioscorea esculenta’ |
| NNG | Manam | kana- | ‘food’ |
| NNG | Mari | ganaŋ | ‘taro’ |
| NNG | Adzera | ganaŋ | ‘banana plant’ |
| MM | Bulu | (ɣani)ɣana | ‘coconut flesh’ |
| SES | Gela | ɣana | ‘food’ |
| Fij | Bauan | kana | ‘meal’ |
| Fij | Bauan | kā-kana | ‘food’ |
| PAdm | *kani-an | ‘staple food’ | |
| Adm | Nyindrou | kani-a | ‘staple food’ |
| PNNG | *kani-ŋa | ‘food’ | |
| NNG | Sengseng | kini-ŋ | ‘food in general; animal protein in particular’ (A. Chowning, pers. comm.) |
| NNG | Bariai | an-ŋa | ‘food’ |
| NNG | Lukep | kani-ŋ | ‘yam’ |
| NNG | Bing | ani-ŋ | ‘banana’ |
| NNG | Manam | ani-ŋa | ‘food’ |
| NNG | Mangseng | ani-ŋ | ‘food’ |
| NNG | Mengen | kani-ŋ | ‘food’ |
| NNG | Hote | ani-ŋ | ‘taro’ |
| PPT | *kani-kani | ‘staple food’ | |
| PT | Misima | an-an | ‘yams; root crops, nuts and fruit; food’ |
| PT | Taboro | ɣani-ɣani | ‘food’ |
| PT | Taboro | ɣani | ‘short cooking banana’ |
| PT | Hula | ani | ‘banana’ |
| PT | Motu | ani-ani | ‘food’ |
| PT | Mekeo | ani-ani | ‘food’ |
| PMM | *k⟨in⟩ani | ‘staple food’ | |
| MM | Nakanai | il-ali | ‘food’ (A. Chowning, pers. comm.) |
| MM | Patpatar | ni-an | ‘food’15 |
| MM | Ramoaaina | ni-an | ‘k.o. yam; food’ |
| MM | Tolai | ni-an | ‘food’ |
| MM | Nehan | ni-eini | ‘food’ |
| MM | Roviana | ɣ⟨in⟩ani | ‘food’ |
| PNCV | *k⟨in⟩ani-ana | ‘staple food’ | |
| NCV | Raga | ɣ⟨in⟩a-ɣani-ana | ‘food’ |
| NCV | Paamese | ani-ene | ‘staple food, as opposed to meat and greens’ |
| NCV | Lewo | k⟨in⟩ani-ena | ‘staple food, as opposed to meat and greens’ |
A further twist is reflected in the forms under ‘cf. also’ above. The base of PMP *kan-an was *kan. The base of the forms under ‘cf. also’ is POc *kani. This reflects PMP *kan-i, where *-i is a reflex of the PAn suffix *-i ‘location focus, atemporal’, reinterpreted as a transitiviser in POc (Pawley & Reid 1980; vol.1, ch.2, §§3 .1.2-3) but lexicalised as part of the POc base when the erstwhile base, e.g. *kan, was a monosyllable. As a result, the forms under ‘cf. also’ have the base *kani, which never co-occurred with a nominalising affix in PAn and PMP.
Several reflexes of POc *kuta denote a major staple in the language concerned, either banana or yam, suggesting that the POc term denoted ‘staple food’. However, it is also possible that *kuta was originally a specialised verb of eating, as the Gumawana, Iduna, Gela and Tolo glosses imply.
| POc | *kuta | ‘staple food’; ‘eat’ | |
| NNG | Lukep | kuta | ‘banana cultivar’ |
| NNG | Bing | (aniŋ) kuta | ‘sweet banana cultivar’ |
| PT | Gumawana | kuta | ‘chew sugar cane’ |
| PT | Molima | ʔuta | ‘chew’ |
| PT | Molima | ʔutaʔuta | ‘Dioscorea pentaphylla’ |
| PT | Iduna | kuda | ‘chew’ (-d- for †-t-) |
| PT | Ubir | ut | ‘greater yam’ |
| PT | Gapapaiwa | uta | ‘yam type’ |
| SES | Gela | kut-i | ‘feed’ (used as causative of vaŋa ‘eat’; Fox 1955) |
| SES | Tolo | kuta | ‘eat’ |
| SV | Kwamera | kə-kətə-n | ‘baked food’ |
The term POc *tamaji ‘additional ingredients to accompany starchy food’ is presented here only because it supports the claim that POc speakers divided meal ingredients into ‘starchy food’ and ‘additional ingredients’ (§6).
Reflexes of *tamaji with meanings supporting the reconstructed meaning are found in Mapos Buang (NNG), Gela, Longgu and Bauro16 (all SES) and perhaps Teop (MM). A larger number of reflexes support another meaning, ‘provisions for a journey’: they are found in Bariai, Takia (both NNG), Dawawa (PT) and Sursurunga (MM). However, these are all Western Oceanic reflexes, implying that this meaning may have developed in the west after Oceanic speakers had begun to spread out across Oceania.
| POc | *tamaji | ‘additional ingredients to accompany starchy food’ | |
| NNG | Bariai | tamad | ‘food for a journey’ |
| NNG | Mangap | temen | ‘small meal prepared as expression of gratitude’ |
| NNG | Takia | (la)tamad | ‘food for journey’ (la ‘move away from speaker; go around the island’; cf. vol.2, ch.8, §3.4.5) |
| NNG | Mapos Buang | tmul | ‘mix various kinds of food together and cook them; stew, soup’ |
| PT | Dawawa | tamasi-na | ‘food for travel’ |
| MM | Sursurunga | t⟨in⟩mas | ‘picnic lunch, food taken and eaten away from the house or on a journey’ |
| MM | Teop | tamari | ‘prepared food’ |
| SES | Gela | tamadi | ‘relish with vegetable food’; ‘eat together different sorts of food’ |
| SES | Gela | tamadi | ‘relish with vegetable food’; ‘eat together different sorts of food’ |
| SES | Longgu | amadi-a | ‘eat something as an accompaniment to something else’ |
| SES | Longgu | amadi-na | ‘food that is eaten as an accompaniment to something else’ |
| SES | Bauro | amasi | ‘eat two things together; use a relish, use betelpepper’ |
| SES | Arosi | amadi | ‘betelpepper’ |
| NCV | Lonwolwol | tamsi- | ‘small pieces of’ |
The concept of ‘additional ingredients’ was also lexicalised in PMic *ta(l,n)ia ‘side dish of meat, fish, or sauce’ (Bender et al. 2003: 89).
A number of Oceanic languages appear to use prefixes or reduplication in plant names as a means of categorising the plant thus named. However, this categorisation is more apparent than real. The prefixes described in §7.1.1 and §7.1.4 and the reduplication described in §7.2 reflect strategies for creating new plant names rather than direct reference to categories of plant. The prefixes described in §7.1.3 are part of the debris left by an earlier numeral classifier system.
Three kinds of prefix stand out in the history of some Oceanic plant names reconstructed in this book:
The first two categories appear at first sight to be rather similar, but the evidence suggests that they are derived from different POc constmctions and have different uses.
In several Western Oceanic languages some plant names begin with a prefix that means ‘tree or shrub’. Examples from three languages are given below, together with the protoforms which they reflect. The three prefixes are Yabem ka-, Muyuw a- and Patpatar i-, each of which is assumed to reflect POc *kayu ‘tree’. Yabem ka- occurs on most tree names, Muyuw a- on many, and Patpatar i- on just a few.17
| ka-toʔ | ‘mangrove’ | POc | *toŋoR | ch.6, § 3.1 |
| ka-dada | ‘k.o. grassland shrub’ | POc | *jajal | ch.13, §6.4 |
| ka-bʷɛŋ | ‘ironwood’ | PWOc | *bʷana | ch.7, §4.9 |
| ka-maʔ | ‘Cordyline sp.’ | PWOc | *mʷa(r,R)ep | ch.13, §6.2 |
| ka-bɔʔ | ‘k.o. mangrove tree’ | PWOc | *baul | ch.6, §2.1 |
| a-mʷakot | ‘Dysoxylum spp.’ | POc *maqota | ch.7, §4.5 |
| a-yayak | ‘Myristica schleinitzii’ | POc *(d,r)aRa(q,k)(a,i) | ch.7, §5.9 |
| a-gi-gaway | ‘Ficus tinctoria’ | POc *qayawan | ch.10, §4 |
| a-nag | ‘Cordia sp.’ | PWOc *nagi | ch.5, §4.1.1 |
| a-simʷal(gayas) | ‘Glochidion sp.’ | PWOc *ji(m,mʷ)(a,i)R | ch.8, §2.4 |
| a-kobʷow | ‘Macaranga tanarius’ | PWOc *kobo | ch.7, §2.5 |
| i-nas-nas | ‘Tournefortia argentea’ | POc | *na[su]-nasu | ch.5, §4.1.5 |
| i-walas | ‘Semecarpus forstenii’ | POc | *[wa]lasi | ch.7, §6.1.6 |
| i-kon | ‘Heritiera littoralis’ | POc | *kayu (ni) qone | ch.6, §4.4 |
A simple preliminary hypothesis is that these prefixes label their denotata as members of the POc *kayu taxon, ‘trees and shrubs’ (ch.3, §4.2). But if this were so, we would expect to find the prefix on every tree name in these languages and we would also expect to find prefixes reflecting the other primary plant taxa, *waroc ‘vine’, *pali[s,j]i ‘grass’, *taliŋa ‘mushroom’ and *l[i,u]mut ’moss, algae. Neither expectation is fulfilled.
Data from other Oceanic languages suggest an alternative hypothesis, namely that a reflex of *kayu occurs in cases where the following root is in some way descriptive of the tree, i.e. cases parallel to English ‘flame tree’, ‘rain tree’, ‘coral tree’, ‘bead tree’ and ‘canoe tree’, all coinages naming trees of the Pacific. Among the data above Patpatar i-kon ‘Heritiera littoralis’ is a strikingly obvious example of this. The cognate set below is presented in ch.6, §4.4, and it is argued there that it reflects POc *kayu qone, which can be glossed ‘beach tree’ (*qone ‘beach’).
| POc | *kayu qone | ‘Heritiera littoralis’ | |
| MM | Patpatar | i-kon | ‘Heritiera littoralis’ |
| MM | Tolai | ka-kono | ‘Heritiera littoralis’ (Record 1945) |
| NCal | Nyelâyu | kʰon | ‘Heritiera littoralis’ |
The two Oceanic languages for which the best information about plant names is available are the SE Solomonic language Kwara’ae and Wayan Fijian.18 Only a small number of Wayan tree names begin with a reflex of POc *kayu. Pawley & Sayaba (2003) list the seven entries below.
| kaidam | ‘wild nutmeg, Myristica chartacea’ | dam | ‘turn reddish’ |
| kaidrisi | ‘wild nutmeg, Myristica chartacea’ | drisi | ‘be reddish’ |
| kailalālau | ‘thorny shrubs, Caesalpinia spp.’ | lau | ‘be pricked, wounded’ |
| kailō | ’small trees, Diospyros spp. ” | lō | ‘be dark, secretive’ |
| kaimoðemoðe | ‘mimosoid tree taxon’ | moðemoðe | ‘have a short sleep’ |
| kaimoku | ‘creeper, Mimosa pudica’ | … | |
| kaivula | ‘tree, probably Endospermum macrophyllum’ | vula | ‘moon’ |
Significantly, for six out of seven the dictionary independently lists a meaning for the root. For four the descriptive meaning is clear. Myristica chartacea has red sap: dam and drisi mean respectively ‘turn reddish’ and ‘be reddish’. The term kaimoðemoðe denotes Albizia saman and Serianthes vitiensis, the leaflets of which fold together at night: moðemoðe means ‘have a sleep’. Endospermum macrophyllum Euphorbiaceae is a timber tree with pale yellow wood, which perhaps accounts for the name kai-vula, literally ‘moon tree’.
Kwa’ioloa & Burt (2001) list thirty-six names of trees, big and not so big, beginning with ʔai-. Most of these are explicitly descriptive, according to Kwa’ioloa and Burt. The first eleven are listed below, and nine out of eleven are descriptive.
| ʔaibū | ‘tree, Diospyros ebenum’ | … | |
| ʔaisarufi | ‘tree, Eugenia effusa’ | sarufi | ‘big tree, Litsea alba’ |
| ʔaisaliŋa | ‘big tree, Aporosa papuana’ | aliŋa | ‘ear’ |
| ʔaisubu | ‘big tree, Pimelodendron amboinicum’ | … | |
| ʔaiuluulu | ‘big tree, Vitex cofassus’ | uluulu | ‘bushy’ |
| ʔaikame | ’big tree, Putranjiva roxburgii | kame | ‘monitor lizard’ |
| ʔaikusi | ’big tree, Cryptocarya alleniana” | kusi | ‘greybird’ |
| ʔaikaʔo | ‘big tree, Xylopia papuana’ | kaʔo | ‘bamboo’ |
| ʔailali | ‘Tahitian chestnut, Inocarpus fagifer’ | lali | ‘kidney’ |
| ʔairade | ‘trees, Dysoxylum’ spp. | rade | ‘stink’ |
| ʔaininiu | ‘big tree, Horsfieldia irya’ | niniu | ‘palm, Gulubia macrospadix’ |
In two cases, ʔaisarufi and ʔaininiu, the tree is named for a resemblance to another plant. The tree ʔaisaliŋa, literally ‘ear tree’, is so named because the leaf has ‘ears’ at its base. Two reasons are given for the name ‘monitor lizard tree’, ʔaikame: its bark is said to be like the skin of a monitor lizard, and the monitor lizard is said to take refuge in this tree when it is chased by a dog. The ‘greybird tree’, ʔaikusi, is so named because its leaf bloom is like the feathers of a kusi bird.
The most striking feature of Kwara’ae ʔai- is its distribution: it is never prefixed to a reflex of a POc tree name. This implies that it is used productively for descriptive coinages and that it does not cooccur with roots that are felt to be tree names in their own right.
The inference to be drawn from these observations is that POc had a construction [*kayu DESCRIPTOR] which was used for coining tree names. Kwara’ae and Fijian evidence suggests that [*waRoc DESCRIPTOR] also occurred (*waRoc ‘vine, creeper’ > Wayan wā, Kwara’ ae kʷalo), e.g. Wayan wā bitubitu ‘creeper taxon of strong-stemmed vines’, literally ‘bamboo creeper’; Kwara’ae kʷalo kakali ‘wild passionfruit, Passiflora foetida’, literally ‘Hornstedtia vine’, because its seeds resemble those of the plant Hornstedtia lycostoma. Like Kwara’ae ʔai-, Kwara’ae kʷalo is used productively for new coinages like the name of the introduced passionfruit.
Among POc plant names trisyllables are rarer than disyllables, reflecting a pattern in POc stems generally. The first syllable of a number of trisyllables is reconstructed as *ka-, with perhaps greater than chance frequency, and it seems probable that at least some of these also reflect the construction with *kayu posited above. Examples are:
| POc | *kanawa(n) | ‘Cordia subcordata’ | ch.5, §4.1.1 |
| POc | *katita | ‘the putty nut’ | ch.7, §5.10 |
| POc | *kasiala | ‘a palm, Caryota sp.’ | ch.7, §5.11.1 |
| POc | *kalaka | ‘Planchonella sp.’ | ch.7, §4.10 |
| POc | *kapika | ‘Malay apple’ | ch.11, §3.7 |
| PWOc | *kasuwai | ‘mango’ | ch.11, §3.4 |
| PWOc | *kapu(r,R)ik | ‘k.o. wild melon’ | ch.13, §7.4 |
| POc | *[ka]ŋaRi | ‘canarium almond’ | ch.11, §2.1 |
| POc | *[ka]timun | ‘cucurbit (generic)’ | ch. 13, §7.4 |
The inference that *ka- was indeed a prefix is strengthened by the fact that in the last two cases there are reflexes of forms with and without *ka-, i.e. *kaŋaRi vs *ŋaRi, *katimun vs *timun, implying either that *ka- was an omissible prefix, or that *ka- was part of the stem but was reanalysed as an exemplar of the prefix and hence deleted.
The Bola (MM) reflex of POc *[ka]ŋaRi is taŋari, implying either replacement of one (perceived) prefix by another or some form of wordplay. The Kairiru (NNG) reflex of *kalaka is lalak, apparently a reduplication of suffixless *laka. POc *[ja]latoŋ ‘nettle tree’ has reflexes with and without POc *ja-, and *ja- is replaced in PNCV by *ga- (ch.7, §6.3.2). Again this implies prefix replacement or wordplay.
In certain NCV languages tree names occur with one of two prefixes. In Tamambo, stems naming trees and large bushes (but not other plant types) are regularly prefixed with vu- ‘tree’, e.g. vu-mambue ‘chestnut tree’, vu-niu ‘coconut palm’, vu-ŋai-ŋai ‘canarium nut tree’ (Jauncey 1997). Vines, ferns, tubers and grasses are not prefixed with vu-, but Jauncey observes that ‘some large kinds of bushes are marked as trees, but only if they do have a main central trunk’. This supports her inference that vu- reflects POc *puqu(n) ‘base of tree; source, origin’ (ch. 4, §2.1).
Contrasting with vu- is the prefix ra- ‘leaf of (ROOT)’, reflecting POc *raun ‘leaf’ (ch. 4, §2.5), e.g. ra-ɣaviɣa ‘leaf of Malay apple’, ra-ɣatabola ‘leaf of Dracontomelon vitiense’, ra-moli ‘leaf of citrus tree’.
Both prefixes derive countable units. Thus [_vu-_ROOT] means ‘a tree of the kind denoted by ROOT’ and [_ra-_ROOT] means ‘a leaf of the kind of tree denoted by ROOT’. This interpretation receives support from the fact that the prefixes also occur with xai ‘tree, wood’: vu-xai is ‘a tree’, i.e. vu- denotes the unit and xai the nature of the unit. POc *puqu(n) meant ‘base of tree’, but by metonymy acquired the sense ‘a tree-like unit’. François (2002: 50) reports that the Araki prefixes vi- and da- have similar functions to Tamambo vu- and ra- respectively.
The corresponding forms in Fijian languages preserve largely unchanged the POc construction from which Tamambo vu- and ra- are derived. In Wayan Fijian we find the construction [NOUN ni NOUN]. In the broadest terms, the second noun serves as an attribute of the first. Thus in a phrase like rau ni kulu ‘a breadfruit leaf’, kulu ‘breadfruit’ specifies the type of rau ’leaf. For example:
Like Tamambo vu-, Wayan vū ‘base, bottom’ is used metonymically to refer to whole trees. As a result, like Tamambo xai, Wayan kai ‘tree, shrub’ can serve as the second noun: vū ni kai ‘a tree/shrub’, rau ni kai ‘a leaf’, vua ni kai ‘fruit of tree’ (i.e. ‘a fruit’ ), tiki ni kai ‘piece of wood, stick’ (tiki ‘part, piece’).
Wayan, like other Fijian languages, here preserves a POc noun phrase construction which had a variety of functions. Sometimes called the ‘associative’ construction in the literature of Oceanic linguistics, this construction allowed one noun to be used as an attribute modifying another. It had two forms, the choice between them depending on whether the first (head) noun was a zero-valency noun or a monovalent noun.19
| Kwara’ae | Kwaio | Lau | PPn | POc | ||
|---|---|---|---|---|---|---|
| compact objects | faʔi | fe?e | … | *foqi | *puaq qi | ‘fruit’ |
| people | … | … | … | *toko | *tau + ? (ni) | ‘person, body’ |
| wooden objects | … | … | ʔai | … | *kayu (ni) | ‘tree’ |
| ‘head’ | gʷaʔi | gʷe?e | … | … | *pʷatu qi | ‘head’ |
| flat objects | ʔaba | … | … | … | … | |
| vertical objects | baʔe | … | … | … | … | |
| containers | taʔe | … | … | … | … | |
| pieces of | afu | me?e | … | … | … | |
| units of class | maʔe | māʔe | … | … | *mata qi? | ‘eye’ |
| leaves | raʔi | (gāʔe) | … | … | *raun qi | ’leaf |
| clumps | fiʔi | fūʔi | fi, fui | *fu(h)i | PEOc *pu qi (?) | ‘clump’ |
In the first variant of the construction, the head noun is a zero-valency noun:
| *a | polo | ni | niuR | |
| ART | juice | of | coconut |
The second noun is a generic (non-specific) possessor, and thus an attribute, of the first (head) noun.
In the second variant the head noun was monovalent and in consequence ni was replaced by *qi:
| *a | qaqe | qi | boRok | |
| ART | leg | of | pig |
In Fijian languages reflexes of *ni and *qi have been redistributed but the construction otherwise survives unchanged. This construction has a complex history in Oceanic languages (Hooper 1985, Ross 1998b, 2001b).
We can infer, for example, that a particular fruit tree of the species Syzygium malaccense, the Malay apple, would have been referred to as *(a) puqu(n) ni kapika (*a ARTICLE; kapika Syzygium malaccense, ch.11, § 3. 7), that a leaf and a fruit of the species would have been referred to respectively as *(a) raun ni kapika and *(a) puaq ni kapika or, if the head nouns were directly possessed in POc, as *(a) raun qi kapika and *(a) puaq qi kapika. We can also infer that Tamambo vu- and ra- and their cognates in other NCV languages reflect a grammaticisation of this construction with considerable phonological attrition.
This POc construction is also reflected in Kwara’ae plant naming, but in a rather different way from NCV. Henderson & Hancock (1988: 277) report that the name of a small tree is optionally preceded by faʔi and the name of a plant that grows straight without branching (and some that do branch) by fiʔi, e.g. faʔi keto or keto ‘Macaranga spp.’, fiʔi arakai or arakai ‘yam sp., Dioscorea pentaphylla’. Kwa’ioloa & Burt (2001) translate fiʔi as ‘clump’, which is probably more accurate than Henderson and Hancock’s characterisation.
The morphemes faʔi and fiʔi belong to a set of classifiers described in Deck’s grammar and used principally for counting (1934: 7-10). These are listed in Table 2.1 along with corresponding forms in closely related Kwaio and Lau, the Proto Polynesian classifiers reconstructed in POLLEX, and the POc nouns from which they are derived.
Several of the Kwara’ae classifiers share the form monosyllable + -ʔi or -ʔe. Their common form reflects their shared origin in the POc construction [MONOVALENT NOUN *qi NOUN] described above.20 The likely POc usage from which the classifiers in Table 2.1 are derived is illustrated below:21
| *i-tolu | puaq | qi | pudi | |
| 3sG-three | fruit | of | banana |
| *i-tolu | puqun | ni | pudi | |
| 3sG-three | trunk | of | banana |
Some classifiers did not combine with *qi, either because they remained disyllabic or because they were derived from zero-valency nouns which instead headed the POc construction [ZERO-VALENCY NOUN *ni NOUN].
The classifier faʔi is described by Deck as being used to count round, compacted or heaped units, but it seems to me that, like reflexes of PMP *buaq / POc *puaq ‘fruit’ (ch.4, §2.8) in a number of Austronesian languages, it is the default classifier for countable inanimate objects, i.e. it is used when no other classifier is more appropriate.22
Plant names which begin with the POc prefix *mala- ‘resembling’ occur quite frequently in Oceanic languages.23 This has evidently been an important means of forming new plant names, most often by exploiting the resemblance of one tree to another.24 As in the reduplicative process described below (§7.2), the tree denoted by a name with *mala- is generally inferior in some way to the one denoted by the plain root.
Examples from Nakanai (MM) are: mala-savula ‘a plant, Ficus sp.’ (< savula ‘the fruit of a tree, candlenut, Aleurites moluccana’); mala-sesege ‘a plant, Acrostichum aureum’ (< e-sesege ‘small black crablike shellfish (PLURAL)’); mala-viva-viva ‘a wild shrub, Clerondendron paniculatum, considered to be related to e-viva ’cultivated shrub with edible leaves, Abelmoschus manihot’. Similar examples are found in other Meso-Melanesian languages: Kara and Patpatar of New Ireland and Nehan of Nissan Island (Ross 2005b).
In Kwara’ae (SES), Kwa’ioloa & Burt (2001) note the following instances:
| mala-ŋali | ‘Canarium asperum’ | ŋali | ‘canarium nut, Canarium indicum’ |
| mala-ʔadoʔa | ‘Canarium harveyi’ | ʔadoʔa | ‘Canarium salomonense’ |
| mala-ʔafiʔo | ’Syzygium aqueum | ʔafiʔo | ‘Malay apple, Syzygium malaccense’ |
| mala-rufa | ‘Metrosideros parviflora’ | rufa | ‘Syzygium lauterbachii’ |
| mala-ʔasai | ‘wild mango tree, Mangifera mucronulata’ | ʔasai | ‘mango tree, Mangifera indica’ |
| mala-kona | ‘Burckella sorei’ | kona | ‘Burckella obovata’ |
| mala-dili | ‘a shrub, Dracaena angustifolia’ | dili | ‘a shrub, Cordyline fruticosa’ |
In Raga (NCV), Walsh (2004) notes mal-buliva ‘unidentified Ficus sp.’ (< buliva ‘Ficus aspera’), mal-ɣaviɣa ‘unidentified tree sp.’ (< ɣaviɣa ‘Malay apple, Syzygium malaccensis’), mal-walahi ‘unidentified tree sp.’ (< walahi ‘Semecarpus vitiensis’) and mal-bei-bei ‘unidentified Polyscias sp.’ (< bei ‘Polyscias sp.’).
Examples are also found in Fijian and Polynesian languages. Like reduplication as a process in forming plant names, POc *mala- as a plant-name formative receives support from evidence outside Oceanic which suggests that it is descended from a PMP form (Ross 2005b).
A reduplicated form (usually with CVCV-) in Oceanic often encodes the perception that the denotatum is inferior to or a diminutive of the denotatum of the unreduplicated form. In the case of plant names, reduplication often means that the denotatum is a wild variety of the cultivated plant denoted by the unreduplicated form. It seems likely that this reduplicative derivational process occurred not only in POc, but at least as early as Proto Malaya-Polynesian.
Thus in Dobu (PT) Arnold (1931) cites rabia ‘sago palm’ vs rabi-rabia ‘useless sago palm’, magi ‘areca palm’ vs magi-magi ‘useless palm resembling areca palm’, boro ‘taro’ vs boro-boro ‘wild taro’, udi ‘banana’ vs udi-udi ‘wild banana’.
From closely related Kilivila (Trobriand Islands) Ralph Lawton (pers. comm.) provides natu ‘a tree with edible fruit like mango’ (probably Burckella obovata - MR) vs gi-natu-natu ‘a tree with inedible fruit’, meku ‘a hardwood tree used for carving’ vs kai-meku-meku ‘a tree no good for carving’ (kai ‘tree’), seda ‘a nut tree’ vs seda-seda ‘a tree without nuts’.
The POc term ancestral to Dobu udi ‘banana’ was *pudi (ch.9, §3). Its reduplicated form *pudi-pudi is a candidate for reconstruction with the meaning ‘wild banana’, at least in PWOc.
| PWOc | *pudi-pudi | ‘wild banana’ (Ross 1996d) | |
| NNG | Kove | puri-puri | ‘wild banana’ (puri ‘banana’; A. Chowning, pers. comm.) |
| NNG | Mangap | pin-pin | ‘wild banana’ (pin ‘banana’) |
| PT | Dobu | udi-udi | ‘wild banana’ (udi ‘banana’) |
| PT | Sudest | ɣudu-ɣudu | ‘wild banana seeds’ (yudu ‘banana’) |
| MM | Ramoaaina | udu-udu | ‘wild banana’ (un ‘banana’) |
Other reduplicated forms in Ramoaaina are ləma ‘coconut tree or fruit’ vs ləma-ləma ‘wild coconut tree or fruit’ and bara ‘breadfruit’ vs bara-bare ‘wild breadfruit’. For nearby Patpatar Peekel (1984) lists tuh ‘sugarcane, Saccharum officinarum’ vs tuh-tuh ‘wild sugarcane, Saccharum spontaneum’, pulaka ‘Polynesian arrowroot, Tacca pinnatifida’ vs pulaka-pulaka ‘wild varieties of Polynesian arrowroot’, sier ‘betelpepper vine, Piper betle’ vs sier-sier ‘a vine, Piper fragile, Piper singkojan or Piper banksii’ .
Wayan Fijian has niu ‘coconut palm’ vs niu-niu ‘cycad, Cycas circinalis’ and vara ‘germinating coconut’ vs vara-vara ‘taxon of fleshy herbs, particularly orchids’.
Biggs (1991: 67-69) notes that reduplication was one of the devices used by the newly arrived Eastern Polynesian ancestors of the Maori to name New Zealand plants which resembled those they had known in their eastern Polynesian homeland. His examples include Proto Polynesian *futi ‘banana’ vs Maori huti-huti ‘sweet potato variety’, Proto Polynesian *kawa ‘Piper methysticum’ vs Maori kawa-kawa ‘Macropiper excelsum’ and Proto Polynesian *koli ‘tree or shrub with perfumed fruit’ vs Maori kori-kori ‘a buttercup, Ranunculus insignis’.
In Marovo (MM) several forms are found with a somewhat different semantic derivation. Here the reduplicated base denotes a feature which somehow characterises the plant denoted by the reduplicated form. Hence vose-vose ‘a tree of the deep forest, used for making paddles etc’ (< vose ‘paddle’), muta-muta ‘a forest tree with sap the smell of which causes vomiting’ (< muta ‘vomit’ ), and ta-talo ‘a sea plant with calcified leaves, Halimeda sp.’ (apparently < talo ‘taro’).