Chapter 3.9 Staple foods: root crops, bananas, breadfruit and sago

Malcolm Ross

1. Introduction⇫

A Proto Oceanic (POc) meal was typically made up of two categories of ingredient, *kanaŋ ‘staple food, starchy food’ (ch.2, §6.1) and *tamaji ‘additional ingredients to accompany starchy food’ (ch.2, §6.2). The main ingredients of *tamaji were - or rather are in today’s Oceanic societies - green vegetables, treated in chapter 10, and coconut cream, treated in §4.2 of chapter 12.

The present chapter treats staple foods. The chapter is divided according to the four categories of staple foods: root crops (§2), bananas (§3), breadfruit (§4) and stem starch, principally sago (§5).

2. Root crops⇫

There is no term for root crops in general in most Oceanic languages, and there was apparently no general term in POc. The only major staple root crops known to POc speakers were yams (Dioscorea spp., §2.1) and taro (Colocasia and Alocasia spp., §2.2).

Of the root crops found in the Pacific today, the widely consumed sweet potato (Ipomoea batatas), as well as cassava (Manihot esculenta) and American or Chinese/Hong Kong taro (Xanthosoma sagittifolium), are relatively recent introductions from South America, brought by Europeans to Indonesia, whence they spread to New Guinea.¹ They were certainly not present in NW Island Melanesia in POc times. Indeed, the sweet potato arrived in the Bismarck Archipelago and the Solomons only with European traders and settlers in the 19th century (Bourke forthcoming). Barrau (1965) suggests that these introduced crops relegated some less significant earlier staples to the status of famine foods: basing himself on Guppy (1906), he names Pueraria lobata, Dioscorea nummularia and Cycas rumphii as possible earlier staples.

Yams and taro differ in their distributions since yams, and especially Dioscorea alata, have more demanding environmental requirements. Unlike Colocasia taro, which thrives in generally rainy areas, in damp soil, and up to altitudes of 2200 m or so, yams require seasonal rainfall and flourish in deep, well-drained soils. They are grown in open gardens as they need sunlight. Although Dioscorea alata and Dioscorea esculenta require a similar environment, Dioscorea alata can be grown at altitudes up to 1900 m, Dioscorea esculenta only to less than 1000 m (M. Bourke, pers. comm.). Thus Ivens (1927) describes a difference in root crops among speakers of dialects of South Malaitan. On the island of Ulawa, only Dioscorea alata was cultivated. On Small Malaita Dioscorea esculenta was also grown on the low hills. On Malaita proper, people lived on the central ridge at 1000 m and had no yams, only taro.

Because of its environmental tolerance the taro tuber Colocasia esculenta is the most important subsistence plant in Melanesia (Barrau 1955: 50). When the ancestors of POc speakers arrived in the region Colocasia esculenta was already the dominant staple over much of New Guinea. In drier lowland and mid-altitude areas Dioscorea species predominated (Bourke 1982, Swadling & Hide 2005). However, it seems likely on circumstantial grounds that the yam Dioscorea alata was an important staple of POc speakers, at least immediately after their arrival in the Bismarcks.

POc speakers lived on small islands and in coastal enclaves (vo1.2, ch.2), and most of their environments were probably suitable for growing Dioscorea alata. Taro would have assumed greater importance as they moved inland on the larger islands. Significantly, one of the POc terms for Colocasia taro was *mʷapo(q), a borrowing from a mainland Papuan language (§2.2.1). Taro was certainly known to POc speakers - they also had their own Austronesian word for it, *talo(s) - but the presence of a Papuan borrowing for such a seemingly central item of diet implies that taro cultivation came to assume greater importance through contact with taro growers on the New Guinea mainland.

Both yams and taro allow some variety in their preparation for eating: they are boiled, braised or baked whole or in large chunks, and sometimes they are eaten grated or mashed.

A huge number of varieties of both yams and Colocasia taro (in the tens, if not the hundreds) are individually named in Oceanic-speaking communities (Malinowski 1935a: 100–104, Fox 1978, Barrau 1962: 99-100). Where more than one name for a species is reconstructed, it may be that one or other name denoted a variety. However, no deliberate attempt has been made here to reconstruct names for varieties, partly because sources do not provide Linnaean names for individual varieties and partly because their names are often transparently descriptive and have probably been innovated over and over again.

2.1. Yams, Dioscorea spp.⇫

Unlike the Araceae (taros and taro-like plants; §2.2) with their large leaves, yam plants form long vines which in the wild twine around tree trunks. When they are cultivated, the farmer usually provides a stake or frame for support. On Ulawa yams were planted communally only in newly cleared gardens.² The men felled the trees, cleared the ground and dug the holes, and the women planted the small tubers or tuber tops or slices (Ivens 1927: 10, 355-357). In other places yams are sometimes planted in old gardens, but only after they have lain fallow for a while, as yams need fertile soil. Generally there is only one yam crop per year, and tubers are harvested 9-12 months after planting. However, on Small Malaita and Ulawa yam tops were also planted as yams were consumed, providing an earlier occasional crop (Ivens 1927: 361).

Barrau (1955: 39–40) points out that yam-growing communities require a good understanding of the seasons. ‘Flowering and fructification of local plants often indicate to the natives the season best suited for the various agricultural activities.’ Thus on Espiritu Santo (Vanuatu) and in Fiji, yams are planted when the coral trees, Erythrina variegata (ch.5, §5.5), are in flower (Baker & Harrison 1936).

Preparing and planting a yam garden requires a major effort, and in many yam-growing communities the planting was associated with rituals and incantations calling on the spirits to stand guard over the crop, protecting it from thieves, yam beetles and other malign beings (Fox 1924: 104-106, Malinowski 1935a). On Ulawa the rituals began the night before planting by blessing the pieces of yam to be planted, followed in the morning by incantations in the garden and culminating with a small feast (Ivens 1927: 358-361). Weeding also had associated ritual. The women bathed, and then ceremonially cleansed themselves with Euodia sprigs (ch.7, §5.6) before they went to weed the yam garden. When the yams were ready for harvesting, the priest took to the beach two yams tied with Dracaena (ch.13, §6.1), where he sacrificed them to the sea spirits on behalf of the village. Only after this might the villagers harvest their yams as they needed them (Ivens 1927: 362-363). On both Small Malaita and Ulawa the yams were first placed in a temporary shelter near the garden, then transferred to a storehouse (Ivens 1927: 363-366). Malinowski’s (1935) account of yam cultivation among the Kilivila is similar to Ivens’ account of the Ulawa; the processes are the same, the rituals pervasive, but yam cultivation is strictly a male domain.

Because they are seasonal, communities dependent on fresh yams alone would undergo regular food shortages while the young yams were growing (Barrau 1955: 58). Before the advent of the sweet potato a few hundred years ago, the perils of shortage were avoided in two ways. First, yams were often co-staples with other starchy foods: in New Guinea societies with either Colocasia taro or bananas (Bourke 1982), in Vanuatu coastal communities with breadfruit (Walter & Sam 2002: 41). Secondly, by storage: of the root crops grown in NW Island Melanesia, yams can be stored longest, kept on shaded or dark shelves in a dry, well aired location for up to six months, providing food between harvest periods (Henderson & Hancock 1988: 19). Yam storage houses are a salient feature in Melanesian villages from the Trobriand Islands of Papua New Guinea to New Caledonia (Barrau 1955: 58).

2.1.1. Why so few yam terms?⇫

Today six yam species are found in Melanesia: Dioscorea alata, Dioscorea bulbifera, Dioscorea esculenta, Dioscorea pentaphylla, Dioscorea hispida, and Dioscorea nummularia (Barrau 1955: 56, Bourke 1982). The first four are present throughout the region, but POc names can be reconstructed for only the first two. This is curious, because the two main garden yam species in NW Island Melanesia today are the greater yam, Dioscorea alata, and the lesser yam, Dioscorea esculenta. We might expect to reconstruct a POc term for at least the first four species, and certainly for Dioscorea esculenta, but no term for the latter can be reconstructed with certainty. Barrau (1965) suggests that Dioscorea nummularia, which he found cultivated in scattered locations throughout the Pacific,³ was also once much more widely cultivated than it is today. No term can be reconstructed for it, but this is likely to be an artifact of the data: no source offers a term for it.

The lesser yam is agronomically superior to the greater yam and is in many areas a major source of nutrition, but it is the greater yam which is ceremonially significant and a source of prestige for the grower (Bourke 1982:55, in preparation and pers. comm.).⁴ From this one may infer that the greater yam is an ancient crop, whilst the lesser yam was cultivated at least a little more recently - but long pre-dates the sweet potato.

The linguistic evidence is consistent with Bourke’s inference. The POc etymon *qupi ‘greater yam, Dioscorea alata’ indicates that POc speakers had this crop, and the fact that it was also the generic term for yams suggests that it was the ‘default’ yam species in the POc economy. Because there is no widely distributed cognate set, no POc term for the lesser yam, Dioscorea esculenta, can be reconstructed. What does this imply? In its wild state the lesser yam seems to have long been a feature of the forests of NW Island Melanesia and was presumably present in the POc period.⁵ The most appropriate conclusion seems to be that when POc was breaking up it had not yet been extensively cultivated, and that it became a significant part of the diet only as early Oceanic speakers moved inland on the larger islands from their coastal and small-island habitats.

By the same token, the apparent existence of a POc term for Dioscorea bulbifera, the potato yam, implies that it was consumed by POc speakers, even though it is today cultivated only as a marginal crop and only in certain parts of New Guinea.

The visible differences between Dioscorea alata, Dioscorea esculenta and Dioscorea bulbifera are shown in Figure 9.1.

2.1.2. Dioscorea alata, greater yam, water yam, winged yam, TP yam, P iam (Dioscoreaceae)⇫

Left to themselves, Dioscorea alata tubers come in an amazing variety of forms, as shown in Figure 9.2. In various parts of Oceanic-speaking Melanesia - the Papuan Tip islands, Fiji, New Caledonia - farmers strove to produce yams with a regular form and giant length. Barrau (1955: 57) mentions a 2.6 m yam seen in New Caledonia in 1953. These long yams were produced by hollowing out a cylindrical growing hole and lining it with wooden sticks to provide growth space for the tuber. More generally, the yield is maximised by staking the growing vine with a support 2-3m high.

It is not nutrition that drives the desire to produce the largest and the most yams in Melanesian societies, but rather competition for prestige, carried out by exhibiting one’s success as a yam grower. Thus among Kilivila speakers of the Trobriands yams are displayed first in conical heaps in the garden, then again in front of the in-law’s storehouse, before being placed in one’s own storehouse. A good feast, from the host’s perspective, is one at which the guests eat their fill of yam, then see how full their host’s storehouse remains after the feast. A large yam with a fantastic form is good for display. The very fact that yams can be stored places them above taro in the food hierarchy because it allows a man to accumulate them until he can distribute them at a feast in the large quantities which bolster his prestige (Howes 2003: 67-68, 96-97).

The most widely reflected term for Dioscorea alata is POc *qupi, which is also widely reflected in a second sense as the generic term for yams.

PMP		*qubi	‘yam’ (Dempwolff 1938)
POc		*qupi	‘greater yam, Dioscorea alata; yam (generic)’
Adm	Drehet	u	‘greater yam; yam (generic)’
NNG	Malasanga	kui-kui	‘greater yam; yam (generic)’
PT	Iduna	kuvi	‘greater yam; yam (generic)’
MM	Vitu	ɣuvi	‘greater yam; yam (generic)’
MM	Tolai	a-up	‘greater yam; yam (generic)’
MM	Marovo	uvi	‘greater yam; yam (generic)’
MM	Maringe	n-ufi	‘greater yam, Dioscorea alata; yam (generic)’ (Henderson and Hancock 1988)
TM	Buma	uwâ	‘yam’
TM	Tanema	uva	‘yam’
TM	Tanibili	no-upio	‘yam’
TM	Vano	upie	‘yam’
SES	Lengo	uvi	‘greater yam, Dioscorea alata; yam (generic)’ (Henderson and Hancock 1988)
SES	Arosi	uhi	‘greater yam; yam (generic)’
NCV	Uripiv	n-ov	‘k.o. yam’
NCV	Paamese	o-uhu	‘greater yam; yam (generic)’
NCV	Lewo	yuwi	‘greater yam; yam (generic)’
NCV	Namakir	ʔuw	‘greater yam; yam (generic)’
NCV	Nguna	na-wii	‘greater yam; yam (generic)’
NCV	Lewo	yui	‘yam (generic)’
SV	Sye	n-up	‘greater yam; yam (generic)’
SV	North Tanna	n-up	‘greater yam; yam (generic)’
SV	Kwamera	n-uk	‘greater yam; yam (generic)’
SV	Southwest Tanna	n-ekw	‘greater yam; yam (generic)’
SV	Lenakel	n-uw	‘greater yam; yam (generic)’
SV	Whitesands	n-u	‘greater yam; yam (generic)’
SV	Anejom̃	n-u	‘greater yam; yam (generic)’
NCal	Pije	kuuk	‘greater yam; yam (generic)’
NCal	Fwâi	kuuk	‘greater yam; yam (generic)’
NCal	Nemi	kuuk	‘greater yam; yam (generic)’
NCal	Jawe	kuic	‘greater yam; yam (generic)’
NCal	Nyelâyu	uvi	‘greater yam; yam (generic)’
NCal	Nêlêmwa	kuvic	‘greater yam; yam (generic)’
NCal	Xârâcùù	ku	‘greater yam; yam (generic)’
NCal	Iaai	u	‘greater yam; yam (generic)’
Fij	Wayan	uvi	‘greater yam; yam (generic)’
Pn	Tongan	ʔufi	‘greater yam; yam (generic)’
Pn	Samoan	uji	‘greater yam; yam (generic)’

Three other terms can be reconstructed from cognate sets some of whose members are glossed ‘large yam’, ‘greater yam’ or ‘Dioscorea alata’. This is weak evidence for inferring that the reconstructed terms denoted a Dioscorea alata cultivar. The terms are POc *mʷaruqe, POc *udu(r,R) and POc *pʷasepe.

POc		*mʷaruqe	‘Dioscorea sp. or perhaps a cultivar of Dioscorea alata’
Adm	Titan	mʷare(n)	‘a big yam’
Adm	Loniu	mʷat	‘k.o. large yam’
MM	Roviana	marihi	‘yam’ (borrowed?) (Henderson and Hancock 1988)
NCV	Paamese	a-marue	‘k.o. wild yam’
NCV	Lewo	mol-malu	‘k.o. wild yam, ready before cultivated yams’
NCV	Nguna	mʷālu	‘k.o. yam like English potato’
NCV	Namakir	māroʔ	‘k.o. taro’
NCV	South Efate	n-mʷal	‘Dioscorea sp.’
PSV		*-mʷariq	‘Dioscorea sp.’ (Lynch 2001c: 234)
SV	Sye	(n-uv)mori	‘Dioscorea sp.’ (n-uv ‘yam (generic)’)
SV	Ura	(n-up)mori	‘Dioscorea sp.’ (n-up ‘yam (generic)’)

POc		*udu(r,R)	‘Dioscorea alata cultivar (?)’
NNG	Amara	(o)udo	‘greater yam’
MM	West Kara	udi	‘greater yam’
MM	Nalik	udur	‘greater yam’
SES	Arosi	ugu-ugu	‘yam sp. with nice smell’

POc *pʷasepe ‘Dioscorea alata variety(?)’ is of interest, because it also seems to be the source of the PT forms listed below the cognate set. However, the sound correspondences attested by the PT forms are not those of PT languages but the distinctive reflexes (*pʷ > kʷ; *s > t) of North Malaitan languages of the SES group. Dictionaries are available for two North Malaitan languages, Lau (Fox 1974) and Kwaio (Keesing 1975), but neither includes this item. I assume it is nonetheless a borrowing from a North Malaitan language, but I do not know the source of the borrowing.

POc		*pʷasepe	‘Dioscorea alata variety(?)’
MM	Petats	waseh	‘Dioscorea alata’
MM	Selau	wesewe	‘Dioscorea alata’

cf. also:

PT	Molima	atea	‘greater yam’
PT	Dobu	kʷatea	‘yam’
PT	Duau	kʷatea	‘yam’; ‘lesser yam’
PT	’Auhelawa	ateya	‘yam’
PT	Saliba	kʷatea	‘greater yam, Dioscorea alata; yam (generic)’
PT	Tubetube	kʷatea	‘greater yam’

2.1.3. Dioscorea bulbifera, potato yam, aerial yam, air potato (Dioscoreaceae)⇫

The terms ‘potato yam’ and ‘aerial yam’ reflect the fact that Dioscorea bulbifera has multiple aerial bulbils rather than a single basal tuber. Except in parts of Papua New Guinea it is not cultivated but quite commonly grows wild (Barrau 1962: 113). Its bulbils contain a toxin, dioscorine, and need to be cooked and grated, then washed continuously in fresh water for several hours (Barrau 1955: 59). In Papua New Guinea it is sometimes cultivated as a minor staple alongside the two major staple species, Dioscorea alata and Dioscorea esculenta (Bourke 1982).

POc *pʷatika evidently referred to the potato yam, Dioscorea bulbifera, because its reflexes denote this species in widely separated languages: Patpatar (MM), Kwara’ae (SES) and Sye and Ura (SV) (the Chuukese form appears to be borrowed). The languages in which it is glossed ‘lesser yam’ (Dioscorea esculenta) are contiguous and appear to represent a semantic innovation local to central New Ireland.

POc		*pʷatika	‘potato yam, aerial yam, Dioscorea bulbifera’
Adm	Lou	puet	‘Dioscorea bulbifera’
PT	Boanaki	posika	‘yam’
MM	Notsi	pias	‘lesser yam’
MM	Tabar	poti-poti	‘lesser yam’
MM	Lamasong	patik	‘lesser yam’
MM	Patpatar	patuk	‘Dioscorea bulbifera’
MM	Halia	pat	‘greater yam’
SES	Kwara’ae	(dau)fasia	‘Dioscorea bulbifera cultivars’ (Henderson and Hancock 1988)
PSV		*(nə-tai-)bʷatɣV	‘yam sp.’ (Lynch 2001c)
SV	Sye	(tai)potɣo(nei)	‘k.o. wild yam’
SV	Ura	(dai)borɣe(ni)	‘Dioscorea bulbifera’

cf. also:

Mic

Chuukese

pʷereka

‘Dioscorea bulbifera’ (-r- for †∅)

2.1.4. Dioscorea esculenta, lesser yam, prickly yam, TP mami, P pana (Dioscoreaceae)⇫

Dioscorea esculenta has a prickly vine stem and winds clockwise around its stake, whereas Dioscorea alata winds anti-clockwise (Figure 9.1).

The reasons why a POc etymon for this widespread species is apparently not reconstructable are discussed in §2.1.1. There are, however, a number of terms in NNG and MM languages which imply that there may have been a PWOc term for the species.⁶ Two forms are quite widely reflected, *ka-misa and *ma-misa. The *ka-/*ma- alternation is typical of stative (adjective-like) verbs in POc,⁷ implying that this was originally an epithet applied to Dioscorea esculenta rather than a noun. However, no other meaning for this word can be reconstructed at present.

The possibility of reconstructing PWOc *ka-misa/*ma-misa ‘lesser yam, Dioscorea esculenta’ perhaps means that the lesser yam was domesticated in NW Island Melanesia after the break-up of POc but before PWOc had dispersed far. Against this inference, however, is the fact that we would also expect to find reflexes of such a significant item in languages of the Papuan Tip linkage, the third member linkage of Western Oceanic, yet none have been found.⁸ The S Efate and Wayan Fijian forms are both very doubtful cognates.

PWOc		*kamisa	‘lesser yam, Dioscorea esculenta’
NNG	Tuam	amez	‘yam’
NNG	Tami	kamit	‘yam’
NNG	Yabem	ame	‘lesser yam, Dioscorea esculenta’
NNG	Yalu	amis	‘lesser yam, Dioscorea esculenta’
NNG	Sissano	emiei	‘greater yam’
MM	Tangga	kam	‘sweet potato; lesser yam’
MM	Tomoip	misa	‘lesser yam, Dioscorea esculenta’

cf. also:

NCV

South Efate

n-am

‘wild varieties of Dioscorea bulbifera’

PWOc		*mamisa	‘lesser yam, Dioscorea esculenta’
NNG	Mengen	mamisa	‘lesser yam, Dioscorea esculenta’
NNG	Mamusi	mamisa	‘yam’
MM	Minigir	mamisa	‘lesser yam’

cf. also:

Fij

Wayan

mami

‘cooking banana with short thick fruit; sweet-tasting, but skin remains green and flesh hard even when ripe’

2.1.5. Other terms for yam⇫

Further terms for yam can be reconstructed, but it is unclear which species they denoted.

PWOc		*gobu	‘Dioscorea sp.’
NNG	Numbami	go-gobu	‘greater yam’
NNG	Patep	ɣɛb	‘yam’
MM	Bali	gobu	‘yam, Dioscorea alata, cultivar’ (Hide 1985)
MM	East Kara	go-gof	‘Dioscorea bulbifera’

PROc		*damu	‘Dioscorea sp.’
NCV	Mota	nam	‘yam’
NCV	Araki	tsam	‘Dioscorea sp.’
NCV	Tamambo	dam	‘yam’
NCV	Ambae	damu	‘Dioscorea sp.’
NCV	Raga	damu	‘yam’
NCV	Uripiv	drum	‘Dioscorea sp.’
NCV	Naman	ne-dum	‘Dioscorea sp.’
NCV	Neve’ei	ne-dam	‘Dioscorea sp.’
NCV	Avava	a-dam	‘Dioscorea sp.’
NCV	Larëvat	dram	‘Dioscorea sp.’
NCV	Nese	na-ram	‘Dioscorea sp.’
NCV	Port Sandwich	na-ⁿram	‘yam’
NCV	Lonwolwol	dɛm	‘yam’
SV	Kwamera	nau-ram	‘kind of wild yam’
SV	Anejom̃	rame	‘k.o. yam, stringy’
NCal	Pije	dimʷa	‘Dioscorea bulbifera’
NCal	Jawe	dimʷa	‘Dioscorea bulbifera’
NCal	Fwâi	niwâ	‘Dioscorea bulbifera’
NCal	Nemi	niwâ	‘Dioscorea bulbifera’
NCal	Nyelâyu	demʷa	‘Dioscorea bulbifera’
NCal	Nêlêmwa	demʷa	‘Dioscorea bulbifera’

cf. also:

Fij

Wayan

damu(ni)

‘k.o. yam with curved tuber and chocolate coloured skin’ (final -ni is unexplained)

PWOc		*ka(p)ul	‘seed yam’
PT	Dobu	awona	‘seed yam’
PT	Kakabai	ko-koya	‘yam’
PT	Misima	ka-kaun	‘seed yam’
PT	Kilivila	kaula	‘yam’
MM	Tiang	ko	‘lesser yam’
MM	East Kara	ko-kau	‘lesser yam, Dioscorea esculenta’
MM	Patpatar	kau-kau	‘lesser yam, Dioscorea esculenta’9
MM	Nehan	ko-ko	‘yam’
MM	Marovo	co-core	‘wild yam sp.’

2.2. Taro (Araceae)⇫

Most major food plant terms both in English and in Oceanic languages refer to a single genus. The English term ‘taro’ is an exception. In much of the Pacific ‘taro’ is used to refer collectively to five genera of the Araceae family, namely Colocasia, Cyrtosperma, Alocasia, Amorphophallus, and Xanthosoma, each represented by a single cultivated species in the Pacific and shown in Figure 9.3. The most important species are - and in POc times probably were - Colocasia esculenta and Cyrtosperma merkusii.

2.2.1. Colocasia esculenta (syn. Colocasia antiquorum), taro, Asian taro, TP taro tru (Araceae)⇫

Like the other Araceae, in its natural state Colocasia taro prefers a shady, damp (and even swampy) environment, which explains its cultivation in rain forest environments and the need for irrigation elsewhere. It is often grown on crudely terraced slopes, the terraces supported by tree trunks or stick fences. It is usually propagated by planting a section from the top of the tuber together with the plant’s stem (the leaves are removed). Less often a cutting or a sucker from the side of the corm is used. Ivens (1927: 355) reports from Ulawa that, because taro is not a seasonal crop, the farmer may simply cut the top off a newly pulled tuber and re-plant it immediately. Tubers are ready for harvesting 7-12 months after planting. It grows throughout the year, obviating the need for storage (Barrau 1955: 50-52).

Two POc terms for Colocasia esculenta can be reconstructed: *talo(s) and *mʷapo(q). This synonymy perhaps reflects the complex history of taro cultivation by early Oceanic speakers. On the one hand they brought with them varieties of taro that they named *talo(s). Evidence for this is that the term was inherited from PMP. On the other hand they interacted with Papuan speakers in NW Melanesia and may have acquired new taro varieties from them, along with the label *mʷapo(q), which apparently entered POc from a mainland Papuan language. Reflexes of a possible source occur in languages of the northern Adelbert Range in the present-day Madang Province (see Z’Graggen 1980), and the widespread currency in Papuan languages of words with a form resembling mao is noted by Dutton (1973: 443) and Hays (2005: 642).

The cognate set from which POc *mʷapo(q) is reconstructed contains several formal difficulties. The alternation of reflexes of POc *m and *mʷ in certain items including this one has been discussed by Blust (1981a). French-Wright (1983: 130–132) also considers the alternation between zero and reflexes of POc *-p- in this set, choosing to reconstruct two POc forms, one with and one without *-p-, but suspecting that they are reflexes of a single form. With more data at our disposal, it is clear on both formal and semantic grounds that this is a single cognate set, but one in which two common sporadic changes often occur, namely that *mʷ becomes *m and that *p is lost between vowels when one is rounded. Some of the NNG reflexes (Kove, Gitua and Mangap) complicate the picture by reflecting POc *-q. These variations may mean that the earliest forms in Oceanic do not date back to POc itself but represent slightly later borrowings from dialect to dialect among early Oceanic dialects.

The distribution of the two cognate sets is also interesting. Whilst POc *mʷapo(q) is distributed throughout Oceania, POc *talo(s) hardly occurs in Western Oceanic. Its only known Western Oceanic reflexes are in Manam (NNG), Motu (PT) and Roviana and Marovo (MM). In the cases of Manam and Motu, no reflex has been found in nearby closely related languages, and it is quite possible that these are borrowings from an Eastern Oceanic language or from an English-based pidgin.¹⁰ In any case, POc *talo(s) has been widely replaced by POc *mʷapo(q) in Western Oceanic languages. John Lynch (pers. comm.) points out that reflexes of *talo(s) have been replaced by reflexes of PSOc *bʷeta in all of North/Central Vanuatu except South Efate (which arguably subgroups with South Vanuatu).

A search for languages in which reflexes of *talo(s) and *mʷapo(q) are in contrast has turned up very little, but this may be an artefact of inadequate data. Only Arosi reflects both terms, each as the first element of names of taro varieties (see below).¹¹ The Arosi generic term for taro is bʷa, reflecting neither POc term.

PMP		*tales	‘taro, Colocasia esculenta’ (Dempwolff 1938)
POc		*talo(s)	‘taro, Colocasia esculenta’ (Capell 1943: *dalo(s))
NNG	Manam	taro	‘taro, Colocasia esculenta’
PT	Motu	talo	‘taro, Colocasia esculenta’
MM	Roviana	talo	‘taro, Colocasia esculenta’
MM	Marovo	talo	‘taro, Colocasia esculenta’
SES	Kwaio	alo	‘taro (generic); unit of a hundred taro for a feast’
SES	Lau	alo	‘taro (generic)’
SES	Arosi	aro(-kamae)	‘taro sp.’
SES	Arosi	aro(-kauhi)	‘taro sp.’
SES	Arosi	aro(-maiki)	‘taro sp.’
SES	Bauro	aro	‘taro’
NCV	South Efate	(n)tal	‘taro, Colocasia esculenta’
SV	Sye	n-tal	‘taro, Colocasia esculenta’
SV	Lenakel	nə-te	‘taro, Colocasia esculenta’
SV	Southwest Tanna	nə-tel	‘taro, Colocasia esculenta’
SV	Anejom̃	n-tal	‘taro, Colocasia esculenta’
Fij	Bauan	dalo	‘taro, Colocasia esculenta’
Pn	Tongan	talo	‘taro, Colocasia esculenta’
Pn	Samoan	talo	‘taro, Colocasia esculenta’

POc		*mʷapo(q)	‘taro, Colocasia esculenta’ (Milke 1968: *mʷao)
Adm	Lou	mʷa	‘wild plant like taro’
Adm	Nali	mah	‘k.o. taro’
Adm	Drehet	muh	‘taro, Colocasia esculenta’
Adm	Loniu	mah	‘taro, Colocasia esculenta’
Adm	Titan	ma	‘taro, Colocasia esculenta’
NNG	Kove	moi	‘taro, Colocasia esculenta’
NNG	Tuam	mo	‘taro, Colocasia esculenta’
NNG	Gitua	mʷai	‘taro, Colocasia esculenta’
NNG	Mangap	mok	‘taro; good food’
NNG	Kilenge	mo	‘taro, Colocasia esculenta’
NNG	Mangseng	mʷa	‘taro, Colocasia esculenta’
NNG	Mengen	mao	‘taro, Colocasia esculenta’
NNG	Bing	muāw	‘k.o. yam’
NNG	Takia	mao	‘taro, Colocasia esculenta’
NNG	Yabem	mo	‘taro, Colocasia esculenta’
NNG	Sirasira	moa	‘taro, Colocasia esculenta’
NNG	Wogeo	mʷau	‘taro, Colocasia esculenta’
PT	Dawawa	mavu	‘taro, Colocasia esculenta’
PT	Sinaugoro	maɣo	‘yam (generic)’
PT	Motu	maho	‘greater yam’
MM	Nakanai	mavo	‘taro (generic); Colocasia esculenta’
MM	Teop	mū	‘taro, Colocasia esculenta’
MM	Kia	mahu	‘taro, Colocasia esculenta’
MM	Maringe	mʰau	‘taro, Colocasia esculenta’ (Henderson and Hancock 1988)
SES	Arosi	mʷa(-rataʔai)	‘taro sp.’
SES	Arosi	mʷa(-ruguʔino)	‘taro sp.’
NCal	Xârâcùù	mʷɛ	‘taro, Colocasia esculenta’
Pn	Hawaiian	nao	‘k.o. taro’

PSOc		*bʷeta	‘taro, Colocasia esculenta’ (Lynch 2004a)
NCV	Mwotlap	pʷet	‘taro, Colocasia esculenta’
NCV	Ambae	pʷeta	‘taro, Colocasia esculenta’
NCV	Araki	pera	‘taro, Colocasia esculenta’
NCV	Raga	bʷeta	‘taro, Colocasia esculenta’
NCV	Nokuku	pʷeta	‘taro, Colocasia esculenta’
NCV	Kiai	peta	‘taro, Colocasia esculenta’
NCV	Tamambo	bueta	‘taro, Colocasia esculenta’
NCV	Sakao	(oe)vad	‘taro, Colocasia esculenta’
NCV	Uripiv	na-bʷet	‘taro, Colocasia esculenta’
NCV	Big Nambas	na-pʷet	‘an edible root (not taro)’
NCV	Lonwolwol	(u)bʷer	‘taro, Colocasia esculenta’
SV	Anejom̃	na-pʷat	‘k.o. taro’
NCal	Nemi	pʷeek	‘wild arrowroot, Amorphophallus paeoniifolius (?)’
NCal	Jawe	pʷeek	‘wild arrowroot, Amorphophallus paeoniifolius (?)’

The leaves of Colocasia esculenta are eaten as a green vegetable in many parts of Oceania, and many languages have a separate word for them (ch.4, §2.5). A possible POc term for taro leaves, POc *gal(a,o) is weakly supported, along with a possible PWOc candidate, *[qa]pʷasu - ‘possible’ because only its NNG and PT reflexes denote leaves; its MM reflexes denote Colocasia esculenta itself.

POc		*gal(a,o)	‘taro leaves’ (?) (Ross 1996d: 190)
Adm	Baluan	gal	‘taro’
NNG	Labu	ka	‘taro’
MM	Vitu	galo	‘taro leaves’
SES	Kwaio	gala-	‘taro shoot’

PWOc		*[qa]pʷasu	‘taro leaves’ (?)
NNG	Kaulong	pasu	‘(mature) taro leaf’
NNG	Sengseng	paso	‘taro leaf’ (A. Chowning, pers. comm.)
NNG	Manam	(minam) ʔapʷas	‘k.o. yam’
PT	Misima	pʷasuwa	‘k.o. tree with edible leaves; Gnetum gnemon’
MM	East Kara	iavas	‘taro, Colocasia esculenta’
MM	Madak	pas	‘taro’
MM	Patpatar	pas	‘taro, Colocasia esculenta’
MM	Tolai	pa	‘taro, Colocasia esculenta’

Other terms for parts of the taro plant are reconstructed in chapter 4. They are

• POc *[s,j]uli(q) ‘banana or taro sucker, slip, cutting, shoot (i.e. propagation material)’ (ch.4, §2.6)¹²
• POc *baRa-baRa ‘stem or stalk of non-woody plants, such as taro and banana, probably also the soft stems of leaves’ (ch.4, §2.2)
• POc *up(e,a) ‘taro seedling’ (ch.4, §2.6)

2.2.2. Cyrtosperma merkusii (syn. Cyrtosperma chamissonis, Cyrtosperma edule), swamp taro, giant taro, P kakake, kakama, B wota taro (Araceae)⇫

At 4 metres tall swamp taro, Cyrtosperma merkusii, is a giant among the Araceae. It grows wild or is cultivated in swampy areas on many of the islands of Melanesia, Polynesia and Micronesia. Cyrtosperma taro is commonly grown in the mud directly behind the mangrove swamp (Barrau 1955: 25-26). In pre-European times it was cultivated in the Bismarcks (on Manus and New Ireland), the Solomons, Vanuatu, Micronesia, western Polynesia and parts of Fiji but not on mainland New Guinea, in New Caledonia, or in eastern Polynesia (Barrau 1962: 103, Andrew Pawley, pers. comm.). It tolerates saline swamps and requires little labour to cultivate (Bourke 1982). Because of its suitability for marginal conditions, it is an important staple on atolls in Micronesia. Elsewhere it has only marginal importance as a food crop. It is propagated by planting cuttings in marshy ground, and the tuber, which may weigh up to 15 kg, is harvested only after three years or longer. It is usually cooked by baking (Barrau 1955: 53). Its huge waxy leaves are used in the Solomons for sealing stone ovens and for laying out food (Henderson & Hancock 1988: 29-30).

The reconstruction of a term for ‘swamp taro’ has a chequered history. In Ross (1996d) I proposed that the POc name for swamp taro was *bulaka, but indicated that there was a formal problem with this reconstruction. Its NCV reflexes lack a consonant corresponding to *-l-, and Clark (1996) accordingly reconstructed PNCV *buaga ‘taro’. Geraghty (1990: 57-58) observed that POc *R, but not *l, is often lost in NCV languages, and took this as evidence for PEOc *buRaka, rather than *bulaka, citing Nakanai bureka in support of this. This is just one of three Nakanai reflexes, however,¹³ and at least two out of the three must be borrowings.

Geraghty suggested that Polynesian reflexes were probably borrowings from Micronesian, and Barrau (1959) and Whistler (1991a: 47, 58-59) suggested that Cyrtosperma merkusii was indeed introduced into Polynesia from Micronesia. Kikusawa (2003) has expanded on this and argues that the Micronesian terms below, other than Marshallese, reflect Proto Chuukic-Ponapeic *pʷulaka ‘swamp taro, Cyrtosperma merkusii’ (Chuukic-Ponapeic is a subgroup of Micronesian), and that all the other terms listed below were borrowed directly or indirectly from a Chuukic-Ponapeic language.

Kikusawa may well be right,¹⁴ as Bourke (1990) suggests that Cyrtosperma taro was introduced into the Bismarcks by Polynesians living on the atolls to the east. Swamp taro is the dominant root crop only in Micronesia, and the irregularities of reflexes outside Micronesia and Polynesia cause difficulties both for my *bulaka and Geraghty’s *buRaka.¹⁵ In order to demonstrate this, I list against each reflex below the POc form that it would reflect if it were directly inherited. The Micronesian terms, other than Marshallese, reflect PChk-Ponapeic *pʷulaka, which reflects a hypothetical POc *bulaka. The Polynesian terms appear to reflect PPn *pulaka, also reflecting a hypothetical POc *bulaka. However, the claim in Ross (1996d) that they actually reflect POc *bulaka collapses as we move westward, as does Geraghty’s reconstruction of *buRaka. NCV terms consistently reflect PNCV *bu(R)aga, whilst MM and Adm reflexes are chaotic and typical of sets most or all of whose members are borrowed.

The items listed below denote ‘swamp taro, Cyrtosperma merkusii’ unless glossed otherwise.

Adm	Mussau	ulaa	‘swamp taro, Cyrtosperma merkusii’ (would reflect *pu(l,r,R)a(q,k)a)
Adm	Seimat	hula	‘swamp taro, Alocasia sp., introduced from Aua’ (would reflect *pula(q,k)a)(Sorensen 1950)
Adm	Aua	fuula	‘taro’ (would reflect *pula(q,k)a)
Adm	Lou	pulak	‘swamp taro, Cyrtosperma merkusii’ (would reflect *(p,b)ulaka)
MM	Nakanai	ula	‘an inedible wild taro, Colocasia sp.’ (would reflect *u(l,r,R)a(q,k)a)
MM	Nakanai	buleha	‘an inedible wild taro, Colocasia sp.’ (would reflect *bu(l,r,R)eqa)
MM	Nakanai	bureka	‘variety of elephant ear taro, Alocasia macrorrhizos’ (would reflect *buseqa)
MM	East Kara	vuləi	‘k.o. taro’ (would reflect *hulaqa)
MM	Patpatar	pulaka	‘Polynesian arrowroot, probably Tacca leontopetaloides’ (would reflect *pulaka)16
MM	Tolai	pulaka	‘Polynesian arrowroot, probably Tacca leontopetaloides’ (would reflect *pulaka)
NCV	Mota	puaka	‘boggy ground, mud’ (would reflect *hu(R)aga)
NCV	Uripiv	ʙuak	‘(rare) taro’ (would reflect *hu(R)aga)
NCV	Big Nambas	bōak	‘taro, Colocasia sp.’ (would reflect *hu(R)aga)
NCV	Port Sandwich	ᵐbuaᵑg	‘hill taro’ (would reflect *hu(R)aga)
NCV	Naman	buag	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hu(R)aga)
NCV	Neve’ei	ni-biaŋ	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hu(R)aga)
NCV	Avava	ʙuaŋ	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hu(R)aga)
NCV	Nese	boak	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hu(R)aga)
NCV	Namakir	buag	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hu(R)aga)
NCal	Xârâcùù	buraa	‘variety of taro’ (would reflect ?)
Mic	Marshallese	pɯəl	‘swamp taro patch’ (would reflect *hulaqa)
Mic	Mortlockese	pʷulᴂ	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hula(q,k)a)
Mic	Chuukese	pʷuɾa	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hula(q,k)a)
Mic	Puluwatese	pʷula	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hula(q,k)a)
Mic	Satawalese	pʷula	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hula(q,k)a)
Mic	Carolinian	bʷula	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hula(q,k)a)
Mic	Woleaian	fʷuraxe	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hulaka)
Mic	Pulo Annian	vʷuraxa	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hulaka)
Pn	Anutan	pulaka	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hulaka)
Pn	Pukapukan	pulaka	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hulaka)
Pn	East Futunan	pulaka	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hulaka)
Pn	East Uvean	pulaka	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hulaka)
Pn	Samoan	pulaʔa	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hulaka)
Pn	Tuvalu	pulaka	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hulaka)
Pn	Tokelauan	pulaka	‘swamp taro, Cyrtosperma merkusii’ (would reflect *hulaka)
Pn	Rarotongan	puraka	‘coarse kind of taro’ (would reflect *hulaka)

2.2.3. Alocasia macrorrhizos (syn._Alocasia macrorrhiza_, Alocasia indica), giant taro, elephant ear taro, P waeltaro (Araceae)⇫

The giant taro, Alocasia macrorrhizos, growing to 3-4 m, is not strictly a root crop; it is the corm (i.e. the thick lower stem) that is eaten. It is present as a minor staple in much of New Guinea and the Bismarcks, but appears to have been grown as much for ceremonial significance as for eating (Bourke 1982). Ann Chowning (pers. comm.) reports that among the Nakanai and the Sengseng of New Britain, giant taro is often eaten by people who are abstaining from Colocasia taro as an act of mourning for a dead relative. Alocasia macrorrhizos is not mentioned at all in Henderson and Hancock’s (1988) Solomons survey. Barrau (1955: 54) says that it is found throughout Melanesia but only sometimes cultivated in gardens, Hay (1990: 41-42) that it is uncommon in rural Papua New Guinea. Its calcium oxalate content is such that it requires lengthy baking or scalding with water several times.

The POc name for giant taro was *piRaq, continuing a PMP etymon. The final *-q is attested in Gumawana, Bwaidoga, Big Nambas and Tape and indirectly in the retention of the final vowel in SV languages (Lynch 2001c: 121).

PMP		*biRaq	‘taro sp.’ (Blust 1972b)
POc		*piRaq	‘giant taro, elephant ear taro, Alocasia macrorrhizos’ (Blust 1972b)
Adm	Mussau	ia	‘Alocasia macrorrhizos’
PT	Gumawana	vilava	‘taro’
PT	Bwaidoga	vilaɣa	‘edible root resembling taro’
PT	Motu	hira	‘large sp. of edible arum’
MM	East Kara	fia	‘Alocasia macrorrhizos’
SES	Gela	vita	‘Alocasia macrorrhizos’
SES	Arosi	hira	‘Alocasia macrorrhizos’
NCV	Mwotlap	vı	‘Alocasia macrorrhizos’
NCV	Mota	via	‘the giant caladium’
NCV	Ambae	via	‘Alocasia macrorrhizos’
NCV	Kiai	via	‘Alocasia macrorrhizos’
NCV	Araki	via	‘Alocasia macrorrhizos’
NCV	Raga	via	‘Alocasia macrorrhizos’
NCV	Big Nambas	ðᫀix	‘Alocasia macrorrhizos’ (J. Lynch, pers. comm.)
NCV	Uripiv	na-vi	‘wild arrowroot, Amorphophallus paeoniifolius’
NCV	Tape	viaɣ	‘Cyrtosperma merkusii’
NCV	Nese	na-dᫀi	‘k.o. taro’
NCV	Big Nambas	viah	‘Alocasia macrorrhizos’
NCV	Southeast Ambrym	o-hia	‘unidentified edible tuber’
NCV	Paamese	u-hiaa	‘wild yam’
NCV	Lewo	ko-yuia	‘(wild) yam’
NCV	Namakir	vi	‘Cyrtosperma?’
NCV	Nguna	na-via	‘taro sp., inedible; very bitter or acidic; said to burn the mouth if eaten’
NCV	South Efate	na-fi	‘taro sp.’
SV	Sye	ne-vye	‘water taro’
SV	Lenakel	nu-via	‘k.o. taro’
SV	Kwamera	nu-via	‘Crytosperma (wild taro)’
SV	Anejom̃	nehei	‘Alocasia macrorrhizos’
NCal	Pije	pia	‘Alocasia macrorrhizos’
NCal	Fwâi	pia	‘Alocasia macrorrhizos’
NCal	Nemi	pia	‘Alocasia macrorrhizos’
NCal	Jawe	pia	‘Alocasia macrorrhizos’
NCal	Nêlêmwa	pia	‘Alocasia macrorrhizos’
Fij	Wayan	via	‘giant taro, Alocasia and Cyrtosperma spp.; cultivated but eaten only in time of famine’

2.2.4. Another term for taro?⇫

The glosses suggest that the denotatum of POc *(p,b)oso below belonged to the Araceae family, but it is impossible to infer its identity from the available data.

POc		*(p,b)oso	‘k.o. taro’
NNG	Barim	bus	‘taro’
NNG	Lukep	bus	‘taro’
MM	Madak	pos	‘grealter yam’
Mic	Kosraean	ɔt	‘taro sp.’
Mic	Marshallese	wetɯ	‘elephant ear taro, Alocasia macrorrhizos’
Mic	Chuukese	wōt	‘taro, Colocasia esculenta’

2.3. Other tubers⇫

Three other tubers are quite widely distributed in the Pacific. Two are the wild arrowroot, Amorphophallus paeoniifolius (syn. Amorphophallus campanulatus), and the Polynesian arrowroot, Tacca leontopetaloides, but I am unable to reconstruct a POc term for either. This suggests either that they, too, are more recent introductions or, more probably, that they are rarely eaten and have undergone frequent renaming. The third tuber is Pueraria lobata, apparently more widely cultivated in earlier times than it is today.

2.3.1. Amorphophallus paeoniifolius (syn. Amorphophallus campanulatus), wild arrowroot, elephant yam (Araceae)⇫

Amorphophallus paeoniifolius is a wild plant. In the Austronesian speaking areas of Papua New Guinea it is a minor staple only in parts of western New Britain and in coastal SE Papua New Guinea (Bourke et al. 1998). Margetts (2005b) notes of the latter region that Amorphophallus paeoniifolius is traditionally grown but not very common as people do not go out of their way to cultivate it. It is not as valued as other taro types and does not play a role in feasts and wealth exchanges. In the Solomon Islands it is reported to have been a garden crop in parts of Choiseul and Malaita (Henderson & Hancock 1988: 32). Barrau (1955: 55) reports that it is no longer cultivated anywhere in Melanesia.

2.3.2. Tacca leontopetaloides, Polynesian arrowroot (Taccaceae)⇫

The tuber of Tacca leontopetaloides is bitter and requires considerable processing to produce a starch pudding somewhat like sago pudding. It was used in this way on the small islands of the SE Solomons and the Temotu Province, and in much of Polynesia (Henderson & Hancock 1988: 34) but apparently had only limited use in the Bismarcks.

Interestingly, almost all the names I have collected for it reflect a reassignment of the name of some other starchy food:

MM:	Patpatar	pulaka	< *bulaka ‘Cyrtosperma merkusii’ (§2.2.2)
MM:	Tolai	pulaka	< *bulaka ‘Cyrtosperma merkusii’ (§2.2.2)
TM:	Aiwoo	(to)piya	< POc *piRaq ‘Alocasia macrorrhizos’ (§2.2.3)
Fij:	Bauan	yabia	< POc *Rabia ‘Metroxylon sagu’ (§5.1)
Pn:	Pileni	pia	< POc *Rabia ‘Metroxylon sagu’ (§5.1)
Pn:	Rarotongan	pia	< POc *Rabia ‘Metroxylon sagu’ (§5.1)

The elements in the PPn binomial below are PPn *ma ‘fermented breadfruit’ (POLLEX) and *soaʔa ‘banana’ (§3).

PPn		*mā-soaʔa	‘Polynesian arrowroot, Tacca sp.’ (POLLEX)
Pn	Tongan	māhoaʔa	‘arrowroot, Tacca sp.’
Pn	Tokelauan	mahoā	‘Tacca leontopetaloides’
Pn	Anutan	maoa	‘arrowroot, Tacca sp.’
Pn	East Futunan	māsoʔā	‘arrowroot, Tacca sp.’
Pn	Samoan	māsoā	‘Tacca leontopetaloides’
Pn	Tuvalu	māsoa	‘arrowroot, Tacca sp.’

For the Micronesian names of Tacca leontopetaloides, however, no origin has been found: Nauru (da)mag-mag, Proto Western Micronesian *mʷaku-mʷaku ’arrowroot, Tacca leontopetaloides (Bender et al. 2003).

2.3.3. Pueraria lobata (syn. P. novo-guineensis, Pueraria thunbergiana, Pueraria triloba, Pueraria hirsuta, P. neo-caledonica, Pachyrrhizus trilobus, Pueraria montanus, P. Dolichos hirsutus, Dolichos tuberosus, Dolichos lobatus) (Fabaceae)⇫

Pueraria species are half-woody lianas, 3-5m high, with stems covered in hairs, brown on older parts, rust-yellow to white on younger. The bast is used to make yam for rope, fishing lines and fish nets. One species, Pueraria lobata (with an enormous number of synonyms), is found as far east as Samoa.

Pueraria lobata has a fleshy, tapioca-like tuber that is widely eaten in times of famine (Barrau 1965; Peekel 1984: 253-255). Barrau (1965) supports Guppy’s (1906: 412) suggestion that Pueraria lobata may also have been more widely cultivated by Oceanic speakers before the arrival of the sweet potato. It is still cultivated in the Kangean Islands (120 km east of Madura, Indonesia) and in New Caledonia, where Haudricourt (1964: 97), cited by Barrau, labels it ‘a food for chiefs’ rather than a famine food. One of Guppy’s mid-nineteenth-century sources noted that it also had ceremonial significance in Fiji.

No POc term for Pueraria lobata is reconstructable, and the cognate set below is restricted to North-Central Vanuatu and Central Pacific, and thus allow only a PROc reconstruction. None of Barrau’s sources make specific mention of the Bismarcks or the Solomons, and it may be that its significance has faded more definitely from folk memory there than on smaller islands where it still serves as a famine food.

The gloss of the Raga reflex refers to famine food, and probably denotes Pueraria lobata.

PROc		*Raka	‘k.vine, Pueraria lobata’ (Geraghty 1990)
NCV	Raga	aga	‘yam with blue flowers, eaten in time of famine’
NCV	Port Sandwich	ni-ax	‘vine sp.’
NCV	Paamese	e-ā	‘kind of tree with very tough roots that are very tough to dig out while hoeing in garden’
NCV	Namakir	ni-ak	‘vine sp. with blue flowers like a yam; Pueraria (?)’
Fij	Wayan	aka	‘creeper, Pueraria lobata’
Fij	Bauan	yaka	‘creeper, Pueraria lobata’ (Geraghty 2004: 79)
Pn	Tongan	aka	‘creeper, Pueraria lobata, whose root is eaten when better food is scarce’
Pn	Niuean	aka	‘Pueraria lobata’
Pn	Ifira-Mele	aka	‘creeper sp., probably Pueraria lobata’
Pn	Pukapukan	aka	‘a root’
Pn	Rennellese	aka	‘bush vine, Pueraria lobata with long roots, eaten in times of shortage’
Pn	Tikopia	aka	‘root of Pueraria lobata’
Pn	Samoan	aʔa	‘creeper, Pueraria lobata’
Pn	Tokelauan	aka	‘root’
Pn	Māori	aka	‘Metrosideros spp. that begin life as climbing vines.’

3. Bananas, cultivars of the genus Musa⇫

Banana plants have a non-woody stem and huge leaves which, along with the stem holding the bunches of bananas, grow directly out of the top of the main stem. The plant as a whole is usually around 3 m in height.

In Papua New Guinea today bananas are the second most widely consumed food crop after the sweet potato. For some PNG communities starchy varieties of bananas are the sole staple, picked and cooked before they are fully ripe. Some varieties lend themselves better to boiling, others to baking. Traditional banana varieties are relatively hard to grow; each year they must be moved to new garden sites with fertile soil. Many of these varieties produce fruit within six months of planting (French & Bridle 1978: 12). Figure 9.5 shows a traditional banana plant.

Banana leaves have a variety of uses. The most common are wrapping food for baking in a stone oven (vol.1, ch.6, §3.1) and laying out food on them at feasts.

Edible bananas are all sterile hybrids with complex genetic histories, cultivars of the genus Musa.¹⁷ In my 1996 attempt to gloss the various reconstructed terms for bananas, I employed the old division, based on morphological characteristics, of the genus Musa into five sections, varieties of which found in NW Melanesia belong to the sections Eumusa and Australimusa. In the past fifty years a new understanding of Musa sections has emerged, resulting in a reduction in their number from five to three, but two are still relevant to NW Melanesia, namely Musa ¹⁸ (including the former Eumusa) and Callimusa (including the former Australimusa) (Simmonds & Shepherd 1955, Wong et al. 2002). In Simmonds and Shepherd’s cultivated banana nomenclature, each section contains species (plus subspecies) and hybrids, the latter labelled with letters denoting the subspecies from which they are derived:

A for the various subspecies of Musa acuminata, B for those of Musa balbisiana, S for the minor New Guinea species Musa schizocarpa (all three in the section Musa) and T for Callimusa-derived genomes. Thus ‘Musa sapientum’ and ‘Musa paradisiaca’, the two alleged species recognised by Linnaeus, are both AAB triploid hybrids. Arnaud and Horry’s (1997) survey of 108 banana varieties in Papua New Guinea includes AA and AB diploid and AAA, AAB and ABB triploid hybrids, as well as hybrids of Musa and Callimusa species.¹⁹

Biomolecular study confirms these sections and is currently bringing about a revision of ideas about the archaeobotany of bananas, which is considerably more complicated than previously thought (J. Kennedy forthcoming). The transformation from wild to cultivated edible bananas occurred independently in the two sections. The wild species of the section Musa that have contributed to edible bananas are found over a region stretching from eastern India and Sri Lanka through island SE Asia into the NW Melanesia as far as the Solomons, but the fact that New Guinea is home to many highly diverse diploid AA cultivars makes it the probable area of their domestication. Wild Callimusa species are confined to a region stretching from Halmahera to the Solomons, indicating that NW Melanesia was the region where Callimusa bananas were also first domesticated.

J. Kennedy (2008) points out that all bananas east of the Solomons were carried there by settlers. These belong to three distinct lineages. The first consists of so-called Fe’i bananas, hybrids among several Callimusa species (and the world’s only edible Callimusa bananas). The other two, both known as ‘Pacific plantains’, are derived from Musa species. One involves Musa acuminata, subspecies banksii, endemic to New Guinea, and the other is a cross between the Musa acuminata subspecies banksii and errans, the latter endemic to the Philippines. Fe’i bananas and banksii derivatives both originated in New Guinea, but the site of banksii/errans hybridisation is unknown. Archaeological evidence from Kuk in the New Guinea Highlands indicates that edible diploids derived from Musa acuminata banksii were under cultivation by 7000 years ago, but the origins of Fe’i bananas and banksii/errans hybrids cannot be dated. The geographic and chronological unknowns do not exclude the possibility that all three lineages were carried into the Pacific by early Oceanic speakers, but they do not confirm it either.

The likelihood that banana domestication first occurred in New Guinea, and that edible bananas spread from there into Asia and Africa, means that any bananas brought to New Guinea and the Bismarcks by Austronesian speakers probably came from this region in the first place. That bananas have long been a part of their diet is confirmed by the very well attested PMP term *punti and its POc continuation *pudi. The fact that its reflexes are so widespread makes it certain that this was the generic term for bananas. Today’s Oceanic speakers usually have numerous terms for the different banana cultivars in their gardens, but the fact that few people use Simmonds and Shepherd’s nomenclature in glosses (and that old terms persist) makes it difficult or impossible to provide scientific names for these. Because of this difficulty, old species terms are retained in glosses.

PMP		*punti	‘banana’ (Dempwolff 1938)
POc		*pudi	‘banana, Musa cultivars’ (Capell 1943: *puti)
Adm	Mussau	uri	‘banana’
Adm	Seimat	pudi	‘Musa sp.’ (Sorensen 1950)
Adm	Drehet	puŋ	‘banana’
NNG	Mengen	puri	‘banana’
NNG	Kove	puri	‘banana’
NNG	Lukep	pur	‘banana’
NNG	Gitua	pudi	‘banana’
NNG	Mapos Buang	vud	‘banana’
PT	Tubetube	udi	‘banana’
MM	Nakanai	vugi	‘banana’
MM	Bulu	vudi	‘banana’
MM	Tigak	ur	‘banana’ (Beney 1980)
MM	Tolai	vudu	‘banana’
TM	Äiwoo	no-u	‘banana’
SES	Gela	vudi	‘banana’
SES	To’aba’ita	fudi	‘banana’ (archaic)
SES	Lau	fudi	‘banana’
SES	Sa’a	huti	‘banana’
NCV	Naman	nə-vəj	‘banana’
NCV	Nese	no-ðᫀic	‘banana’
NCV	Uripiv	na-vij	‘banana’
NCV	Lonwolwol	vih	‘banana’
NCV	Port Sandwich	na-vüc	‘banana’
NCV	Nguna	na-adi	‘banana’
SV	Sye	no-voh	‘banana’
SV	Anejom̃	no-hos	‘banana’
NCal	Pije	piji(ŋ)	‘Musa paradisiaca’
NCal	Xârâcùù	pwî	‘banana’
NCal	Iaai	o-vic	‘banana’
Mic	Kosraean	uṣ	‘banana’
Mic	Mokilese	wus	‘banana’
Mic	Ponapean	ūt	‘banana’
Mic	Chuukese	wūc̣	‘banana’
Mic	Puluwatese	wūṛ	‘banana’
Mic	Satawalese	wūṛ	‘banana’
Mic	Woleaian	wīṣi	‘banana’
Fij	Wayan	vudi	‘cooking banana; sometimes used as generic term for banana’
Pn	Tongan	fusi	‘cooking banana or banana of any kind’

The term Musa troglodytarum is used (inaccurately) both of wild varieties of Musa balbisiana and of Fe’i bananas. In the Uripiv and Samoan glosses below it is likely that it denotes the latter, in agreement with the Bauan gloss. POc *joRaga is glossed accordingly, but with a question mark, because this was the meaning of the PROc reflex but not necessarily of the POc term. Fe’i bananas are uncommon in New Guinea today, and it is not clear whether they were present in the Bismarcks in Proto Oceanic times.

POc		*joRaga	‘banana, Fe”i (?) cultivars’
NNG	Middle Watut	cok	‘banana’
NNG	Patep	joŋ	‘banana (generic); banana, Musa sapientum’
MM	Vaghua	soga	‘banana’
SES	Arosi	toraga	‘banana sp.’
NCV	Raga	hoaga	‘k.o. banana’
NCV	Uripiv	jok	‘Musa troglodytarum, a kind of banana’
NCV	Paamese	sōko	‘kind of banana with large orange coloured fruit’
NCV	Nguna	soaga	‘cooking banana’
Fij	Bauan	soaga	‘banana sp., Musa fehi’
Pn	Samoan	soaʔa	‘mountain plantain, Musa troglodytarum’ (Whistler 2000: 7)

The glosses of the four terms below are vague, both because the glosses of present-day forms are often vague and for the reasons given above, but they appear to have denoted banana cultivars. POc *sakup perhaps denoted a cultivar with long fruit, as the glosses of its Gumawana, Motu and Kwara’ae reflexes suggest. In view of its upright bunches Kwara’ae sa-sao appears to denote a Fe’i cultivar (J. Kennedy, pers. comm.), but this is neither confirmed nor disconfirmed by the glosses of its cognates. It is probable that POc *bʷera was a descriptive term meaning ‘white’, since Puluwat pʷer retains this meaning. The POc form is probably a reflex of PMP *burak ‘white’. If so, one may infer that it referred to bananas of the Musa group, since Fe’i bananas are typically copper-coloured.

POc		*sakup	‘banana cultivar with long fruit’ (?)
PT	Gumawana	yagowa	‘a long non-sweet banana’
PT	Taupota	hakova	‘banana’
PT	Taboro	daua	‘k.o. banana: white flesh’
PT	Motu	dau	‘k.o. banana: very long’
MM	Roviana	hakua	‘banana’
MM	Maringe	cau	‘banana’
SES	Kwara’ae	sa-sao	‘k.o. banana with upright bunches and large fruit’
SES	’Are’are	sao-sao	‘k.o. wild banana’
NCV	Ambae	haka	‘banana’
NCV	Larëvat	(nəv)saɣ	‘banana’
NCV	Tape	(ni)saɣ	‘banana’
NCV	Paamese	sou-sou	‘k.o. banana’

POc		*bʷera	‘Musa cultivar’
PT	Gapapaiwa	bora-bora	‘k.o. banana’
PT	Budibud	bʷela-mʷela	‘banana’
NCV	Paamese	a-voi	‘k.o. banana’
Mic	Puluwatese	(wuṛu)pʷeṛ	‘k.o. banana enjoyed cooked’

POc		*baqun	‘banana cultivar’
Adm	Loniu	pakɔw	‘k.o. wild banana’
MM	Teop	pauna	‘banana; kidney’
SES	To’aba’ita	baʔu	‘banana plant and fruit’ (syn. fudi, now archaic)
SES	Dori’o	baʔu	‘banana’
PSV		*n-ban	‘banana (generic)’ (Lynch 2004a)
SV	Sye	ni-mpa	‘k.o. banana with long fruit’
SV	Lenakel	nə-pən	‘banana’
SV	Southwest Tanna	nə-pʷan	‘banana’
Mic	Puluwatese	(wuṛu)pāwo	‘k.o. cooking banana’

cf. also:

NNG

Mato

baɣup

‘banana’(Stober 2005)

PWOc		*bʷatiq	‘banana cultivar’
PT	Tawala	bihiya	‘banana plant’
PT	Misima	bʷahiki	‘banana’
PT	Nimoa	bʷasihe	‘banana’
MM	Vitu	beti	‘banana’
MM	Vangunu	batia	‘banana’

There is evidence for PWOc *pudi-pudi ‘wild banana’, a transparent reduplication of the generic POc term for banana (see ch. 2, §7.2).

NNG	Mangap	pin-pin	‘wild banana’
PT	Sudest	ɣudu-ɣudu	‘wild banana seeds’
MM	Ramoaaina	udu-udu	‘wild banana’
MM	Tolai	vudu-vudu	‘wild banana, Ensete glaucum’(Arnaud and Horry 1997: 19) 20

Names for the various parts of the banana plant are discussed in chapter 4. Some are also mentioned in §2.2.1, as they are also used for the taro (and probably for other plants, too). Sorting out glosses is sometimes difficult, but the following inferences appear to be well founded (section numbers refer to ch. 4): the banana plant has a main stem, *baRa-baRa (§2.2) of its own, from which protrudes a stem holding a bunch, *puŋu (§2.8) of bananas, from which in turn grow small stems, *kulo (§2.8) each holding a hand, *qitiŋ (§2.8) of bananas.²¹ A new banana plant is planted with a slip or cutting, *[s,j]uli(q) (§2.6), and the new growth first appears as a shoot, *qili (§2.6).

4. Artocarpus altilis (syn. Artocarpus communis, Artocarpus incisus), breadfruit, TP kapiak, P beletiutu (Moraceae)⇫

Breadfruit trees grow throughout the Oceanic-speaking region, except for New Zealand and the Chatham Islands. Breadfruit tend to be co-staples with other starchy foods, providing starch when the other staples are out of season and vice versa. In Melanesia it is most significant as a staple in the Temotu Province of the Solomons, but is also important in SE Papua, the Bismarcks and Bougainville,²² on the coasts of the islands of Vanuatu, in Micronesia, in Fiji and in parts of Polynesia (Henderson & Hancock 1988:36, Walter & Sam 2002: 107-108).

Breadfruit trees grow to a height of 20 m and usually at altitudes below 1200 m. They are fairly tolerant with regard to soil, and grow in the coral soils of atolls. In the South Pacific, a tree yields 50 to 150 ovoid grapefruit-sized pale yellow to yellow-orange fruit per (annual) season. In some areas there is a second smaller fruiting halfway between main annual fruitings. The breadfruit is a syncarp, a compound fruit with many segments arranged around the core, which itself is the spike of the original flower. Each segment contains a seed surrounded by flesh and growing from the core on a fleshy receptacle. The segments are visible on the skin of the fruit as hexagon-like shapes, as shown in Figure 9.6. A breadfruit may weigh as much as 5 kg.

Breadfruit seeds have a high protein content and are nutritionally the most valuable part of the fruit (Paijmans 1976: 123). Breadfruit are very rich in starch and contain about 25% carbohydrates. They are usually roasted, baked, fried, or boiled before being eaten. The cooking method depends on the cultivar. They can be eaten raw, but in Papua New Guinea and the Solomons they are commonly baked whole in a fire or an oven, and then the skin and seeds are removed. Some fruit have seeds about the size and texture of chestnuts and taste rather like very dense young potatoes. They are removed from the baked fruit and sometimes roasted over the fire before eating. In New Guinea this is usually the only part that is eaten (Walter & Sam 2002: 107-110, Bourke & Allen forthcoming).

The description above applies to breadfruit varieties in New Guinea and the Bismarck Archipelago. As one moves eastward across the Pacific, one finds breadfruit varieties that have been increasingly selected for flesh rather than seeds, so that most cultivars in Polynesia are seedless and are propagated by root cuttings. The genetic history is more complex than this implies, however, as two other species have contributed to the varieties subsumed under Artocarpus altilis. One is Artocarpus camansi, sometimes called ‘breadnut’, which occurred both wild and cultivated in New Guinea, the other Artocarpus mariannensis, endemic to the high islands of western Micronesia (Yen 1991, Zerega 2003, cited by Kennedy & Clarke 2004).

Because breadfruit trees usually produce large crops at certain times of the year, preservation is an issue. In parts of Papua New Guinea and in the Temotu archipelago, breadfruit are dried into a sort of biscuit (French & Bridle 1978:40, Henderson & Hancock 1988: 37). A traditional preservation technique in Vanuatu, Micronesia and Polynesia is to bury peeled and washed fruits in a leaf-lined pit where they ferment over several weeks and produce a sour, sticky paste. So stored, the product may last a year or more, and some pits are reported to have produced edible contents more than 30 years later (for terms associated with fermentation, see vol.1, ch.6, §4.2).

A common breadfruit product is a mixture of cooked or fermented breadfruit mash mixed with coconut milk and baked in banana leaves. In Vanuatu breadfruit are sometimes grated before cooking, or made into small balls and cooked in coconut milk.

All parts of the tree yield latex, a milky juice, which is used in some locations for boat caulking. At least in parts of New Britain and Vanuatu it is used to catch birds (Powell 1976, Walter & Sam 2002: 109). It also has medicinal uses. On Manus (Admiralties) the latex is diluted and drunk as treatment for dysentry, diarrhoea and stomach ache (O’Collins & Lamothe 1989). On Vanua Lava (Banks Islands, Vanuatu) latex from the breadfruit is mixed with latex from Ficus adenosperma as a potion against excessive menstrual discharge (Bourdy & Walter 1994). It is also used in Vanuatu to ‘patch’ damaged yams for storage (Walter & Sam 2002: 109).

The Bola and the Nakanai of New Britain beat the inner bark into cloth and use it for articles of clothing, including rain capes and it was formerly also used in this way in Tonga (Floyd 1954, Powell 1976, Walter & Sam 2002: 109). Infusions of the bark and leaves are used to treat disorders of the digestive tract in parts of Vanuatu. The wood sometimes serves for making outriggers and paddles and for firewood.

Two POc terms for ‘breadfruit’ are reconstructed with reasonable certainty, *kuluR and *baReko, whilst a third, *beta, is less well supported.

By far the most widespread and frequently reflected of the three is POc *kuluR, inherited from PMP, with reflexes in the Admiralties and Mussau, North New Guinea, Papuan Tip, Bali-Vitu (MM), the Willaumez group (MM), New Caledonia and Central Pacific. It is apparently also reflected in the Chuukic subgroup of Micronesian with a change in denotation to Barringtonia asiatica. A variant form *kunuR is reflected in North New Guinea, languages of the Ngero-Vitiaz and Markham groups and in the Willaumez (MM) language Meramera (which has probably borrowed it from a Ngero-Vitiaz language).²³ There are also possible reflexes of *kunuR in Papuan Tip languages, but these are open to two interpretations. One is that they indeed reflect *kunuR. The other is that the Dobu, Kalokalo and Wedau forms below reflect *kuluR, as POc *l is regularly refllected as n in these languages, and that Suau and the Central Papuan languages (Balawaia, Motu, Roro and Mekeo) have borrowed from a language with a n. This second hypothesis encounters a problem. Although the Central Papuan languages have a number of loans from the Are-Taupota group to which Wedau belongs (Ross 1994a), Suau shows no sign of such loans. What is more, Suau (Daui) unuli and Suau (Kwato) ʔunuli are the expected reflexes of *kunuR, displaying a uniquely Suauic paragogic -i after the earlier final consonant, and Wedau kunori is clearly a loan from Suauic. Thus the first hypothesis is apparently correct: Papuan Tip languages also reflect *kunuR. In consequence, *kunuR can be reconstructed to PNGOc, and it is tentatively marked thus below - ‘tentatively’, because we would expect all North New Guinea terms to reflect the PNGOc etymon, but some of the Bel languages (Bilibil, Gedaged and Takia) and all the Schouten languages (Wogeo, Manam, Ali and Sissano are cited below) instead reflect POc *kuluR, and I have no explanation for this.

Reflexes of POc *baReko are found in New Ireland (MM), NW Solomonic (MM), SE Solomonic, Temotu, North-Central and Southern Vanuatu. In three Central Papuan languages, Lala, Roro and Kuni, the reflex of *baReko denotes ‘sago palm’, a shift from one starch source to another. In the sense ‘breadfruit’, reflexes of *kuluR~*kunuR and *baReko are geographically in complementary distribution. The distribution of *baReko forms an uninterrupted block across much of Island Melanesia excluding New Caledonia, with reflexes of *kuluR~*kunuR across most of the remaining Oceanic-speaking region. Just two reflexes of *baReko in the sense ‘breadfruit’ occur outside the block. One is Tawala beleha, which displays the wrong vowels and may be a chance resemblance. The other is Tomoip buŋ-biria.²⁴ Tomoip is situated near the languages of the Mengen family (NNG), but it is a New Ireland outlier (Ross 1988: 292-293) and as such is an extension from the block.

How are we to explain the complementary distribution? It is clear from non-Oceanic cognates and from its Oceanic distribution that the usual POc term for breadfruit was *kuluR. But on our criteria for POc reconstruction (ch. 1, §3.2.3) *baReko must also be reconstructed to POc, as it is reflected in Westem Oceanic (MM), SE Solomonic, Temotu, and North and South Vanuatu. The distributions suggest that POc *kuluR was the default term for Artocarpus altilis and that it was then replaced by *baReko in a block from New Ireland to South Vanuatu. However, the reflexes of *baReko are generally regular, suggesting that replacement took place very early in the history of Oceanic.²⁵

POc *baReko or perhaps *beta may have denoted Artocarpus camansi or a particular breadfruit cultivar or, for example, the cooked breadfruit, but these possibilities remain speculations.

Scattered areas - many North New Guinea languages on New Britain, scattered languages in the islands of the Papuan Tip area, much of Central and South Vanuatu, and Micronesia - have a breadfruit term other than *kuluR~*kunuR and *baReko. Two of these, POc(?) *beta and PROc *maRi, are sufficiently widespread to warrant mention.

POc (?) *beta has a more restricted distribution: its reflexes occur in the New Georgia group, within SE Solomonic only in Gela, and in North-Central Vanuatu. This distribution requires that it be reconstructed for POc. The fact that its only non-Eastern Oceanic reflexes are in Nduke and Roviana in New Georgia in the northwest Solomons raises the possibility that *beta is a PEOc term that has been borrowed westward, but there is no other evidence I know of for Eastern Oceanic loans in New Georgia.

A fourth term, PROc *maRi ‘breadfruit’, is limited to South Vanuatu, Micronesian and Polynesian, but Geraghty (2004: 88) attributes Polynesian reflexes of the form mei (presumably for †mai or †mal) to borrowing, presumably from a Micronesian source. Again the distribution requires an explanation, this time because it is bipartite, but I can provide none.

PMP		*kuluR	‘breadfruit, Artocarpus altilis’ (Dempwolff 1938)
POc		*kuluR	‘breadfruit, Artocarpus altilis’
Adm	Mussau	ulu	‘breadfruit’
Adm	Loniu	kun	‘breadfruit’
Adm	Titan	kul	‘breadfruit’
Adm	Ponam	gul	‘breadfruit’
NNG	Bilibil	uli	‘breadfruit’
NNG	Gedaged	ul	‘breadfruit’
NNG	Takia	ul	‘breadfruit’
NNG	Wogeo	kul	‘breadfruit’
NNG	Manam	kulu	‘breadfruit’
NNG	Ali	kul	‘breadfruit’
NNG	Sissano	uʎ	‘breadfruit’
MM	Vitu	kulu	‘breadfruit’
MM	Bola	ulu	‘breadfruit’
MM	Nakanai	ulu	‘breadfruit’
NCal	Pije	cin	‘breadfruit’
NCal	Nemi	cin	‘breadfruit’
NCal	Nyelâyu	yen	‘breadfruit’
NCal	Nêlêmwa	cen	‘breadfruit’
NCal	Iaai	i-oun	‘breadfruit’
Fij	Wayan	kulu	‘breadfruit’
Pn	Samoan	ʔulu	‘breadfruit’
PChk		*kulu	‘Barringtonia asiatica’ (Bender et al. 2003)
Mic	Chuukese	kuun	‘Barringtonia asiatica, Barringtonia racemosa’
Mic	Puluwatese	kuul	‘tree, used for fish poison’

PNGOc		*kunuR	‘breadfruit’
NNG	Kove	unu	‘breadfruit’
NNG	Gitua	unu	‘breadfruit’
NNG	Lukep	kun	‘breadfruit’
NNG	Sio	kunu	‘breadfruit’
NNG	Tami	kun	‘breadfruit’
NNG	Sukurum	gunik	‘breadfruit’
NNG	Adzera	guni	‘breadfruit’
NNG	Mato	uŋ	‘breadfruit’
NNG	Wab	un	‘breadfruit’
NNG	Bing	un	‘breadfruit’
PT	Dobu	ʔunu	‘breadfruit’
PT	Kalokalo	kunu	‘breadfruit’
PT	Wedau	kunori	‘breadfruit’ (loan from Suauic)
PT	Suau	unuli	‘breadfruit’
PT	Suau	ʔunuli	‘breadfruit’
PT	Balawaia	ʔunu	‘breadfruit’
PT	Hula	ʔunu	‘breadfruit’
PT	Motu	unu	‘breadfruit’
PT	Roro	unu	‘breadfruit’
PT	Mekeo	unu	‘breadfruit’
MM	Meramera	unu	‘breadfruit’ (borrowed from a NNG source)

POc		*baReko	‘breadfruit’ (French-Wright 1983)
PT	Tawala	beleha	‘breadfruit’ (for †bale[h]o)
PT	Lala	baleʔo	‘sago palm’
PT	Roro	pareʔo	‘sago palm’
MM	Lavongai	beiŋo	‘breadfruit’
MM	Tigak	bego	‘breadfruit’ (Beney 1980)
MM	Nalik	bərəuə	‘breadfruit’
MM	Tabar	bareu	‘breadfruit’
MM	Lihir	bale	‘Artocarpus altilis’ (Burley 2006)
MM	Tangga	bīe	‘breadfruit’
MM	Tomoip	(buŋ)biria	‘breadfruit’
MM	Nehan	bario	‘breadfruit’
MM	Halia	baleo	‘breadfruit’
MM	Tinputz	baniu	‘breadfruit’
MM	Teop	banio	‘Artocarpus altilis’ (Record 1945)
MM	Uruava	bareo	‘breadfruit’
MM	Babatana	bario	‘Artocarpus altilis’ (McClatchey et al. 2005)
TM	Äiwoo	ñi-bälo	‘breadfruit’
TM	Natügu	bia	‘breadfruit’
TM	Tanema	baloe	‘breadfruit’
TM	Buma	bale	‘breadfruit’
SES	Gela	baleɣo	‘a pair of breadfruit tied together’
SES	Lengo	beɣo	‘breadfruit’
SES	Talise	baleɣo	‘breadfruit’
SES	Longgu	baleʔo	‘breadfruit’
SES	Kwaio	baleʔo	‘breadfruit tree, breadfruit’
SES	Birao	baleho	‘breadfruit’
SES	Lau	bʷaleo	‘breadfruit’
SES	Kwara’ae	baleoʔ	‘breadfruit’
SES	’Are’are	pareʔo	‘breadfruit’
SES	Sa’a	pʷaleʔo	‘breadfruit’
SES	Arosi	bʷareʔo	‘breadfruit’
SES	Bauro	pareɣo	‘breadfruit’
NCV	Mota	pego	‘breadfruit sp.’
NCV	Nduindui	baeko	‘breadfruit’
NCV	Ambae	baego	‘breadfruit’
NCV	Nokuku	peko	‘breadfruit sp.’
NCV	Tamambo	baeho	‘breadfruit’
NCV	Raga	baeɣo	‘Artocarpus altilis’ (Walsh 2004)
SV	Lenakel	nu-(və)vaau	‘k.o. breadfruit with very large fruit’

POc		*beta	‘breadfruit’
MM	Nduke	beta	‘breadfruit’
MM	Roviana	beta	‘breadfruit’
SES	Gela	beta	‘breadfruit’
NCV	Mwotlap	b[e]te	‘breadfruit’
NCV	Mwesen	per	‘breadfruit’ (Bourdy and Walter 1994)
NCV	Naman	ne-bet	‘breadfruit’
NCV	Tape	pəte	‘breadfruit’
NCV	Avava	e-bet	‘breadfruit’
NCV	Lonwolwol	beta	‘breadfruit’
NCV	Paamese	vetā	‘breadfruit’

PROc		*maRi	‘breadfruit’ (Geraghty 1990)
SV	Ura	ni-mal	‘breadfruit’
SV	Sye	n-mar	‘breadfruit’
SV	Whitesands	nə-mei	‘breadfruit’
SV	Kwamera	ne-mer	‘breadfruit’
SV	Anejom̃	in-ma, in-mer-	‘breadfruit’
SV	Anejom̃	in-mer-	‘breadfruit’ (in compounds)
Mic	Kiribati	mai	‘breadfruit’
Mic	Marshallese	may	‘breadfruit’
Mic	Woleaian	māy	‘breadfruit’
Mic	Puluwatese	mǣy	‘breadfruit’
Mic	Chuukese	mǣy	‘breadfruit’
Pn	Tongan	mei	‘breadfruit’
Pn	East Futunan	mei	‘breadfruit’
Pn	Marquesan	mei	‘breadfruit’

Names of some breadfruit parts are reconstructed in chapter 4. They are:

• PWOc *kalijo ‘edible kernel of breadfruit segments’ (ch.4, §2.9)
• POc *malo- ‘breadfruit flower, breadfruit core’ (ch.4, §2.9)
• POc *bul[i,u]t ‘sap (of plant) or other sticky substance; be sticky’ (Micronesian and Polynesian reflexes denote the latex of the breadfruit tree; ch.4, §2.11)

5. Plants which store starch in their stems⇫

5.1. Metroxylon spp., sago palm, TP saksak, P hevinat, aevrinat, B natanggura (Arecaceae)⇫

The starch source in a Metroxylon sagu palm is the pith within its trunk. The palm only flowers and fruits once, and large stocks of carbohydrate are accumulated in the trunk in preparation for this event, after which the palm dies. This reserve of starch reaches its maximum just before flowering, when sago-eaters fell the palm and harvest its starch.

Sago palms are species of the genus Metroxylon. They grow in fresh-water swamps and wetlands, but not on permanently flooded sites. The species which produces most food starch, Metroxylon sagu, reaches 15 m in height and has a bole diameter without leaf sheaths of 35-60 em. Each palm is produced from a root stock which gives out several shoots, with the result that sago stands are naturally very dense. Other species vary in height, some smaller, some larger, the very largest, Metroxylon amicarum, growing to 33m.

There are a number of different species of Metroxylon found in the Oceanic-speaking region. Those upon which the literature (Barrau 1955:45–46, 1962: 140–143, Henderson & Hancock 1988: 34–35, McClatchey et al. 2006b) appears to agree are: ²⁶

Metroxylon sagu

New Guinea, Bismarcks, Bougainville, Choiseul, New Georgia²⁷

Metroxylon bougainvillense

New Guinea, Bismarcks, Bougainville, Choiseul, New Georgia²⁸

M. solomonense

Solomon Islands

Metroxylon warburgii

Solomon Islands, Vanuatu, Fiji, Rotuma, Samoa²⁹

Memecylon vitiense

Fiji

Metroxylon amicarum

Micronesia

It seems probable that sago starch was consumed by POc speakers. The POc homeland was in the Bismarcks (vo1.2, ch.2) and it is there and in areas nearby (New Guinea, Bougainville, Choiseul and New Georgia) that Metroxylon palms are used for food purposes. This is almost certainly related to the fact that this region is the domain of the two species, Metroxylon sagu and Metroxylon bougainvillense, which provide the best supplies of starch (Henderson & Hancock 1988: 34–35). Outside the sago-eating region sago species usually do not grow naturally but are cultivated for their leaves.

It is probably true to say that in Oceanic-speaking communities sago is only a staple where a marshy environment causes other staples to be in short supply. Except at locations in the Admiralties, it is always a supplement to garden staples, not a replacement. Although there are communities in New Guinea which get their sago entirely from natural palm stands, Oceanic-speaking communities probably also plant palms from suckers (transplanting seedlings is rare; Rhoads 1986).

Sago palms grow very quickly, up to 1.5 m of vertical stem growth per year. They are harvested at the age of 7 to 15 years just before they flower. The palm is felled, the trunk is cut into sections. These are either split lengthways or the bark is half peeled off, and in pre-contact times the pith was extracted with tools made from hard wood, rough stone or sharpened bamboo. Commonly the crushed pith is washed and pounded, crushed or wrung (POc *poRos, vol.1 ,ch.9, §7) in an inclined trough made from a sago leaf petiole (see Figure 9.7). The starchy liquid thus extracted from the fibrous residue runs through a coconut frondnetting strainer (POc *Runut, vol.1, ch.6, §5.6; this volume, ch.12, §5.2) and is collected in leaf pans to settle. The water is decanted so that the starch can dry. The dried starch is sago flour, which is almost pure carbohydrate, for which reason it is rarely eaten alone. To preserve it, sago flour is wrapped in sago leaves or placed in clay pots and simply dampened with water from time to time. It is sometimes eaten as a porridge with additional ingredients, sometimes made into a paste which is baked into a pancake or biscuit with other ingredients, and sometimes baked in a hollow bamboo over a fire (May 1984: 54).

At least on Malaita and in the Temotu archipelago chunks of unprocessed pith are sometimes baked as an emergency food (Kwa’ioloa & Burt 2001: 191, Henderson & Hancock 1988: 34). Throughout Oceania, sago leaves, which are pinnate, not palmate, provide excellent roofing thatch. They were evidently used for this purpose by POc speakers, who called the leaves and thatch *qatop (vol.1, ch.3, §3.4). Parts of the sago palm have a variety of uses, and not just in the sago-making process itself. As well as thatching, among the Bola of New Britain the leaves are used for making screens, pillows, and canoe sails, laths of skin from the leaf midrib are made into matting which serves as a fish trap, and midribs are used to make internal partitions and heavy stiff rope (Powell 1976). The Kwara’ae feed the pith to pigs. They use the leaf spines to make needles for stitching sago thatch and darts for shooting birds, as well as for making brooms. The leaf shoot is stained and used to make decorative armbands. The frond bases used to be used as stools, and a fallen palm becomes a source of sago grubs and sago fungus, both of which are eaten (Kwa’ioloa & Burt 2001: 191).

There was apparently just one POc term for the sago palm, namely POc *Rabia. As the Fijian and Polynesian reflexes below show, early Central Pacific speakers reassigned PCP *abia, reflecting POc *Rabia, to another starch source, Tacca leontopetaloides, Polynesian arrowroot (§2.3.2). Kikusawa (2003: 49) argues - correctly, I think - that PPn *pia ‘Polynesian arrowroot’ is cognate with Bauan yabia ‘arrowroot’, and reflects loss of initial *a- from PCP *abia.³⁰

PMP		*Rambia	‘sago palm’ (Blust 1989)
POc		*Rabia	‘sago, Metroxylon spp., mainly Metroxylon sagu (syn. Metroxylon rumphii)’ (Grace 1969: *r(a,u)bia)
Adm	Seimat	api	‘sago’ (Sorensen 1950)
Adm	Titan	api	‘sago’
Adm	Bipi	abi	‘sago’
NNG	Mengen	lamvia	‘sago’
NNG	Malasanga	labia	‘sago’
NNG	Kaiwa	labi	‘sago’
NNG	Kairiru	rabi	‘sago’
PT	Muyuw	yabiy	‘sago’ (F. Damon, pers. comm.)
PT	Kilivila	yabia	‘sago’
PT	Bwaidoga	labia	‘sago’
PT	Saliba	labia	‘sago’
PT	Balawaia	labia	‘sago’
MM	Bola	rabia	‘sago’
MM	Babatana	rabia	‘Nypa fruticans’ (McClatchey et al. 2005)
TM	Nebao	në-rë	‘sago’
TM	Asuboa	(lov)ne-te	‘sago’
TM	Tanibili	no-kʷo	‘sago’
TM	Buma	ɔtɔvɔ	‘sago’
SV	Kwamera	ni-epi	‘Metroxylon warburgii’
PCP		*abia	‘Polynesian arrowroot, Tacca leontopetaloides’
Fij	Bauan	yabia	‘arrowroot, starch, Tacca leontopetaloides’
PPn		*pia	‘Polynesian arrowroot, Tacca leontopetaloides’
Pn	Samoan	pia	‘Polynesian arrowroot, Tacca leontopetaloides’ (old term; Whistler 2000: 192)
Pn	Pileni	pia	‘Polynesian arrowroot, Tacca leontopetaloides’
Pn	Rarotongan	pia	‘Polynesian arrowroot, Tacca leontopetaloides’

In Ross (1993) I followed Dutton (1994) in reconstructing POc *sag(u) ‘sago starch’ as a continuation of Dempwolff’s (1938) PMP *sa[ŋ]gu ‘sago starch’. I now think this reconstruction is too insecure to stand. As Chowning (2001: 79) points out, some of the supporting data are semantically doubtful reflexes, and others may represent borrowings from Tok Pisin saksak ‘sago’. I still think that Kara (E) sa-sak ‘sago’, Solos sa-sak and Haku saka-saka, both ‘sago pancake’, may represent a continuation of Dempwolff’s etymon, but I cannot be sure.

A glance through chapter 4 shows that where the parts of a sago palm resemble those of the coconut palm, the same terms are used for both. Thus most of the terms in §5.1 of ch.12 probably also applied to the fronds of the sago palm. The petioles of some sago species are covered in thorns, for which the POc term was *ruRi or *druRi (ch.4, 2.12).

5.2. Cycas spp., cycad, Malayan palm fern, TP baibai, B namele (Cycadaceae)⇫

Cycads are palm-like plants of the family Cycadaceae (unlike sago, coconut, betelnut and the black, fan and nipa palms, all members of the family Arecaceae). Until a decade ago Cycas rumphii (syn. Cycas circinalis) was believed to be the only cycad species in Oceanic-speaking territory. However, recent work on the genus reported by Laubenfels & Adema (1998) has established that two species, Cycas rumphii and c, have almost identical distributions extending from the north coast of New Guinea eastward through the Bismarcks and into the Solomons with outliers in Micronesia. At least one of these is evidently also present in Vanuatu and New Caledonia, to judge from the data listed below.³¹

The description here is limited to Cycas rumphii, but, as some sources treat Cycas scratchleyana as a synonym of Cycas rumphii, I assume that the two species have almost identical features. Cycas rumphii grows 3-10m tall. Guppy (1906: 413), cited by Barrau (1965), considered C. rumphii to have been one of the early food sources of the Pacific Islands. Barrau mentions oral traditions about its use in New Guinea, Vanuatu, New Caledonia and Fiji. The seeds were and sometimes still are used at least as famine foods on islands throughout its Pacific distribution. Such is the strength of these oral traditions that Barrau concurs with Guppy’s view that it was at one time more than a famine food, pointing to the popularity of food made from the seeds, for example, on Guam. Other sources mention the seed pudding as something prized in places as far apart as the Ninigo Islands (Sorensen 1950) and Bellona (Christiansen 1975). Exploitation of the stem starch seems to have been much rarer in the Pacific (it is more common in parts of Asia; Thieret 1958, Whiting 1963), but it occurred at Wedau (north coast of SE Papua, Kahn 1986), in the Solomons and in New Caledonia. In the Solomons the starch was processed in much the same way as sago starch (and as with sago, the starch was harvested before fruiting; ch.9, §5.1 Parkinson 1907). In New Caledonia the stem starch was cut up and cooked (Jumelle 1907, cited by Thieret 1958). In Tonga the grated stem starch was rinsed and the starch that settled out was collected as a famine food (Whistler 1991b).

The seeds need considerable processing before they can be eaten, as they contain the toxin hydrocyanic acid. In the Ninigo Islands the otherwise poisonous fruit is submerged in salt water until the skin begins to peel off, then washed in fresh water and dried in the sun. This process removes the toxin, and the seeds are then pounded into flour, made into a pudding and baked (Sorensen 1950). The process on Guam was similar (Safford 1905, cited by Whiting 1963). On Rennell and Bellona Islands the seeds are wrapped in a fern leaf (Microsorium scolopendria) and soaked in water for five or more days, then pounded and baked (Henderson & Hancock 1988: 140). In Fiji the seeds were boiled until they were soft, then eaten (H. B. R. Parham 1943, cited by Thieret 1958).

Cycas rumphii also has a variety of other uses. Its bark sap is reported as a wood glue in Marovo and northern Malaita. In a number of places in the Bismarcks and the Solomons seeds are threaded on a string and used as a bull-roarer toy (Peekel 1984: 35, Henderson & Hancock 1988: 140, Hviding 2005: 143).

Various parts of the plant are used to cover small wounds and tropical ulcers. In the Bismarcks it is the resin of the fruit (Peekel 1984: 35, Holdsworth et al. 1982); in the Morobe Province of Papua New Guinea, the raw seeds (Barrau 1965); among Drehet speakers in Manus, it is sap obtained by cutting a notch in the bark (Beard n.d.); on Santa Ana (Solomons), the pulp of a scraped fruit. In the Reefs a bark preparation is used to treat a stomach ailment (Henderson & Hancock 1988: 140).

Barrau’s assertion that Cycas rumphii played an important role in earlier Oceanic cultures receives some support from the fact that several terms for it are reconstructable: POc *(p,pʷ)atoRu and *bai-bai(t), PEOc *mʷa(q)ele and PCP *loŋo-loŋo. The geographic distributions of these terms scarcely overlap.

The glosses of many of the terms listed below refer simply to ‘cycad’ or ‘Cycas’. None is identified as ‘Cycas scratchleyana’, and there is reasonable support for glossing all of them ‘Cycas rumphii’, but this is likely to be due to the fact that the two species were not distinguished until recently.

PMP		*patuRu	‘a cycad, Cycas rumphii’32
POc		*(p,pʷ)atoRu	‘a cycad, Cycas rumphii’
Adm	Mussau	otou	‘Cycas rumphii’
Adm	Seimat	haato	‘Cycas rumphii’ (Sorensen 1950)
Adm	Lou	paro	‘Cycas rumphii’
Adm	Drehet	potop	‘Cycas palm’
MM	Nakanai	patola	‘cycad’ (-a for †-u)

PMP		*ba(y)it	‘a cycad, Cycas rumphii’ (Ross 1996d)
POc		*bai-bai(t)	‘a cycad, Cycas rumphii’ (Ross 1996d)
PT	Minaveha	hai-hai	‘palm’
MM	Patpatar	be-be	‘cycad palm’
MM	Tolai	bai-bai	‘arborescent fern sp.; cycad palm’
MM	Nehan	be-be	‘cycad’
TM	Äiwoo	(nʷasi)poyi	‘Cycas rumphii’
TM	Äiwoo	(nʷasi)poyi	‘Cycas rumphii’
SES	Longgu	bai-bai	‘cycad palm’
SES	Lau	bai-bai	‘k.o. cycad; nut of this tied to strings and made to hum, a bull-roarer’
SES	Kwara’ae	bai-bai	‘Cycas rumphii’
SES	Baelelea	gʷae	‘cycad palm’
SES	Kwaio	bai-bai	‘k.o. cycad; nut used in whirling toy’
SES	’Are’are	pai-pai	‘k.o. cycad’

PCP		*loŋo-loŋo	‘a cycad, Cycas rumphii’
Fij	Bauan	loŋo-loŋo	‘Cycas rumphii’ (Ryan 1988: 140)
Pn	Tongan	loŋo-loŋo	‘Cycas rumphii’
Pn	Niuean	loŋo-loŋo	‘Cycas rumphii’
Pn	West Uvea	loŋo-loŋo	‘Cycas rumphii’
Pn	Tikopia	roŋo-roŋo	‘Cycas rumphii’
Pn	Emae	roŋo-roŋo	‘cycad’

Most of the forms below point to PEOc *mʷele. However, Namakir and Wayan point to *mʷali, Santa Ana and Paamese to *mʷa(q)ele, and Naman and Neve’ei also reflect a vowel sequence. Since the vowel sequence *ae is not known to have occurred in POc, I posit a medial *-q- and reconstruct PEOc *mʷa(q)ele.

PEOc		*mʷa(q)ele	‘a cycad, Cycas rumphii’
SES	Ulawa	mʷaele	‘Cycas rumphii’ (W. McClatchey, pers. comm.)
SES	Santa Ana	mʷaere	‘Cycas rumphii’
NCV	Mwotlap	mʷel	‘Cycas rumphii’
NCV	Mota	mʷele	‘cycad’
NCV	Ambae	mʷele	‘Cycas rumphii’
NCV	Kiai	mele	‘Cycas rumphii’
NCV	Raga	mʷele	‘Cycas sp.’ (Walsh 2004)
NCV	Naman	meil	‘Cycas rumphii’
NCV	Neve’ei	ni-mʷiyil	‘Cycas rumphii’
NCV	Tape	ni-mʷil	‘Cycas rumphii’
NCV	Uripiv	na-mʷel	‘Cycas rumphii’
NCV	Araki	(vi)mele	‘Cycas rumphii’
NCV	Tamambo	vu-mʷele	‘Cycas rumphii’
NCV	Paamese	maili	‘Cycas rumphii’
NCV	Lewo	(puru)mʷela	‘Cycas rumphii’
NCV	Namakir	mʷal	‘Cycas rumphii’
NCV	Nguna	na-mʷele	‘Cycas rumphii’
NCV	South Efate	na-mʷel	‘Cycas rumphii’
SV	Sye	no-mol	‘Cycas rumphii’
SV	Lenakel	nə-məl	‘Cycas rumphii’
SV	Kwamera	na-mur	‘Cycas rumphii’
SV	Anejom̃	no-mʷoc	‘Cycas rumphii’
NCal	Pije	ŋēn	‘Cycas rumphii’
NCal	Fwâi	ŋēn	‘Cycas rumphii’
NCal	Nemi	ŋēn	‘Cycas rumphii’
NCal	Jawe	ŋēn	‘Cycas rumphii’
NCal	Nyelâyu	mʷēŋ	‘Cycas rumphii’

6. Summary⇫

This chapter has treated POc terms for staple foods, i.e. sources of starch. ‘Starchy food’ was evidently a POc category of meal ingredient and was the hyponymous sense of POc *kanaŋ ‘food’. Additional ingredients to accompany starch were probably POc *tamaji.

The main root crops grown by POc speakers were evidently *qupi, the greater yam (Dioscorea alata), *pʷatika, the potato yam (Dioscorea bulbifera), and *talo(s) or *mʷapo(q), taro (Colocasia esculenta) (*talo(s) was the inherited term, *mʷapo(q) a borrowing from a New Guinea mainland Papuan language). They also grew *piRaq, giant taro (Alocasia macrorrhizos). The importance of the greater yam and Colocasia taro is attested by the fact that several terms can be reconstructed for (cultivars of?) each. The lesser yam, PWOc *kamisa or *mamisa (Dioscorea esculenta), seems to have been domesticated by early Oceanic speakers shortly after the break-up of POc.

The swamp taro, Cyrtosperma merkusii, was a later introduction, reaching Melanesia probably from Micronesia. The arrival of other modern root crops postdates the European presence in Central and South America: they are the sweet potato (Ipomoea batatas), cassava (Manihot esculenta) and Xanthosoma taro.

The banana, *pudi (Musa cultivars), and the breadfruit, *kuluR and *baReko (Artocarpus altilis) were also important sources of starch for POc speakers, but the sago palm, *Rabia (Metroxylon species), was probably no more than a supplementary or emergency starch source.

Notes⇫