Chapter 1.7 Canoes and Seafaring

Andrew Pawley and Medina Pawley

1. Introduction⇫

1.1. Questions⇫

Between 3000 and 1000 BC, speakers of Austronesian languages spread across Island Southeast Asia and western Melanesia and into the previously uninhabited islands of the Central Pacific.¹ By about 1000 AD they had colonised the whole of Polynesia and Micronesia and had settled Madagascar, more than half a world away.² This farflung maritime dispersal of closely related peoples, without parallel in human history until the West European expansion of the 15th-20th centuries AD, must have been based on a fairly efficient sailing technology. What was this technology? When and where did it develop? To what extent was it modified during the Austronesian diaspora?

There is already an extensive literature on these questions, much of it by culture historians who apply the typological comparative and distributional methods (discussed in §1.2) to sailing craft and sailing techniques. Our contribution will be to see what light can be thrown on these questions by comparing the vocabulary for watercraft and seafaring across the Austronesian family, using the genetic comparative method of historical linguistics. We will focus on the early phases of Austronesian expansion, that is, on cognate sets that probably go back more than 2,000 years. Little will be said here about the complex developments in boatbuilding which have taken place in Island Southeast Asia over the last couple of millenia, well after the main Austronesian dispersal across and beyond this region, or about modifications in the design of craft that according to Haddon and Hornell (Haddon 1937, Haddon & Hornell 1938, Hornell 1936) have taken place in various regions of Oceania within the last millenium. An exhaustive study of material relevant to reconstructing terms for vessel design and seafaring at all stages in the history of the Austronesian family would take several years and fill a very large book.

1.2. Reconstructions based on comparative typology: methodological problems⇫

Describing the sailing craft and navigation techniques in the Indo-Pacific region, and theorising about their origins and development have been popular pastimes among Western scholars since the first European explorers visited the region.³ The fairly close correlation between the distribution of outrigger canoes and that of the Austronesian language family has not escaped the attention of culture historians.

Those writers who have drawn conclusions about the nature of early Austronesian sailing craft and navigation have generally relied on the ‘typological comparative method’ of historical reconstruction. The typological method is a theory of structural types, making assumptions about how structural features are linked in systems, what kinds of changes are possible or likely, which types are logically prior to others, etc. The kinds of historical inferences this method can give, when applied to a range of contemporary systems, are probabilistic ones: for example, the inference that type X is more likely to have given rise to type Y than vice versa. The typological method is often allied to a distributional one. The assumption is that one can infer much about the antiquity of a given structural type or feature from its geographic distribution. For instance, the observation that a certain complex of cultural features is found among widely scattered peoples speaking related languages can be taken as evidence that the complex was present in a common ancestral culture. On the other hand, if such a complex has a more restricted but continuous geographic distribution it is likely to have been an innovation that diffused over this region after the dispersal of the ancestral population.

In his Sailing Craft of Indonesia Adrian Horridge (1986:2-3) employs a distributional argument when he writes:

The very wide distribution of their homogeneous cultural heritage shows that [the ancestral Malayo-Polynesians]⁴ had an excellent knowledge of outrigger canoes for transporting fire, family, pigs, chickens and dogs, not to mention dozens of useful plants, by sea.

Other passages in the same book present some fairly detailed conclusions about the design of the vessels. These are based partly on the logic of technology—what is possible and what is likely in the development of techniques, given certain equipment and natural resources, engineering knowledge, navigational knowledge, climatic conditions, etc.—and partly on distributional grounds.

The signs are that the original Malayo-Polynesian rig was a two-boom triangular sail fixed by the point (tack) in the bows of the boat and held up by a loose prop, with a rope to the outriggers to prevent it falling sideways. (p.56)

All Malayo-Polynesian rigs therefore had the fundamental property that the sail could be tilted fore and aft to balance the sail with reference to the balance of the hull and load on the steering paddle. In principle they could all be steered like a windsurfer, simply by tilting the sail fore and aft. These rigs, like the outrigger canoes for which they were adapted, were invented in Island South-East Asia, and spread with the Malayo-Polynesian expansion, although they may have been known in Indonesia before that time. (p.58)

The idea of a boat built from sewn planks was known to the Polynesian migrants into the Pacific, and perhaps came from mainland Asia before 5000 Be, but the idea of fixing the planks edge-to-edge with dowels seems to have spread later from the mainland, with the same distribution as the use of metal for boat-building tools … All the techniques needed to make a lashed-lug boat, sewn of course, were known to the earliest Polynesians, and the projecting lugs carved in situ are a feature of many traditional Polynesian, Micronesian and Melanesian boats. (pp.57-58)

In the following passage, Horridge combines structural and distributional arguments with archaeological evidence:

The earliest evidence of trading by boats that could beat against the wind is provided by pottery with a particular design, known as Lapita, that spread rapidly from Western Melanesia far into what is now Polynesia about 3000 years ago. The vessels carrying the potters were probably double canoes because nothing else in the region is sufficiently seaworthy. (p.4)

The strength of technology-based comparative studies is in the detailed historical inferences concerning design and construction which they yield. That is not to say such inferences are necessarily reliable. There are some serious methodological weaknesses associated with all reconstructions based purely on the logic of types and the distributional method.

Because innovations in material culture sometimes spread and replace older usages and because some innovations are made independently in different places, the widespread geographic distribution of a feature or even a complex of features does not guarantee its great antiquity. Nor does the restricted distribution of a feature guarantee that it is a recent innovation. The method of comparative typology is unable to distinguish in a reliable way between ‘inherited’ and ‘borrowed’ elements or features within a continuing community or cultural tradition, that is, between institutions which have been handed down from generation to generation within the community or tradition since a given point in time, and institutions that have entered the tradition from outside since that point in time. The method is also unable to distinguish reliably between ‘retentions’ and ‘innovations’ within the same tradition, that is, between features that have been part of a tradition since a given point in time and those that were developed later.

In Indonesia and contiguous regions of Island Southeast Asia the problems of distinguishing retentions, innovations and borrowings in sailing technology are particularly acute. There, the local sailing traditions of Austronesian-speaking communities have not only diversified and influenced one another but have been exposed to numerous alien traditions coming from mainland Asia. Horridge acknowledges this point:

The Malayo-Polynesians diversified as they spread, and from them are descended the specialized boat-building and boat-loving maritime groups of Indonesia, namely the Bajau or Sea Gypsies, the Buginese from the Gulf of Bone, the Makassarese, the Mandar people from West Sulawesi, and the Butungese from South-east Sulawesi, and the Madurese … [and] the fishennen of the islands of Bawean, Masalembu and Sepudi in the Java Sea, the traders of Bonerate and Pulau Palu’e in the Flores Sea, the whalers of Lamalerap on Lomblen in the Timor Straits, the men of Luang in the Barat Daya Islands, and the numerous Buginese colonies which control a wide network of trade in miscellaneous goods. All these very diverse groups have inherited the Malayo-Polynesian seafaring tradition, and methods of building outrigger canoes which over the past two millenia have been mixed with traditions from the Indian Ocean and the West to give the modern hatch-patch of boat and canoe styles. (pp.3-4) [our italics: AP & MP.]

Horridge’s remarks remind us that the testimony of central Pacific cultures is likely to be crucial in reconstructing early Austronesian sailing technology. The relative isolation of the peoples of the more remote Pacific islands may have allowed some of them to continue the early Austronesian sailing culture with fewer changes than most peoples in Island Southeast Asia.

A problem arises when students of comparative technology dabble in comparative linguistics without underpinning them by the careful studies of sound correspondences and subgroupings needed to distinguish between cognates, accidental resemblances and borrowings or to determine the relative chronology of linguistic innovations. The results of such dabblings will be largely worthless. Haddon and Hornell’s (Haddon 1937, Haddon & Hornell 1938, Hornell 1936) admirable survey, Canoes of Oceania, is marred by a number of fanciful historical speculations which rest in part on naive readings of linguistic similarities. To a much larger extent, however, their fanciful historical inferences rest on a diffusionist interpretation of the comparative material culture, uninformed by systematic study of comparative vocabulary or by archaeological evidence.

1.3. The Genetic Comparative Method⇫

There is a means of escape from the limitations of comparative typology. It is the special virtue of the comparative or genetic method of historical linguistics that it can, in principle, (a) define genetic continuity in certain parts of the vocabulary of each language in a language family, distinguishing resemblances due to common origin from resemblances due to borrowing, and (b) assign a relative chronology to innovations occurring within the languages of a linguistic family.⁵

However, linguists need not feel unduly smug about the genetic comparative method. It applies only to a restricted part of each language-culture system, namely the stock of morphemes that have cognates in genetically related languages. Often it happens that linguistics is silent in the face of competing hypotheses derived from comparative technology, say, about techniques of manufacture, because no distinctive terms can be reconstructed for the technological elements in question. And while the genetic method often allows us to make strong inferences about the presence of some things in a reconstructed language-culture system, it does not tell us what was absent. That is to say, our inability to reconstruct a term for a particular element in protolanguage L is not conclusive proof that speakers of L lacked a term for that element. Within these limitations, the genetic comparative method remains a powerful tool.

1.4. Lexical sources⇫

Many previous publications isolate Austronesian cognate sets referring to canoe parts and seafaring. However, works compiled by linguists only list such cognate sets as isolated items within a larger body of semantically unordered material. This chapter draws together published comparisons and extends them. Sources giving relevant cognate sets or sound correspondences relevant to determining cognation include Bender et al. (1983), Biggs (1978, 1993),⁶ Blust (1970, 1972b, 1976a, 1978a, 1978b, 1980b, 1983–84a, 1986, 1989), Clark (1994), Collins (1983), Dahl (1973), Dempwolff (1938), Ferrell (1969), Geraghty (1983), Jackson (1983), Milke (1961, 1968), Pallesen (1985), Pawley (1972), Reid (1971), Ross (1988), Sneddon (1984) and Tryon (1976). We have also consulted dictionaries of many contemporary languages.

1.5. Organisation of reconstructions and cognate sets⇫

Some fifty cognate sets will be discussed. These are ordered by semantic domains, such as type of vessel, beginning with types of vessels and hull construction and going on to outrigger structure, superstructure, sail and rigging, accessories, launching, beaching and anchoring and ending with terms for seafaring and seafarers.

Each reconstructed word or etymon is attributed to a certain level of the Austronesian family tree, the highest level justified by the distribution of the cognate set across subgroups. Below it are listed the cognates from contemporary languages. These are ordered according to subgroup, usually proceeding roughly in a west to east direction, e.g. Western Malayo-Polynesian cognates precede CMP cognates which in turn precede Eastern Malayo-Polynesian cognates. As well as giving the highest-level reconstruction we often give intermediate-level reconstructions for named interstages, especially when the the interstage reconstruction differs significantly in form or meaning from the highest-order reconstruction. When listing cognates we occasionally acknowledge sources when the evidence comes from Haddon (1937), Hornell (1936) or Haddon and Hornell (1938) or a source other than a published dictionary.

2. Types of vessel and hull construction⇫

2.1. Types of vessel⇫

Haddon and Hornell (1938: 70-72) referred to ten or so putative cognate sets having the general sense of ‘canoe’ or ‘boat’ that go back to an early stage or stages of Austronesian. While early Austronesian speakers probably had several different named types of craft, most of Haddon and Hornell’s etymologies are false. Only one of their putative cognate sets can be attributed to an early Austronesian interstage—that pointing to PCEMP *waŋka, POc *waga.

POc *waga (phonetically [wáŋga] with [ŋg] being a single phoneme) is widely reflected within Oceanic as a general term for a canoe or boat with a hull, as opposed to a raft. In many Oceanic languages, however, the reflex of *waga refers chiefly to large sailing canoes and other large vessels, in contrast to dugout canoes and small outrigger canoes. This range of meanings, taken together with the cognate set under PMP *katiR, suggests that *waga may have had two senses in POc.

PCEMP		*waŋka	‘outrigger canoe’
CMP	Buru	waga	‘boat, canoe’
CMP	Komodo	waŋka	‘boat, canoe’
CMP	Manggarai	waŋka	‘canoe’
SHWNG	Mor	waʔa	‘canoe’
SHWNG	Dusner	wak	‘canoe’
SHWNG	Numfor	wa(i)	‘outrigger canoe’
SHWNG	Waropen	wa	‘boat, canoe’
POc		*waga	(1) ‘large sailing canoe’; (2) ‘canoe (generic)’
Adm	Wuvulu	wa	‘canoe’
Adm	Seimat	wa	‘canoe’
NNG	Gedaged	wag	‘large canoe that goes out on the high seas, has one or two masts and a large platform, ship or boat’
NNG	Gitua	waŋga	‘canoe’
NNG	Yabem	waŋ	‘canoe, boat, ship’
PT	Dobu	waga	‘sailing canoe’
PT	Molima	waga	‘canoe in general’
PT	Kilivila	waga	‘generic term for all kinds of sailing craft’
PT	Wedau	waga	‘large canoe’
NCV	Mota	aka	‘canoe’
NCV	Mota	aka (paspasau)	‘canoe with plank sides’
NCal	Nyelâyu	waŋga-n	‘his canoe’
NCal	Xârâcùù	kwā	‘canoe’
Mic	Kiribati	wā	‘canoe’
Mic	Marshallese	wā	‘canoe’
Mic	Puluwatese	wā	(1) ‘canoe, vehicle of any kind’; (2) ‘container, people in a canoe’
Mic	Puluwatese	wā(herak)	‘large sailing canoe able to face the high seas’
Mic	Chuukese	wā	‘canoe, boat, vehicle’
Mic	Woleaian	wa	‘generic for all canoes’
Fij	Bauan	waga	‘generic for boats (traditionally canoes) of all kinds’
Pn	Tongan	vaka	‘boat (generic)’
Pn	Samoan	vaʔa	‘boat (generic)’
Pn	Samoan	vaʔa (folau)	‘sailing boat (for long voyages)’
Pn	Tokelauan	vaka	‘canoe, craft, boat’
Pn	Tikopia	vaka	‘canoe (generic)’
Pn	Tikopia	vaka (tapu)	‘consecrated voyaging canoe’
Pn	Hawaiian	waʔa	‘canoe’

The primary sense of POc *waga was perhaps ‘large sailing canoe’ in contrast to paddling canoes (dugouts and small outrigger canoes). This sense is widely reflected across subgroups of Oceanic. As the name of the largest and most prestigious type, *waga would have been a natural choice as a generic term for all types of canoe and in a number of languages the generic sense has become primary. An example is from Kilivila, the language of the Kiriwina people of the Trobriand Islands (reported by Haddon (1937:267-269), based on Malinowski (1922) and correspondence with Malinowski). In Kiriwina there are four named types of canoes. The kewoʔu is a simple dugout with outrigger, used in the lagoon. The kalipoulo, a fishing canoe, is a larger dugout with several designs. These have in common a hull built up with a washstrake on each side, and transverse carved and painted breakwaters, and the hull usually has pointed ends, carved and painted. There is often a platform over the booms. The sail is a lateen. The masawa is a large trading canoe, similarly constructed to the kalipoulo but with two clinker-built washstrakes on each side and twenty or more booms covered by a continuous platform. It carries a large, elongated steering oar worked by two men. The nagega is larger and more seaworthy than the masawa, with higher sides and more carrying capacity and a central standing mast, as opposed to a leaning mast stepped within the hull and shored by a prop. The generic term for all these craft is waga. In another part of Papua, around Samarai, Abel (1902:62) reports that vaga is the generic for all kinds of sailing canoes, but evidently excluded canoes without outriggers or sails.

Cognates belonging to the set above are sometimes placed together with forms listed below:

WMP	Malay	waŋkaŋ	‘Chinese junk’
WMP	Javanese	waŋkaŋ	‘Chinese junk’
WMP	Ngaju Dayak	vaŋkaŋ	‘Chinese junk’

and:

PAn		*ba(ŋ)ka(q)	‘outrigger canoe, dugout canoe’
Fma	Kuvalan	baŋka	‘canoe’7
WMP	Aklanon	baŋca	‘canoe’
WMP	Balangaw	baŋka	‘canoe’
WMP	Cebuano	baŋkaʔ	‘one-piece dugout between 5 and 15 metres, optionally with one or two masts and outrigger’
WMP	Ilongot	baŋka	‘canoe’
WMP	Tagalog	baŋkaʔ	‘canoe’
WMP	Tausug	baŋkaq	‘dugout canoe (without outrigger)’
WMP	Laiyolo	biŋka	‘canoe, boat’
WMP	Tolaki	baŋga	‘canoe’
PCEMP		*ba(ŋ)ka	‘canoe’
CMP	Larike	haka	‘ship’
CMP	S. Nuaulu	haka	‘ship’
CMP	Kola	boka	‘ship’
POc		*paka	‘ship’
MM	Roviana	vaka	‘ship’
SES	Gela	vaka	‘foreign vessel, European ship’
SES	Sa’a	haka	‘ship’

Although the forms in the above two sets show a striking resemblance to POc *waga, they are not demonstrably cognate. In the first set, initial w- is irregular, suggesting that these are borrowed words, though the source of the borrowing is unclear (Alexander Adelaar, pers.comm.). In the second set, the uncharacteristic occurrence of the cluster ŋk in the Philippine and Formosan forms suggests borrowing from a Malayo-Javanic or Sulawesi source (Lawrence Reid, pers.comm.). Tagalog and Cebuano b are not the regular reflexes of PMP *w. At present, most of the few coastal Austronesian-speaking communities left on Taiwan proper use large bamboo sailing rafts for fishing. However, there is evidence that in the nineteenth century other kinds of craft including double outriggers were in use. Scott (1982:337) cites an eighteenth century Chinese observer, Huang Shi-ching, describing a built-up dugout canoe called a manka [or banka] as follows:

A manka is a single tree trunk hollowed out, with wooden planks fastened on both sides with rattans; since they have no putty for caulking and water easily enters, the barbarians keep baling with a ladle.

It is likely (Mark Donohue, pers.comm.) that the Chinese characters in question are from the Hokkien dialect and should be read as banka, not manka as reported by Scott. The form *ba(ŋ)ka is however reflected both in CMP and Oceanic, and must be attributed to their common ancestor, which Blust (1993) labels PCEMP.

The next term is reconstructable as far back as PMP. But note the disagreement between WMP and Oceanic witnesses as to its meaning.

PMP		*katiR	‘(small) outrigger canoe or canoe hull’
PWMP		*katiR	‘outrigger float’
WMP	Malay	katir	‘(small) outrigger canoe or canoe hull’
WMP	Madurese	kater	‘(small) outrigger canoe or canoe hull’
WMP	Sundanese	katir	‘(small) outrigger canoe or canoe hull’
WMP	Maranao	katig	‘(small) outrigger canoe or canoe hull’
WMP	Tausug	katig	‘(small) outrigger canoe or canoe hull’
WMP	Cebuano	katig	[N] ‘float of the outrigger’; [V] ‘provide a boat with outrigger’
WMP	Sasak	katir	‘carry between two persons’
POc		*kati(R)	‘(small) outrigger canoe or canoe hull’
NNG	Manam	kati	‘outrigger canoe’
NNG	Kairiru	qat	‘outrigger canoe’
NNG	Tuam	kat	‘canoe platform’
PT	Motu	asi	(1) ‘hull of large multi-hulled canoe (finished more roughly than single-hulled canoe)’; (2) ‘large canoe’
PT	Motu	asi-asi	‘temporary small double canoe’
PT	Sinaugoro	gasi	‘outrigger canoe’
PT	Roro	ahi	‘canoe, hull’
MM	Lavongai	kati	‘large outrigger canoe 50 or more feet in length’ (Haddon 1937: 141)

Although the WMP cognates denote ‘outrigger float’, there is a much stronger candidate for that meaning in PMP, namely *(c,s)a(R)man (§4.1). Therefore we conclude that *katiR is unlikely to have meant ‘outrigger float’ in PMP and that the WMP cognates probably show semantic change. In Oceanic, reflexes of *katiR are confined to Western Oceanic but are widely dispersed within that large group. The Western Oceanic comparisons strongly point to an earlier meaning such as ‘canoe hull’ or ‘small outrigger canoe’.

If POc *kati(R) referred to a small outrigger canoe, then the following set may reflect a term for a large outrigger canoe (perhaps synonymous with *waga in its narrower sense):

POc		*tola	‘large canoe’ (?)
PEAd		*n-tol(V)	‘large canoe’
Adm	Titan	drol	‘large canoe’
Adm	Papitalai	dual	‘large canoe’
Adm	Nali	droi	‘large canoe’
Adm	Lele	dol	‘large canoe’
Adm	Ere	drual	‘large canoe’
Adm	Bohuai	coal	‘large canoe’
Adm	Mondropolon	col	‘large canoe’
Adm	Drehet	kxoŋ	‘canoe, boat’
Adm	Ponam	hol	‘large canoe’
Adm	Loniu	ton	‘large canoe’
PSES		*tola	‘plank-built canoe’
SES	Bugotu	tola	‘large canoe’
SES	Gela	tola	‘plank-built canoe with both ends turned up not very high’ (Haddon 1937: 100)
SES	Lau	ola	‘plank-built canoe (built up from a plank centre keel, with bow and stern keels), with no outrigger’
SES	Baegu	ola	‘canoe’
SES	Kwai	ola	‘canoe’
SES	Arosi	ora	‘large canoe’

Evidence for the reconstruction occurs in two geographically separated primary subgroups of Oceanic, but apparently nowhere else. Despite this distribution, the evidence would be clear enough, were it not for (i) the meaning of the PSES term, reconstructable as ‘plank-built canoe’, and (ii) the fact that the Lau, Baegu, Kwai and Arosi reflexes could well reflect, not *tola, but POc *(q)oRa ’strake, probably topstrake (washstrake) (§2.4) (Cristobal-Malaitan languages regularly lose POc *t). It is possible that the forms ola/ora reflect a falling together of *tola and *(q)oRa, the meaning narrowing in favour of *(q)oRa and then transferred to Bugotu, Gela tola.

Evidence from other Southeast Solomonic languages further confuses the issue. Gela and Tolo (Guadalcanal) tiola ‘generic term for plank-built canoes’ and ‘Are’are iora, Sa’a iola ’plank-built canoe’ might be explained as deriving from a bimorphemic form *ti-(q)oRa. This might be liable to conflation with reflexes of *tola also.

2.2. Double canoes⇫

It has been suggested (Roger Green, pers.comm.) that the ocean-going double-hulled canoe was an innovation of Oceanic speakers. He argues that it was large double canoes, stable and able to carry big loads while being sailed with traditional rig, that were the key to the transport of people, crops and domestic animals to the far-flung islands of Remote Oceania (the Pacific islands east and north of the New Guinea area, the Bismarck Archipelago and the main Solomon Islands chain).⁸ The design of double canoes varied in some details from place to place but the main structure consisted of two dugout hulls, placed parallel and usually one to two metres apart, joined by booms, with a platform built amidships. In the most efficient craft, represented by the Fijian drua, one hull was slightly smaller than the other. At the time of first European contact such craft were almost wholly confined to Remote Oceania, being present in New Caledonia and Fiji, in many parts of Polynesia and in a restricted region of the central Caroline Islands. In western Melanesia double canoes were made by the Mailu, of south-east coast Papua. The Mailu speak a non-Austronesian language but many of their canoe terms (and other parts of their vocabulary) are from Austronesian; it seems likely that the population of the Mailu area was once largely Austronesian-speaking. The Motu laɣatoi, a multi-hulled craft, can be derived from the double canoe.

No term for double canoe can safely be reconstructed for any very early Austronesian interstage. However, a number of Oceanic languages reflect the following form:

PEOc		*paqurua	‘double canoe’
SES	Lau	foorua	‘outrigger canoe’
Mic	Kiribati	baurua	‘large single outrigger voyaging canoe’9
Fij	Rotuman	foulua	‘ship’ (probably a Pn borrowing)
PPn		*faʔurua	‘double canoe’
Pn	Niuean	faulua, foulua	‘ship’
Pn	Rennellese	haʔugua	‘the double canoe in which the ancestor Kaituʔu sailed when he discovered Rennell and Bellona’
Pn	Tuvalu	foulua	‘double canoe’10
Pn	Māori	hourua	‘double canoe’

cf. also:

Fij	Bauan	drua, waga drua	‘ocean-going double canoe’
Pn	Samoan	fau-tasi	‘large whaleboat’ (possibly a nineteenth century coinage)
Pn	Manihiki	waka tau-rua	‘double canoe’
Pn	Hawaiian	waʔa kau-lua	‘double canoe’ (vs waʔa kau-kasi ‘single-hulled canoe’)

PEOc *paqurua is a compound formed from POc *paqu(s) ‘bind, lash; construct by tying together’ (Ch. 9, § 10) and *rua ‘two’. While it is possible that Lau foorua is borrowed from a Polynesian language, the details of form do not support this notion. Lau has the independent bases foo ‘bind’ and rua ‘two’. Haddon (1937: 77) reports Kinilaulau (Carteret Islands) haulau, holua ‘outrigger canoe’, but this term is probably borrowed from a Polynesian language believed to have been formerly spoken on the island or from nearby Takuu.

Friederici (1928:31) suggests that the double canoe of Oceania (or at least the type with smaller and finer-pointed port hull) originated from a canoe with a single outrigger, on the grounds that (a) in the double canoes of Polynesia and Fiji one of the two hulls is usually smaller, the smaller hull being called by the term for the outrigger float (e.g. hama in Tongan), and (b) the connecting poles between the hulls are also called by the word for the outrigger booms (kiato in Tongan). On the other hand, Haddon and Hornell (1938: 43) argue that the most likely origin of the double canoe is from two dugouts lashed together or a short distance apart. They also note the possibility that the single outrigger canoe may ultimately be derived from the double canoe by reducing the port hull to form a float. As the smaller hull and the float and the connecting poles serve the same purpose in both types of vessels, the extension of terms seen in Tongan hama and hato is natural. On logical grounds alone it is hard to choose between these historical interpretations. And in this case the linguistic evidence is relatively unhelpful.

Numerous other reconstructions, attributable to interstages lower than PMP or POc, can be made for types of craft. For example, the following doublets (distinct words coexisting in a language whose similar form and meaning suggest they ultimately trace back to the same source) can be reconstructed:

PWMP		*padaw	‘kind of sailboat’ (Blust 1983–84a: 90)
WMP	Maranao	padao	‘sailboat’
WMP	Malay	layar padau	‘storm sail’
WMP	Cebuano	paraw	‘schooner, galleon’

PWMP		*paraSu	‘boat’ (Dempwolff 1938)
WMP	Toba Batak	parau	‘boat’
WMP	Javanese	perau	‘boat’
WMP	Malay	perau	‘boat’
WMP	Ngaju Dayak	parau	‘boat’

These forms have sometimes been compared with Oceanic forms represented by Tongan folau ‘voyage, travel by sea’, Fijian volau ‘boat shed’ (cf. *pa-laSud, §9.2).

A PPn reconstruction for a small canoe used close to shore is well supported:

PPn		*paopao	‘small outrigger or dugout canoe for inshore use’
Pn	Tongan	pōpao	‘roughly made dugout canoe with two boom outrigger’
Pn	Samoan	paopao	‘small outrigger canoe with two booms’
Pn	Tikopia	paopao	(1) ‘craft made from log simply hollowed out, not built up’; (2) ‘canoe, sea-going but not sacralised, in contrast to vaka tapu’
Pn	Nukuoro	paopao	‘double-ended single outrigger with two booms’

A possible cognate occurs in a single Western Oceanic language, Mekeo of the Central Province of Papua. East Mekeo has papao (dialect fafao) ‘small canoe for children, also used as a trough for feeding pigs’. Reduction of the first vowel cluster of a reduplicated word is characteristic of Mekeo (Alan Jones, pers.comm.). This comparison points to POc *paopao ‘small outrigger canoe’, though it needs strengthening by further cognates beyond Polynesian.

Blust (1986:33) reconstructs the term *dakit, noting a number of reflexes within the WMP region and a regular reflex in one Oceanic language, Motu. To these may be added many other Philippine and Northern Sulawesi reflexes (given in Reid 1971) and possible reflexes in CMP and Western Oceanic.

PMP		*dakit	[N] ‘raft’; [V] ‘join along the length’ (Blust 1986)
WMP	Aklanon	gākit	‘raft’
WMP	Gaddang	gakit	‘raft’
WMP	Isneg	gākit	‘raft’
WMP	Ivatan	dākit	‘raft’
WMP	Kankanay	lakit	‘raft’
WMP	Malay	rakit	‘raft’
WMP	Sangir	hakiq	‘raft’
WMP	Sundanese	rakit	‘raft’
WMP	Wolio	rakii	[v] ‘join along the length’
WMP	Balangaw	dakit	‘put very close’
WMP	Sika	dakit (wii)	‘(earth, flour, resin) cleave, stick together’
WMP	Maloh	daʔit	‘raft’
WMP	Maloh	dakit	‘join along the length’
CMP	Taliabo	(ka)haŋkiʔ	‘raft’ (Charles Grimes, pers.comm.)
POc		*raki(t)	‘raft; join outrigger, join two hulls’
PT	Motu	rai	‘prepare a canoe for the sea; tie the outrigger on; tie two canoes to make an irai (double-hulled canoe)’

Chowning (1985: 59) gives the following as possible cognates, although the final vowel is problematic:

NNG	Kove	laɣe	‘raft’
MM	Nakanai	lage	‘raft’

2.3. Hull construction⇫

In pre-European times three basic hull designs were found in Austronesian-speaking communities: (a) a dugout built from a single log; (b) a five-part canoe, composed of dug-out hull, with the sides raised by sewing one side piece or strake to each side, with forked, crutch-shaped pieces at each end; and (c) a built-up canoe, in which a number of planks or strakes are added to a keel. The keel may consist of a thick plank or a dugout underbody (or sometimes two or even three dugouts joined).

Ethnologists have argued on logical grounds that types (b) and (c) developed from the dugout prototype. In his account of Philippine boat-building in the sixteenth and seventeenth centuries, Scott (1982: 337-338) sketches a theory of the evolution of the hull of Philippine boats:

In the ship-building technique [developed in China and Europe in the Middle Ages] … a rigid framework of keel and ribs is first constructed … and the wooden planking of the hull then nailed to it with metal spikes or wooden trenails. The older technique was to build the hull first, plank by plank carved to fit, and to fasten the ribs in afterwards. This technique is probably a natural development of the one-log dugout canoe by adding one board to each side to obtain higher freeboard.

By increasing the number of such additional planks, a fully developed boat or ship is produced. But as the sides of the canoe, or banca, are thinned, some transverse strengthening is required, and this can be provided by running strut-like thwarts across the vessel, securing them to the sides without nails by means of tambukos [lugs] and lashing. For this purpose a flexible rib can be pressed down across all of them and lashed securely to the matching tambukos carved on each plank. Finally a combination of such thwarts and ribs lashed together … produces a sturdy vessel whose hull and other structural parts are held firm under prestressed tension.

The question arises as to whether any or all of these features of hull design are attested in the vocabulary of PMP and other interstages.

2.4. Planking⇫

The comparisons below and in §2.5 point to the use by speakers of PMP and its immediate descendants, of boats built up by planking and strengthened by thwarts:

PMP		*papan	‘plank (of boat +), strake’
WMP	Malay	papan	‘plank’
WMP	Madurese	papan	(Horridge)
WMP	Bajau	papan	(Horridge)
WMP	Buginese	papan	(Horridge)
CMP	Buru	papa-n	‘plank’
POc		*baban, *bapan	(1) ‘plank’; (2) ‘canoe plank or strake’
PT	Molima	baba	(1) ‘lower plank on canoe’; (2) ‘put plank on canoe’
PT	Wedau	papana	‘built-up canoe’
SES	Lau	baba	‘long side board of canoe’
SES	Sa’a	hapa	(1) ‘plank’; (2) ‘thwart of canoe’
SES	Bugotu	pava	‘plank’
Mic	Kosraean	pæp	‘sides of canoe’
Mic	Puluwatese	pāp	‘board; canoe planks’
Mic	Chuukese	pape-n (wā)	‘canoe strake’ (pape- ‘plank’, -n ‘construct suffix’, wā ‘canoe’)
Fij	Bauan	bava	‘washstrake or upper planks of canoe’
Fij	Bauan	baba	‘side planks of canoe’
Pn	Tongan	papa	‘board, flat, hard surface’
Pn	Hawaiian	papa	‘flat, hard surface, board’

PMP *papan evidently referred to any plank or board (see Ch. 3, §3. S). It might be argued that this term could have been independently applied to canoe strakes or planks by different daughter languages after the break-up of PMP. However, the fact that in diverse Malayo-Polynesian languages reflexes of *papan are consistently used of canoe planks, even in cases when another general word for plank or board has developed, is a fairly strong indication that this application goes back to PMP times.

POc		*(q)oRa	‘strake, probably topstrake (washstrake)’
MM	Mono-Alu	ora	‘median strake at each end, above keel strakes on a canoe with dugout underbody’
Pn	Niuean	oa	‘washstrake, grooved and drilled for lashing to the hull’
Pn	Samoan	oa	‘gunwale and gunwale flange’
Pn	Pukapukan	oa	‘washstrake’
Pn	Rarotongan	oa	‘sideplank or planks of a canoe, lashed to the main body; the gunwale or seaboard of a canoe’
Pn	Tokelauan	oa	‘gunwale’
Pn	Tikopia	oa	‘top strake, gunwale’
Pn	Māori	oa	‘side boards of a canoe’

Possible cognates of *(q)oRa occur in Cristobal-Malaitan languages of the Southeast Solomonic family: Arosi ora, Lau ola ‘plank-built canoe (built up from a plank centre keel, with bow and stern keels), with no outrigger’. As noted in §2.1, however, their interpretation is difficult, as they could also reflect POc *tola ‘large canoe’.

The question of how planks were joined is connected to the question of whether early Austronesian communities used stone or metal tools, as the following remarks by Scott (1982:338-339) make clear:

Historically there are two methods by which the planks in [Philippine] plank-built boats are fastened together—sewing and edge-pegging. Sewing—or, better said, lacing—the boards together is done by drilling a matching row of holes through the two boards near their adjoining edges, and running rattan strips through them in the manner of lacing up a shoe or basketball. This is the older technique and it can be performed with even a simple stone or bone drill, as was still being done in remote Pacific islands … Stone tools are probably inadequate for drilling deep holes in the thin edges of boards, and it is therefore not surprising that edge-pegging does not appear in those distant Pacific islands whose inhabitants presumably migrated there without metal.

We can reconstruct PMP *kiRam ‘adze/axe’ and *taRaqi (V) ‘adze, carve’ (Ch. 4, §4.1.1), with reflexes in Oceanic as well as in WMP witnesses, but there are no secure PMP or POc reconstructions for other equipment likely to have been used in boat-building.¹¹ The current consensus among archaeologists is that the introduction of metal in eastern Asia postdated the break-up of PAn and PMP. Although Blust (1976b) has pointed to comparisons between Formosan and WMP forms suggesting that PAn speakers may have had a knowledge of metal tools, it is likely that this knowledge spread after the break-up of PAn and PMP.

2.5. Thwart, cross-seat⇫

PMP		*seŋkar	‘cross-seat in boat, thwart’ (Blust 1972b: *soka)
WMP	Iban	seŋka	‘thwart’
WMP	Malay	seŋkar	‘thwart’
WMP	Tagalog	saŋkal	‘thwart’
POc		*soka(r)	(1) ‘thwart’; (2) ‘bracing timber, crossbeam’ (cf. Ch. 3, §3.4)
Mic	Kiribati	oka	‘rafter of house going from horizontal beam to ridgepole’
Fij	Bauan	soka	‘thwart’
Fij	Bauan	(i)ðoka	‘collar-beam in house’ (i- < POc *i- INS)
Pn	Tongan	hoka	‘upright timber supporting ridgepole’
Pn	Samoan	soʔa	‘collar beam of house’

2.6. Underbody, keel⇫

There is no well-supported PMP reconstruction for ‘dugout underbody of canoe (to which planking is added)’ or for ‘keel of built-up canoe’ although *ba(ŋ)ka(q) (§2.1) is a candidate for the former. Such a term is, however, clearly reconstructable for PCP:

PCP		*takele	‘keel or dugout underbody to which planking is added’
Fij	Wayan	takele	‘keel or dugout underbody to which planking is added’
Fij	Bauan	(i)takele	(i- < POc *i- INS)
Pn	Tongan	takele	‘keel or dugout underbody to which planking is added’
Pn	Samoan	taʔele	‘keel or dugout underbody to which planking is added’
Pn	Tikopia	takere	‘bottom of container, bilge of a canoe hull’
Pn	Rarotongan	takere	‘dugout underbody when washstrakes are present’
Pn	Māori	takere	‘keel or dugout underbody to which planking is added’
Pn	Hawaiian	kaʔele	‘canoe hull; inside bottom of a container’

2.7. Bilge, interior of hull⇫

A PPn term for ‘bilge, interior of hull’ is well supported but cognates have not been noted elsewhere. In some Polynesian languages the same term refers to the interior of any container, such as a cup or basket.

PPn		*liu	‘bilge, interior of hull’
Pn	Tongan	liu	‘bilge’
Pn	Niuean	liu	‘inside of a cup, canoe’
Pn	Samoan	liu	‘bilge water’
Pn	Rennellese	giu	‘bilge, interior of basket, bowl +’
Pn	Tikopia	riu	‘inside of a container; bilge of a canoe’
Pn	Māori	riu	‘bilge, valley, basin’

2.8. Bow and stern⇫

Two well-attested POc locative nouns having the general sense of ‘rear, back part’, and ‘front, front part’, respectively, are reconstructable also with the specialised senses ‘stern’ and ‘bow’. The first of these terms goes back at least to PMP.

POc		*muri-	‘rear, stern’
Adm	Mussau	muri	‘stern’
NNG	Gedaged	muzi-n	‘rear, stern’
MM	Nakanai	mori	‘stern’
MM	Vitu	muri	‘stern’
SES	Ghari	muri-na	‘stern of a boat, hindquarters’
SES	Lau	buri	‘stern, rear’
SES	Sa’a	puri	‘stern’

See also PPn *tau-muri ‘afterdeck’ (§2.10)

POc		*muqa-	‘front, bow of boat’
Adm	Mussau	mua	‘bow’
NNG	Gedaged	muga(ŋ)	‘front part, bow of boat’12
MM	Roviana	(ke)mua	‘bow’
Fij	Bauan	mua	‘tip, point, front’
Fij	Bauan	mua (e liu)	‘prow of boat’
Pn	Tongan	(tau-)muʔa	[V] ‘steer for s.t.’; [N] ‘prow’; [N] ‘aim, goal’
Pn	Samoan	(tau-)mua	‘bows of boats’

See also PPn *tau-mu ʔa ‘foredeck’ (§2.10).

2.9. Carved projecting end-pieces⇫

In many regions of Oceania larger canoes are constructed with a projecting headboard, part of it resting on the underbody abutting the washstrake, the rest sometimes extending several feet beyond. The end of such headboards usually consists of an elaborately carved figurehead, often a human or animal head or figure. There is a corresponding carved end-board at the stern, often standing more or less vertical. Such carved end-pieces are common, for example, in eastern Polynesia, the Solomon Islands, the Massim, the north coast of New Guinea and the Bismarck Archipelago. A POc reconstruction can be made for the name of the headboard:

POc		*ijuŋ	‘projecting headboard of prow, often with ornately carved figurehead’
Adm	Mussau	uru(gila)	‘beak like projection on bow, in shape of bird’
MM	New Georgia language (unspecified) (Haddon 1937: 104, 106)	(toto)ishu	‘figurehead on canoe prow’
SES	Sa’a	ʔisu	‘pieces erected on bow and stern’
SES	Lau	isu	‘prow’
Pn	Tikopia	isu (fana)	‘pointed end of canoe’
Pn	Tongarevan	isu	‘projecting headboard’
Pn	Manihiki	ihu	‘projecting headboard’
Pn	Māori	ihu (waka)	‘carved figurehead on prow’13
Pn	Tahitian	ihu, ihu (vaʔa)	‘projecting headboard on prow’
Pn	Hawaiian	ihu	‘prow’
Pn	Hawaiian	(laʔau) ihu	‘end-piece at head’ (vs Laʔau hope ‘end-piece at stern’)
PWOc		*ŋuju	‘carved prow’
PT	Iduna	mudu	‘carved prow’
MM	Roviana	ŋuzu-ŋuzu	‘figurehead of a tomoko war canoe’

FIXME: From vol2: In vol.1,189 the form POc *ijuŋ ‘projecting headboard of prow’ is erroneously given for *(i,u)cuŋ. This is almost certainly the same etymon as that reconstructed here.

The central meaning of POc *ijuŋ was ‘nose’ and that of *ŋuju ‘beak, snout, mouth’. Reflexes of both commonly have the extended sense ‘projecting point’. The conjunction of Southeast Solomonic and Polynesian evidence suggests *ijuŋ as the more likely POc form carrying the sense ‘projecting headboard’. No widespread cognate set for the stempiece has been noted.

2.10. Bow and stern covers, end-decking⇫

While outrigger canoes with simple dugout hull are suitable for inshore sailing, seaworthy canoes require at least ‘end-decking’, minimally a V-shaped piece covering the bow and stern of the dugout hull. The next comparison points to a PCP term for this piece; PCP *tau may come from POc *taRu ‘cover up’:

PCP		*tau	‘end-decking, end-piece covering bow and stern of canoe hull’
Fij	Rotuman	fau	[V] ‘cover’; [N] ‘cover’
Fij	Bauan	tau	‘triangular decking covering bow and stern’
Pn	Tongan	tau(-muʔa)	‘bow, foredeck’
Pn	Samoan	tau	‘deck’
Pn	Samoan	tau(-mua)	‘bow, foredeck’
Pn	Samoan	tau(-muli)	‘stern, afterdeck’
Pn	Nukuria	tau	‘bow-board, stern-board’
Pn	Tikopia	tau(-muli)	‘stern’
Pn	Tokelauan	tau	‘point at which the keel meets the curve of the bow or stern’
Pn	Tokelauan	tau(-mua)	‘bow, forepart of boat’
Pn	Tokelauan	tau(-muli)	‘stern’
Pn	Māori	tau(-rapa)	‘carved end-piece affixed to stern’
Pn	Māori	tau(-ihu)	‘carved figurehead affixed to prow’

Rick Jackson (pers.comm.) suggests Proto Trukic *tau (?) ‘thwart, cross-seat in canoe’ based on Carolinian, Trukese sō, Puluwat ho ‘thwart’ . The above Polynesian comparisons also indicate PPn *tau-muʔa ‘foredeck’ and *tau-muri ‘afterdeck’.

3. Superstructure⇫

3.1. Platform⇫

On medium-sized and large canoes a platform is sometimes built over the hull and the outrigger (or between the two hulls on double canoes); or two platforms are built, one on the outrigger and one on the starboard side. No PMP term for such a platform is reconstructable on present evidence but there is a strong candidate for a POc term, namely:

PMP		*pa(n)tar	‘shelf; bed-frame of wooden or bamboo laths’ (Blust 1980b: 123)
POc		*patar	‘platform of any kind, including that erected over hull and outrigger framework’ (cf Ch. 3, §3.5)
Adm	Seimat	paca	‘canoe platform’ (Haddon 1937)
PT	Kilivila	pita-patile	‘canoe platform’
PT	Molima	vata-vata	‘platform of any kind’
PT	Molima	vata-vatala	‘canoe platform’
PT	Suau	pata-patari	‘canoe platform of poles stretching across all the booms’ (Haddon 1937)
PT	Motu	pata	‘shelf, table’
MM	Tolai	vatar	‘bamboo or board platform on canoe’
SES	Arosi	hā	‘platform’
SES	’Are’are	hā	‘generic name for stage, shelf, small platform above fireplace’
SES	Tolo	pata	‘raft’
NCV	Raga	bata	‘canoe platform’ (Haddon 1937)
Fij	Bauan	vata	‘platform, shelf’
Pn	Rennellese	hata	‘platform, canoe platform’
Pn	Samoan	fata	‘platform (over canoe +)’

3.2. Cabin, deck hut⇫

A deck hut is often added to large outrigger canoes but no widespread cognate set distinctively denoting such a structure has so far been noted. In a number of languages, reflexes of PMP *balay, POc *pale ‘hut, shed, open-sided house’ (Ch. 3, §3.3) are used for a deck hut, for example, Fijian vale waqa (lit. ‘canoe house’), and valevale ‘hut’ (Hornell 1936:323), and this may also have been the case in POc.

4. Outrigger structure⇫

4.1. Outrigger float, outrigger side of canoe⇫

The meaning ‘outrigger float’ can be attributed to PMP *(c,s)a(R)man. This term is well attested in Oceanic and Central/Eastern Malayo-Polynesian, but rare in WMP. In many WMP languages it has been replaced in this meaning by a reflex of *katiR (§2.1). In Oceanic languages the reflex of POc *saman also has the sense of ‘outrigger side of the canoe’ in contrast to *katae ‘free side of canoe’ (§4.4).

PMP		*(c,s)a(R)man	‘outrigger float’
WMP	Sangir	sahemaŋ	‘outrigger float’
WMP	Tonsea	sareman	‘paddle’
WMP	Chamorro	sakman	‘large canoe from Polynesia or Papua. No outrigger, capable of carrying over 100 people’
WMP	Ambonese Malay	semaŋ	‘wooden strut which supports float’
CMP	Buru	seman	‘wooden strut which supports float’
CMP	Larike	simanu	‘wooden strut which supports float’
CMP	Kola	ama	‘wooden strut which supports float’
CMP	Dobel	yerʔman	‘wooden strut which supports float’
CMP	Ujir (Aru Is.)	arman	‘outrigger float’
POc		*saman	‘outrigger float’
Adm	Wuvulu	tama-ne	‘outrigger float’
Adm	Seimat	cam	‘outrigger float’
Adm	Mussau	samana	‘outrigger float’
NNG	Gedaged	sam	‘outrigger float’
NNG	Manam	rama	‘outrigger’14
MM	Nehan	haman	‘boom’ (vs niag ‘float’)
NCV	Mota	sama	(1) ‘outrigger of a canoe’; (2) ‘tack, lie on the other tack’
NCV	Paamese	a-sem	‘outrigger’
Mic	Kiribati	rama	‘outrigger float’
Mic	Marshallese	tam	‘outrigger float’
Mic	Puluwatese	tām	‘outrigger float’
Mic	Chuukese	tām	‘outrigger float’
Fij	Bauan	ðama	(1) ‘outrigger float’; (2) ‘smaller hull of double canoe’
Pn	Tongan	hama	(1) ‘outrigger’; (2) ‘smaller hull of outrigger’
Pn	Tikopia	ama	‘outrigger including float, always on port side’

Oceanic languages of the Papuan Tip subgroup reflect *sarima rather than *saman (e.g. Motu darima, Suau (Daui) salima, Dobu salime, Molima salima). The *sarima forms possibly continue PMP *(c,s)a(R)man with irregular insertion of i. A similar insertion occurs in the PMP verbal prefix *paR-, continued as POc *paRi-.¹⁵

4.2. Outrigger booms⇫

The outrigger float is connected to the hull by booms, two or three in the case of small canoes but often five or more in larger, ocean-going canoes. A term for ‘outrigger boom’ can be constructed for POc but not for PMP:

POc		*kiajo	‘outrigger boom’
Adm	Mussau	iaro	‘outrigger boom’
Adm	Loniu	kiec	‘outrigger boom’
NNG	Wogeo	kiajo	‘outrigger boom’
NNG	Manam	kiazo	‘outrigger boom’
NNG	Gedaged	aia	‘outrigger boom’
NNG	Yabem	kion	‘outrigger boom’
NNG	Barim	kiada	‘outrigger boom’
PT	Kilivila	kiaro	‘outrigger boom’
PT	Aroma	iaro	‘sticks connecting floats to boom’ (Haddon 1937)
NCV	Nguna	kiato	‘outrigger boom’ (prob. Pn borrowing)
Mic	Kiribati	kiaro	‘outrigger boom’
Mic	Kosraean	kiyɛs	‘outrigger boom’
Pn	Tongan	kiato	‘outrigger boom’
Pn	Rennellese	kiato	‘outrigger boom’
Pn	Samoan	ʔiato	‘outrigger boom’

Fijian i-kaso ‘outrigger boom’ (zero for *i irregular) has sometimes been included in this set, but probably belongs to a separate set, along with such forms as Mota gaso ‘rafter’, Lau ʔato ‘rafter’, which derive from a well-established etymon PMP *kasaw, POc *kaso ‘rafter’ (Ch. 3, §3.4). However, the comparison with Lau ato ‘outrigger boom’ suggests that POc *kaso probably had the general meaning ‘connecting beam or brace’, and as well as denoting crossbeams in a house may have been used as a synonym of *kiajo.

4.3. Connective sticks attaching float⇫

Three main methods of connecting the outrigger float to the booms can be distinguished: direct attachment, in which all of the booms are curved and lashed directly to the outrigger; indirect attachment, in which all of the booms are lashed to sticks that are implanted in or lashed to the float; and mixed attachment, in which some booms are attached directly and others indirectly. A great diversity of methods of indirect attachment is found.

A POc reconstruction for the connective sticks (stanchions, struts) was made by Milke (1968) and is well attested. A single cognate in the north-west New Guinea language, Numfor, allows tentative attribution of this etymon (with indeterminate final vowel) to Proto Eastern Malayo-Polynesian.

PEMP		*patotV	‘connective sticks or stanchions attaching floats to booms’
SHWNG	Numfor	fakok	(k for *t regular)
POc		*patoto	‘connective sticks attaching float’
NNG	Gedaged	patot	‘connective sticks attaching float’
NNG	Tuam	patot	‘connective sticks attaching float’
NNG	Mandok	patot	‘connective sticks attaching float’
NNG	Kilenge	patutu	‘connective sticks attaching float’
PT	Arifama	batoto	‘connective sticks attaching float’
MM	Lihir	hidudu	‘connective sticks attaching float’
NCV	Ambae	batoto	‘connective sticks attaching float’
Fij	Bauan	(i)vatoto	(i- < POc: *i- INS)

4.4. Starboard or hull side of outrigger canoe⇫

A POc term can be reconstructed with the sense ’starboard or free side of the canoe, opposite the outrigger side (*saman). In Central Pacific languages this term was also applied to the larger hull of a double canoe.

POc		*katae, *katea	‘free side of canoe, opposite the outrigger’16
NNG	Gedaged	atai	‘projecting part of the canoe platform, opposite outrigger’
NNG	Barim	kat	‘platform of canoe’
NNG	Manam	eteʔa	‘port side of canoe’
NNG	Tami	kataŋ	‘free side of canoe, opposite the outrigger’
MM	Vitu	kata	‘free side of canoe, opposite the outrigger’
NCV	Mota	gatae	‘free side of canoe, opposite the outrigger’
Mic	Kiribati	katea	‘leeward side of canoe’
Mic	Marshallese	kooca	‘lee side of canoe’
Mic	Ponapean	(pali)kasa	‘lee side of canoe’
Mic	Chuukese	asa	‘lee side of canoe’
Mic	Woleaian	(i)geta	‘lee side of canoe’
Fij	Bauan	katā	(1) ‘starboard side of canoe’; (2) ‘larger hull of double canoe’
Pn	Tongan	katea	‘larger hull, main part of canoe as distinct from the ham’
Pn	Samoan	atea	‘hull side of a canoe, as opposed to the ama’
Pn	Tikopia	katea	‘free side of canoe, opposite the outrigger’
Pn	Hawaiian	ʔakea	‘starboard hull of double canoe’

No non-Oceanic cognates of this set are known.

5. Sail and rigging⇫

The characteristic Oceanic sail types are: (a) a triangular sprit sail with apex downwards and a spar along each of the two sides stemming from the apex, and (b) the (crab) claw sail—a triangular sail with the foreside fixed to a vertical mast and the afterside to a strongly curved sprit, whose lower end is attached to the foot of the mast. In Indonesia a number of other types of sail are found. Both fixed masts and moveable masts or props are used in Indonesia and in the Pacific Islands. The pole is stepped in a socket or on a thwart amidships and can be rotated and raked towards either end by means of running stays.

We give here a further quote from Horridge (1986:56-57):

It is my belief, based on distribution, comparative vocabularies and engineering principles, that the fixed mast with a halyard spread into Malayo-Polynesian communities from the Indian Ocean along with the introduction of the pulley. A large sail of matting could not be raised at sea without a pulley unless it was pushed up by a loose pole. There are no signs that even the last Austronesian-speaking migrants to move out into the Pacific knew about the tripod mast, the tilted rectangular sail, the pulley or the quarter rudder lashed to a rudder support. All of these features also spread only a little way up the mainland coast towards China … The tilted rectangular sail seems to have spread from the Indian Ocean, perhaps even from … Egypt, and to have arrived in Indonesia about 2,000 years ago … It has spread eastwards about as far as the quarter rudder and the pulley.

On some of these points the linguistic evidence appears to be silent. However, several terms for parts of the rig are attributable to POc or to later interstages of Oceanic:

5.1. Sail⇫

A PMP term for ‘sail’ (the object) is continued in all the major subgroups:

PMP		*layaR	‘sail’
WMP	Malay	layar	‘sail’
WMP	Tagalog	layag	‘sail’
WMP	Maranao	laiag	‘sail’
WMP	Tausug	layag	‘sail’
WMP	Cebuano	layag	[N] ‘sail of a boat’; [V] ‘put up the sail for a boat to sail’
CMP	Buru	lā	‘sail’
POc		*layaR	‘sail’
NNG	Gedaged	lai	‘sail’
NNG	Barim	lai	‘sail’
NNG	Yabem	lac	[N] ‘sail of a boat, a canoe’; [v] ‘sail, sail across, sail or steer a canoe towards a goal’
PT	Dobu	naia	‘sail’
PT	Motu	lara	‘sail’
NCV	Namakir	na-la	‘sail’
NCV	Nguna	na-lae	‘sail’
NCV	Paamese	alā	‘sail’
Mic	Marshallese	(wec)lay	‘sail’
Fij	Bauan	laða	‘sail’
Pn	Tongan	lā	‘sail’
Pn	Samoan	lā	‘sail’
Pn	Māori	rā	‘sail’

Traditional Oceanic sails are made of matting, woven from pandanus leaves or other plant fibres. It is therefore not surprising to find that reflexes of POc *qebal ‘pandanus mat’ also have the meaning ‘sail’ in a few languages. Given that POc *layaR ‘sail’ is well established, it is likely that *qebal independently acquired the sense ‘sail’ in more than one daughter language. For the cognate set, see Ch. 4, §3.1.

5.2. Boom and yard of sail⇫

The following form is well attested though its reflexes show a range of meanings:

POc		*jila	‘boom or yard of (triangular) sail’
Adm	Seimat	sil	‘booms of triangular sail’ (Haddon 1937)
Adm	Penchal	cil	‘sheet of sail’
Adm	Lou	(e)sil	‘boom or yard of (triangular) sail’
NNG	Tuam	(na)sila	‘boom or yard of (triangular) sail’
NCV	Mota	(pane) sila	‘boom or yard of (triangular) sail’
NCV	Paamese	a-sil	‘mast; central trunk of tree that grows straight up’
Fij	Bauan	sila	‘sheet of a sail’
Pn	Tongan	sila	[N] ‘yard, for a sail to hang from’; [V] ‘shorten the sheet of a sail’
Pn	Samoan	tila	(1) ‘sprit or spar of sail’ (tila lalo ‘lower sprit’ vs tila tu ‘yard, upper sprit’); (2) ‘mast’
Pn	Pukapukan	tila	‘yard of sail’
Pn	Tikopia	tira	‘mast or spar of sailing canoe’
Pn	Rarotongan	tira	‘mast’
Pn	Māori	tira (tū)	‘yard of sail, upper sprit’

POc or PCP *jila has commonly (e.g. Blust 1976a, Geraghty 1986, Ross 1988) been glossed ‘sheet’, referring to the rope fastened to the lower corner of a sail to hold it and control its angle. The weight of the evidence, however, suggests that the term referred in POc to the booms or poles used to extend and support a triangular sail, denoting either the upper pole (the yard) or the lower pole (the boom). The use of reflexes of *jila to denote a fixed mast is confined to certain parts of Polynesia and this sense probably represents a post-PPn innovation. Most Oceanic craft with triangular sails do not use a fixed mast. In one kind of rigging (spritsail) the longer of the two poles extending the sail serves as the mast, with a mast-shore or stays to secure it, or with the apex of the two poles resting on the deck. In another kind of rigging (lateen), the mast is a separate moveable pole which pivots on a thwart in the dugout or on a socket on the deck and is supported by ropes (stays or sheets) tied to the hull or outrigger and sometimes by a mast-shore. The use of reflexes of *jila for ‘mast’ and for ‘sheet’ in various Oceanic languages can thus be derived from the original functions of the booms as supporting and controlling the sail.

5.3. Mast or props supporting mast⇫

PEOc		*kaiu-tuqu(r)	‘vertical supporting timber, prop supporting rig’
SES	Arosi	auu	(1) ‘centre post of house’; (2) ‘mast’
Mic	Carolinian	ayu	‘mast’
Pn	Tongan	kau-tuʔu	‘yard on a mast’
Pn	Samoan	ʔau-tū	[N] ‘core, centre, main theme’; [V] ‘centre around, revolve around s.t.’
Pn	Māori	(rā) kau-tū	‘mast and sail’17

cf. also:

Adm	Seimat	kau ehu	‘mast’
PT	Motu	au tubua	‘mast, centre post of a house’ (au ‘wood, stick’, tubua ‘upright’)

POc *kaiu-tuqu(r) (evidently a compound of *kaiu ‘wood, stick, pole’ and *tuqu(r) ‘stand; fixed’) possibly referred to any main supporting timber including the prop or mast of a boat. This compound has reflexes in Southeast Solomonic, Polynesian and Nuclear Micronesian and possibly in Motu. It appears to have been already lexicalised in the immediate common ancestor of these groups.

PEOc		*pana	‘mast, boom stepped on foot of mast’
NCV	Mota	pane	‘boom with forked end stepped on the foot of the mast’
NCV	Mota	pane (sila)	‘projecting boom of a sail’
Fij	Bauan	(i)vanā	‘mast’ (i- < POc *i- INS)
Pn	Tongan	fanā	‘mast’
Pn	Samoan	fanā	‘mast’
Pn	Tokelauan	fanā	‘mast’

As the Central Pacific languages in this comparison are all in the Fiji-West Polynesia region and innovations in canoe design are known to have moved freely within this region, the PCP antiquity of *vanā may be questioned. On the other hand, the correspondences Fijian v, Tongan, Samoan f are regular, unlike recent Fijian-Polynesian borrowings, where Fijian v has been borrowed as v. There is a possible cognate in Mota.

5.4. Mast stays⇫

The following cognate set appears to be confined to Central Pacific:

PCP		*tuku	‘running stay supporting sail’
Fij	Bauan	tuku	‘running stay supporting sail’
Pn	Tongan	tuku	‘running stay supporting sail’
Pn	Samoan	tuʔu	‘running stay from foredeck’
Pn	Tokelauan	tuku	‘guy rope of traditional sail, fastened to the outrigger’

These forms may be cognate with PCP *tuku ‘let go, slacken’. Again, the narrow geographic range of this cognate set makes it hard to rule out diffusion.

6. Steering oar⇫

A PMP term for steering oar is well supported. Contemporary languages which retain this term now apply it to rudders, but there is no reason to believe that rudders were in use in PMP times. A verbal use ‘steer (a boat from the stem)’ can also be reconstructed. In many languages this verbal use has now generalised to any kind of steering of a vessel or vehicle.

PMP		*quli(n,ŋ)	[N] ‘steering oar’; [V] ‘steer’
WMP	Cebuano	ulin	[N] ‘stem’; [V] ‘steer (a boat from the stem)’
WMP	Maranao	olin	‘steer (a vessel); manage affairs of another’
WMP	Maranao	olin(aq)	‘steering mechanism’
WMP	Maranao	(paŋ)olin	‘rudder’
WMP	Bajau	uli	‘steer from the stem’
WMP	Sangir	uliŋ	[N] ‘rudder’; [v] ‘steer’
WMP	Wolio	uli	‘rudder’
CMP	Roti	uli	[N] ‘rudder’; [V] ‘steer’
POc		*quliŋ	[N] ‘rudder’; [V] ‘steer’
Adm	Lou	kuli(p)	‘steering oar’
NNG	Tami	gul	‘steering oar’
NNG	Gedaged	ulu(m)	‘rudder, steering’
NNG	Yabem	(ŋa)goliŋ	‘rudder’
PT	Molima	kuliga	[N] ‘steering oar’; [V] ‘steer’
PT	Kilivila	kuliga	‘steering oar’
Fij	Bauan	uli	[N] ‘steering oar, rudder, helm’; [V] ‘steer’
Pn	Tongan	ʔuli	‘steer’ (fohe ʔuli ‘rudder, steering oar’)
Pn	Samoan	uli	‘steer’ (faʔa -uli ‘steer, helmsman’; foe uli ‘rudder’)

7. Accessories⇫

Under this heading fall objects used or carried on board, such as cargo, anchor, paddles, punting pole, bailer and Triton shell for use as a trumpet.

7.1. Cargo⇫

A single base can be reconstructed, used both as a verb ‘(boat +) be loaded, carry a cargo’ and as a noun ‘cargo, load’. However, it occurs in the form of a doublet, with and without initial *l-.

PMP		*lujan, *ujan	[V] ‘load (a vessel)’; [N] ‘load, cargo’ (Blust 1986)
WMP	Tagalog	lulan	‘load, cargo, capacity of a vessel or vehicle’
WMP	Makassarese	luraŋ	[V] ‘load’
POc		*ujan, *ujan-i-, *lujan	[v] ‘load (a boat)’; [N] ‘cargo, freight’
Adm	Nauna	us	‘load as cargo in a boat’
Adm	Seimat	uxan-i	‘load as cargo in a boat’
PT	Dobu	usana	‘load a canoe’
PT	Sinaugoro	ɣura(udi)	‘load a canoe’
PT	Motu	uda-uda	‘load a canoe’
SES	Gela	luda	‘load a vessel’
SES	Bugotu	luja	‘load a vessel’ (j for *d unexplained)
SES	Lau	luda	‘load a vessel’
SES	Kwaio	luda	‘load a vessel’
Mic	Puluwatese	wutan	[v] ‘loaded’; [N] ‘cargo, load’
Fij	Bauan	usa	‘carry a cargo’
Pn	Tongan	uta	[v] ‘carry a cargo’; [N] ‘cargo, freight’
Pn	Samoan	uta	[V] ‘carry a cargo’; [N] ‘cargo, freight’

7.2. Paddles, paddling and punting⇫

There are two well-established PMP terms to do with paddling, *be(R)(c,s)ay and *pa-luja:

PMP		*be(R)(c,s)ay	[N] ‘(canoe) paddle’; [V] ‘paddle’
WMP	Aklanon	bugsay	[N] ‘paddle’
WMP	Cebuano	bugsay	‘paddle or row a boat’
WMP	Ngaju Dayak	besei	[v] ‘paddle’
WMP	Buginese	wise	[N] ‘paddle’
WMP	Wolio	bose	[N] ‘paddle’
CMP	Taliabo	bose	[N] ‘paddle’
CMP	Buru	sai	[V] ‘paddle’
CMP	Buru	sahi-n	[N] ‘paddle’
POc		*pose	[N] ‘(canoe) paddle’; [V] ‘paddle’
NNG	Manam	ore	[N] ‘paddle’
PT	Motu	hode	[N] ‘paddle’
PT	Suau	wose	[N] ‘paddle’
MM	Roviana	vose	[N] ‘paddle’
SES	Gela	vohe	[N] ‘paddle’
SES	Lau	fote	[N] ‘paddle’
SES	Sa’a	hote	‘paddle’
NCV	Mota	wose	[N] ‘paddle’
Fij	Bauan	voðe	[V] ‘paddle’
Fij	Bauan	(i)voðe	[N] ‘paddle’ (i- < POc *i- INS)
Pn	Tongan	fohe	[N] ‘paddle’
Pn	Samoan	foe	[N] ‘paddle’
Pn	Hawaiian	hoe	‘paddle’

PMP		*pa-luja	[V] ‘paddle’; [N] ‘paddle’
WMP	Nias	aluxa	[N] ‘paddle’
WMP	Sichule	luga	[N] ‘paddle’
WMP	Sichule	feluxa	[V] ‘paddle’
WMP	Toba Batak	luga	[v] ‘row’
WMP	Isneg	piloxa	‘oars and paddles of a canoe’
POc		*paluca	[V] ‘paddle’; [N] ‘paddle’
Adm	Likum	heluh	‘paddle’
MM	Roviana	valusa	‘bonito fishing’
SES	Arosi	haruta	[V] ‘paddle’
SES	Lau	falita, faluta	[V] ‘paddle’; [N] ‘canoe’
NCV	Lonwolwol	faloh	[V] ‘paddle’; [N] ‘paddle’
NCV	Paamese	valis, valus	‘row, paddle’
Mic	Kiribati	arina	‘paddle’ (metathesis)
Mic	Mokilese	patil	‘paddle’ (metathesis)
Mic	Ponapean	patil	‘paddle’ (metathesis)
SV	Anejom̃	aheleθ	‘paddle’

It is not clear whether PMP *be(R)(c,s)ay and *paluja differed in meaning. As the Sichule comparisons show, PMP *paluja probably derives from a root *luja denoting a paddle, with the verb ‘paddle’ derived by adding the causative prefix *pa-. *paluja is now widely reflected, meaning both ‘a paddle’ and ‘to paddle’, but it may be that its use as a noun developed independently in various languages.

A third form attributable to PEOc, *sua(C), also has some reflexes glossed ‘to paddle’. However, the meanings associated with its putative reflexes are quite varied. These meanings include (i) ‘scull, in which a standing person holds the oar vertically’, (ii) ‘punt or pole a boat in shallow water’, (iii) ‘to paddle’, (iv) ‘a paddle’ and (v) ‘to steer’. The range of meanings suggests an original reference to a standing person using an oar or pole to propel or to steer a boat. (cf. also PCP *sua ‘tack’, §9.2.)

PEOc		*sua(C)	[V] ‘scull, row with oar held vertically’
SES	’Are’are	sua(hi)	‘paddle against the wind’
SES	Lau	sua(la)	‘punt, push a canoe with a pole’
SES	Lau	sua(li)	‘push against’
NCV	Mota	sua	[v] ‘paddle, make a canoe voyage’
NCV	Mota	sua(va)	[N] ‘paddling, canoe voyage’
NCV	Raga	hua	[V] ‘paddle’
NCV	Tangoa	sua	[V] ‘paddle’
Fij	Rotuman	sua	‘scull, paddle, oar’
Fij	Bauan	sua	‘scull, row, put an oar in two transversal poles lashed across the crossbeams near the deck of a canoe to help in rowing’
Pn	Rennellese	sua	‘ceremonial paddle with wide blade; go to a ship at anchor (? by paddling)’
Pn	Māori	hua	‘steer, paddle’

cf. also:

Fij	Wayan	due	‘scull, propel a boat by putting an oar or pole vertically into the water and twisting it about’
Fij	Wayan	i-due	[N] ‘paddle, oar’
Pn	Samoan	sua(ti)	‘balance pole of canoe’
Pn	Tikopia	sua(ti)	‘balance pole of sailing canoe’
Pn	West Uvea	hua(ge)	‘balance pole of sailing canoe’

7.3. Punting pole⇫

PMP *teken ‘pole, staff’ was evidently continued in POc both as a noun *tokon and as a transitive verb *tokon-i- (v) ‘punt or pole (a boat)’:

POc		*tokon	‘staff, punting pole’
POc		*tokon-i	‘punt or pole (a boat)’
NNG	Gedaged	tok	‘pole, stick, staff’
NNG	Tuam	to	‘punting pole’
NNG	Barim	to	‘punting pole’
PT	Motu	do, doa	[V] ‘pole (a canoe)’
PT	Motu	to	‘stay of house, prop of fence, brace’
SES	Ghari	togon-i	‘stick for stirring stones’
Fij	Wayan	(i)toko	‘staff, punting pole’ (i- < POc *i- INS)
Fij	Wayan	tokon-i	‘punt a boat’
Fij	Bauan	(i)toko	‘staff, punting pole’ (i- < POc *i- INS)
Fij	Bauan	tokon(a)	‘punt a boat’
Pn	Tongan	toko	‘punting pole, to punt’
Pn	Samoan	toʔo	‘punting pole, to punt’
Pn	Rennellese	toko	‘punting pole, to punt’
Pn	Tuvalu	toko	‘punting pole, to punt’

cf. also:

WMP	Sangir	tekiŋ	‘staff’
WMP	Malay	teken	‘lean on, pressure’
WMP	Javanese	teken	‘cane, walking stick’

7.4. Bailer, bailing⇫

Three terms to do with bailing water from a vessel can be reconstructed at the PMP level, all of which are continued in PWMP, POc and PCP.

PMP		*limas	‘bailer’
WMP	Cebuano	limas	(1) ‘bailer’; (2) ‘bilge water’; (3) ‘bail water out’
WMP	Tagalog	limas	‘bailer’
WMP	Hova	dima	‘bailer’
CMP	Boano	limate	‘bailer’
POc		*lima(s), *nima(s)	‘bailer’
NNG	Kilenge	na-lima	‘bail’
NNG	Mangseng	lima	‘bail’
Mic	Kiribati	a-nima	‘bailer’
Mic	Puluwatese	niim	‘bailer’
Fij	Bauan	(i)nima	‘bailer’ (i- < POc *i- INS)
Fij	Bauan	nimat(a)	‘bail it out (canoe)’

The following PMP reconstruction is given by Blust (1978b: 94). Although ‘scoop or ladle out’ was probably the basic meaning, a number of Oceanic languages use the reflex of *asu to denote a bailer.

PMP		*aŋsu	‘scoop or bail out’
WMP	Javanese	aŋsu	‘draw water’
CMP	Buru	asu-k	‘scoop, dip or bail (water) with a scooper’
POc		*asu	[V] ‘scoop or ladle out’; [N] ‘ladle, bailer’
Adm	Wuvulu	atu	‘bailer, spoon, ladle’
Adm	Aua	atu	‘bailer, spoon, ladle’
NNG	Gedaged	yasi	‘scoop or ladle out’
NCV	Nguna	na-asu	‘canoe bailer’
NCV	Nguna	(m)asi	‘bail’
NCV	Raga	ahu(a)	‘bail water, scoop up’
Fij	Bauan	yaðu	[V] ‘ladle, scoop’
Pn	Tongan	ohu	‘ladle or bail out liquid’
Pn	Samoan	asu	[v] ‘scoop, ladle or bail out liquid’
Pn	Māori	ahu	‘bailer’

The next comparison shows a PPn term for ‘bailer’ and ‘bail’, derived from a PMP form which may have had a more restricted use.

PPn		*tatā	[V] ‘bail out’; [N] ‘bailer’
Pn	Tongan	tatā	‘bail out’
Pn	Samoan	tatā	‘bail, scoop; anything used as a bailer’
Pn	Tokelauan	tatā	‘bail, scoop; anything used as a bailer’
Pn	Rarotongan	tatā	‘bail, scoop; anything used as a bailer’
Pn	Māori	tatā	‘bail, scoop; anything used as a bailer’

Compare WMP: Cebuano tata ‘empty a container by turning it upside down’. This suggests PMP *tata ‘empty water from something’.

7.5. Portable fireplace⇫

Seagoing canoes in many places carry a heap of sand, or an old clay pot or wooden basin filled with sand or earth, for use as a fireplace. A reconstruction for such an apparatus can be made only at PCP level:

PCP		*tā-dravu	‘portable fireplace or oven’
Fij	Bauan	tā-dravu	‘portable fireplace or oven’
Pn	Tongan	tā-lafu	‘portable fireplace or oven’
Pn	Rennellese	tā-ŋahu	[N] ‘fire’; [v] ‘build a fire for illumination’
Pn	East Uvean	ta-lafu	‘cooking box filled with earth and small stones, formerly used in large sailing canoes’
Pn	Tokelauan	tā-lefu-lefu	‘ashtray’

cf. also:

SES	Arosi	dohu	‘white ashes, a fireplace’
NCV	Mota	tarowo	‘white ashes’

The PCP form can be analysed into two elements: the second is clearly *dravu ‘fireplace, hearth’, which reflects POc *rapu(R) ‘hearth, fireplace; ashes’ (Ch. 6, §2.4). The origin of *tā is less clear.

8. Launching and beaching, anchoring, sheltering vessels⇫

Outrigger canoes are normally launched from beaches and hauled ashore rather than anchored when not in use. Anchors are not carried on small canoes (which may be left to drift or tethered to the reef while fishing) but stone anchors were commonly carried aboard larger vessels.

8.1. Rollers or skids⇫

A PMP term for canoe rollers or skids is well attested, with reflexes in Philippine, Maluku and Oceanic languages.

PMP		*laŋen	‘rollers, skids or blocks to move or raise a boat’
WMP	Maranao	laŋen	‘rollers’
CMP	Asilulu	lane-t	‘rollers’
CMP	Buru	laŋe	‘rollers’
POc		*laŋon	[N] ‘rollers’; [V] ‘place rollers under a canoe’
SES	Arosi	(i)raŋo	[N] ‘roller for canoe’; [V] ‘place rollers under a canoe’ (i- < POc *i- INS)
Mic	Kiribati	naŋo	‘rollers for canoe’
Mic	Marshallese	lᶭaŋ	‘rollers for canoe’
Mic	Woleaian	laŋo	‘rollers for canoe’
Fij	Bauan	laŋo	‘place rollers (for canoe +)’
Fij	Bauan	laŋoni ni waga	‘canoe rollers’
Pn	Tongan	laŋo	‘supporting block or beam’
Pn	Samoan	laŋo	[N] ‘support, prop, pillow, bolster’
Pn	Rennellese	gaŋo	‘coaster (butt ends of coconut fronds or sticks) for dragging a canoe over the beach’
Pn	Tikopia	raŋo	[N] ‘canoe skid or block’; [V] ‘support (canoe +)’

8.2. Anchoring⇫

There are two fairly well-supported PMP reconstructions to do with anchoring. In the following comparison the WMP terms refer to an anchor while the Oceanic terms support a verbal reconstruction ‘be anchored or moored’.

PMP		*sauq	[N] ‘anchor’; [V] ‘be anchored’
WMP	Tagalog	sawoq	‘anchor’
WMP	Toba Batak	sawo	‘anchor’
WMP	Malay	sauh	‘anchor’
WMP	Ngaju Dayak	sauh	‘anchor’
CMP	Buru	sau	‘anchor (possibly borrowed from Malay)’
SHWNG	Numfor	sau	‘anchorage’
POc		*jau(q)	‘be anchored or moored, come to anchor or rest’
SES	Arosi	dau	‘(canoe) come to rest’
SES	Lau	dau	‘come to anchor; alight, be stationary, at rest’
SES	Sa’a	deu	‘(canoe) settle, be stationary’
Pn	Tongan	tau	‘anchor or moor a boat, park a car’
Pn	Samoan	tau	‘moor, anchor’
Pn	Rennellese	tau	‘come to land’
Pn	Tikopia	tau	‘(vessel) fetch up, come in to land’
Pn	Māori	tau	‘come to anchor, ride at anchor, lie to’

8.3. Passage or channel, landing place⇫

The PMP and POc term *sawaŋ or *sawaq appears to have been the conventional name for a channel for boats to pass through or to land, or an area of calm water giving safe anchorage. Blust (1983–84a: 113) reconstructs *sawaq ‘channel’.

PMP		*sawa(ŋ,q)	‘opening used by boats to pass through, channel or strait, safe passage or anchorage’
WMP	Chamorro	sagwaʔ	‘channel, inlet of water, narrow passage in reef’
WMP	Malay	sawaŋ	‘breakwater’
POc		*sawaŋ	‘channel in reef giving passage to boats, landing place, anchorage’
Adm	Lou	(mara)sa	‘channel, passage between islands’
Adm	Titan	(mata)ca	‘channel, passage between islands’
Adm	Wuvulu	tawa	‘channel, passage between islands’
NNG	Yabem	sawa	‘space, empty area’
PT	Motu	dava	‘lagoon in atoll, water in chasm or ditch’
MM	Roviana	savaŋa	‘strait between two islands’
SES	’Are’are	tawa	‘channel in reef, landing place’
SES	Sa’a	tawa	‘landing place’
Mic	Kiribati	rawa	‘channel in reef’
Mic	Marshallese	tew	‘channel in reef’
Mic	Ponapean	dāw	‘channel in reef’
Mic	Chuukese	tāw	‘channel in reef’
Mic	Woleaian	tāwa	‘channel in reef’
Fij	Rotuman	sava	‘passage or opening in a coral reef’
Fij	Wayan	(mata)ðawa	‘beach’
Fij	Bauan	(mata)sawa	‘landing place’
Pn	Niuean	ava	‘harbour, opening in reef, channel’
Pn	Samoan	ava	‘channel, passage in reef; anchorage’
Pn	East Futunan	ava	‘anchorage’
Pn	Māori	awa	‘channel, landing place for canoes’

cf. also:

Pn	Tongan	ava	‘hole, aperture’
Pn	Tongan	vaha	‘space between, strait, channel’

There is clear evidence for a POc compound consisting of *mata ‘opening, entrance; focal point’ plus *sawaŋ, with the sense ‘landing place for boats’. The Southeast Solomonic and Kiribatese forms suggest the POc compound had the form *mata-ni-sawaŋ, or *mata-qi-sawaŋ, with one of the two POc genitive particles *ni or *qi (Hooper 1985) linking the two nouns. However, the Admiralty Islands and Fijian forms do not reflect the genitive.

POc		*mata-sawaŋ, *mata-[ni,qi]-sawaŋ	‘landing place, channel in fringing reef giving passage to boats’
Adm	Lou	mara-sa	‘channel, landing place’
Adm	Titan	mata-ca	‘channel, landing place’
SES	Arosi	maeta-wa	‘boat landing, landing place where the sea is calm’
SES	Kwaio	maa-li-takwa	‘landing place, salt water’ (the Kwaio are an inland people)
SES	Sa’a	maa-li-tawa	‘opening in shore reef, land place’
SES	’Are’are	ma-ri-tawa	‘landing place, channel’
Mic	Kiribati	mata n rawa-rawa	‘channel, gap in reef’
Fij	Wayan	mata-ðawa	‘beach’
Fij	Bauan	mata-sawa	‘landing place’

8.4. Boatshed⇫

While boatsheds, for building and sheltering boats, are common in the Malayo-Polynesian speaking region, no really widespread cognate set for such a building has been noted outside the Central Pacific group. Some Oceanic languages use a simple term reflecting PMP *balay ‘house or building, probably with open sides’ as Lou (Admiralties) pal ‘boatshed’. Others use a compound nominal whose components, as in English, are the words for ‘boat’ and ‘house’ . It may be that speakers of PMP used such a compound but it is hard to rule out the possibility of independent parallel developments in the daughter languages. However, there is good evidence for a distinctive PCP term for boatshed, evidently incorporating the term for ‘make a sea voyage’ (see POc *palau(r), §9.2):

PCP		*(a)valau	‘boatshed’
Fij	Wayan	volau	‘boatshed’
Fij	Bauan	volau	‘boatshed’
Pn	Tongan	ala folau	‘boatshed’ (first -l- unexpected; cf. ala folau ‘fit to go to sea’)
Pn	Niuean	afolau	‘temporary shelter’
Pn	Samoan	āfolau	‘long house, used for e.g. receiving guests’
Pn	Tikopia	aforau	‘canoe shed’
Pn	Tikopia	(mata) aforau	‘canoe yard for sacred canoes’
Pn	Māori	farau	(1) ‘temporary shed or booth’; (2) ‘canoe shed’

9. Seafaring terms⇫

A number of terms can be reconstructed at PMP or lower levels for concepts to do with going to sea and navigation. The following is not an exhaustive list.

9.1. Embark, ride⇫

PMP		*sakay	‘embark, be aboard, ride (on a vessel +)’
WMP	Cebuano	sakay	‘travel by sea, embark, ride on (a boat +)’
WMP	Ilokano	sakay	‘ride in a boat’
WMP	Bikol	(mag)-sakay	‘ride in a boat’
WMP	Sangir	sakaeŋ	‘boat’ (from PMP *sakay-an lit. ‘thing to ride on’)
CMP	Asilulu	saka	‘ascend, climb’
CMP	Selaru	sai	‘climb, go up, ride, mount’
POc		*sake	‘embark, ride on a canoe’ (prefix to numerals denoting number of crew carried by a canoe)
SES	Lau	tae	‘embark’
SES	’Are’are	taʔe	‘ride, embark’ (prefix in taʔe taʔai ‘one-man canoe’, taʔe rua ‘two-man canoe +’)
NCV	Mota	sage	‘prefix with numerals when men on board a canoe are numbered’
Mic	Mokilese	tak	‘ride on (a vessel +)’
Mic	Ponapean	take	‘ride’
Mic	Carolinian	tāta	‘ride on s.t.’
Mic	Woleaian	tagē(a)	‘ride on it, sail in it’
Mic	Woleaian	(tet)tag	‘ride’
Fij	Wayan	ðake	‘embark, go aboard’

This was the PMP/POc term for ‘ascend, climb, mount’.

9.2. Voyaging⇫

In the following set the Central Pacific forms clearly refer to long-distance sailing. There is some question whether the WMP forms are cognate with the Oceanic.

PMP		*pa-laSud	‘go down to the sea or coast’
WMP	Cebuano	palawud	‘go to sea’
WMP	Ilokano	palaud	‘go to the west, go down to the coast’
WMP	Tukang Besi	henaʔu	‘descend, go seawards, go west’
POc		*palau(r)	‘go to sea, make a sea voyage’
SES	Tolo	vola-volau	‘run, race’
NCV	Raga	walau	‘guide, steer, direct’
NCV	Mota	wala-walau	‘paddle all together’
NCV	Nguna	wo-wolau	‘steer canoe’
Mic	Kiribati	borau, bo-borau	‘travel by sea’ (prob. borrowed from a Pn source)
Fij	Bauan	volau	[v] ‘make a sea voyage’; [N] ‘boat house’
Pn	Tongan	folau	‘voyage, travel by sea’
Pn	Tongan	folau(ʔaŋa)	‘boat in which one voyages’ (-ʔaŋa < NOM)
Pn	Tongan	folau(ʔia)	‘be constantly visited by ships’
Pn	Rennellese	hogau	(1) ‘ocean voyage’; (2) ‘canoe making an ocean voyage’
Pn	Samoan	folau	‘travel by sea, make a voyage; depart, sail’
Pn	Samoan	fōlau(ŋa)	‘voyage’ (-ŋa < NOM)
Pn	Samoan	folau(vaʔa)	‘sailor’
Pn	Tikopia	forau	‘voyage overseas, travel abroad’
Pn	Māori	farau	(1) ‘travel, particularly by water’; (2) ‘company of travellers’

cf. also:

SES	Gela	vinau	‘go by sea’
Pn	Nukuoro	balia	‘expert navigator’

Lawrence Reid (pers.comm.) suggests that the Malayo-Polynesian forms derive from a PAn phrase whose constituents were *pa ‘go, towards’ and *iaSud ‘sea, ocean’, giving the meaning ‘seawards (from inland)’ contrasting with *daya ‘landwards, towards the interior (from the sea or coast)’. In some languages, reflexes of *pa-laSud are opposed to a phrase or parallel structure meaning ‘go inland, go to the mountains’. POc continued *iaSud as *lau(r) in its original sense (e.g. Mota lau ‘seawards, coastwards’) but evidently reanalysed the sequence *pa-iau(r) as a single morpheme.

Blust (1978a:216) offers the following:

PEMP		*ta(d,R)i	‘steer a course (in navigating)’
SHWNG	Numfor	kar	‘row (while facing one’s destination)’
POc		*taRi	‘steer a course’
PT	Motu	tari	[N] ‘rudder, steer oar’; [v] ‘steer a canoe’

The sound correspondences are regular, but one would like further cognates to strengthen the comparison. Lau tari ‘steer, keep straight on’ and ‘Are’are tari/roro ’steer a canoe (the steersman drawing the paddle towards himself with big strokes) show a superficial likeness but are not demonstrably cognate, because t in Malaitan languages derives from POc *s. Rick Jackson (pers.comm.) reconstructs Proto Western Micronesian *taraki ’sail, travel by sea’ with reflexes in Trukese, Ponapean and Marshallese but again the resemblance to PEMP *ta(d,R)i seems to be superficial.

PMP		*biluk	‘tack, sail to windward’
WMP	Javanese	biluk	‘tack’
WMP	Malay	belok	‘tack’
WMP	Ngaju Dayak	biluk	‘tack’
WMP	Tausug	biluk	‘tack’
POc		*pilu(k)	‘tack’
SES	Sa’a	hilu-hilu	‘zigzag’
PCP		*sua	‘tack’
Fij	Rotuman	sua	‘tack about, change tack’
Pn	Tongan	hua	‘(boat) change from one tack to another’

9.3. Expert sailor or fisherman⇫

The following reconstruction consists of a compound with *tau ‘person, expert, owner’ as the first element and *tasi(k) ‘sea’ as the second.

PEOc		*tau-tasik	‘expert fisherman or sailor, mariner’
SV	Sye	ntoɣ	‘sailor’
SV	Southwest Tanna	tahik	‘sailor’
Mic	Chuukese	sowu-set	‘master fisherman’
Mic	Carolinian	sou-lē-set	‘skilled fisherman’
Pn	Tongan	toutai	‘mariner, sailor; go fishing’
Pn	Tongan	toutai(ʔi)	‘steer or pilot (a fishing boat)’
Pn	Rennellese	tautai	‘go fishing in the sea’
Pn	Samoan	tautai	‘master fisherman, captain of a boat or ship’
Pn	Tikopia	tautai	‘skilled seaman, expert fisherman’
Pn	Tokelauan	tautai	‘master fisherman, skipper or captain of a boat’

cf. also:

MM

Tolai

te-na ta

‘sailor’ (lit. ‘one belonging to the sea’)

9.4. Boat owner or captain⇫

In the following set the first element is again *tau ‘person, owner’. The agreement between Polynesian and Papuan Tip languages may be the result of parallel development. However, in both groups the formation of compounds with *tau is no longer productive, so the chances are that the compounds in question are quite old.

POc		*tau (ni) waga	‘owner of a boat’
PT	Molima	to ni waga	‘canoe owner or captain’
PT	Muyuw	ta-ga-ni-wag	‘canoe owner or captain’
Fij	Bauan	tau-kei ni waga	‘boat owner’
Pn	Rennellese	tau baka	‘canoe owner, act as a captain’
Pn	Tikopia	tau vaka	‘canoe owner’

10. Conclusions⇫

Upwards of twenty terms to do with watercraft and seafaring can be reconstructed for PMP. The comparative lexical evidence allows the following inferences. PMP speakers were familiar with outrigger sailing canoes. Various clues indicate that craft could be quite large. Hulls could be built up with planking. Skids or rollers were used to move vessels on land. A steering paddle was used. Large canoes probably carried anchors. Cargo and paddles, punting poles and bailers were carried on board.

Virtually all the reconstructable PMP canoe and seafaring terms were continued in POc and PCP. In addition, around ten terms can be attributed to POc and PCP that have not so far been reconstructed for PMP. While these figures are impressive, they probably represent only a small proportion of the total body of terms for canoes and seafaring used by the speech communities in question. In contemporary societies where large sailing canoes remain in use, it is usual for a language to have over one hundred terms for parts of the vessel alone.

Some of the POc and PCP terms which do not have known sources in PMP may be formal innovations. It seems likely, however, that in most cases the innovative forms replaced functionally equivalent PMP terms whose forms are not recoverable on present evidence. This conclusion rests on logical grounds: the presence of certain terms strongly implies that other functionally connected terms also existed. Thus, while we cannot reconstruct with certainty a PMP name for ‘outrigger sailing canoe’, we can reconstruct PMP terms for ‘outrigger float’, ‘outrigger boom’, ‘sail’ and other relevant parts and equipment. It can therefore safely be inferred that PMP speakers were familiar with outrigger sailing canoes. Furthermore, it seems that all Oceanic languages have a general name for outrigger sailing canoes (as well as, usually, a variety of terms for specific types). It would be very surprising if such a name did not exist in PMP, even if it was not cognate with the term *waŋka/*waga that has been reconstructed for PEMP and POc. The same reasoning applies, say, to ‘canoe platform’, though with slightly less force. A term for such a platform, placed amidships over hull and outrigger booms, is well supported for POc (*patar) but not for PMP. However, a PMP term for ‘load a vessel’ and ‘cargo, load carried by a vessel’ is reconstructable and it is therefore likely that PMP speakers built platforms on their larger, cargo-carrying outrigger canoes. Although a PMP term for ‘strake, plank (of canoe +)’ is recoverable, implying familiarity with built-up canoes, no term for the end-decking of a built-up canoe-minimally, triangular end-pieces abutting the topstrake fore and aft-is attributable to PMP. But as end-decking of some sort is a functional necessity in built-up seagoing outrigger canoes, it is unlikely that PMP seagoing canoes would have lacked these essential parts.

Further work will undoubtedly add to the body of relevant lexical reconstructions. However, it is unlikely to fill all the gaps. Often the lexical evidence is not fine-grained enough to allow us to recover certain details of vessel design. For example, the lexical reconstructions for the sailing rig do not indicate whether the sails used by PMP speakers were triangular, crab claw or rectangular or whether their vessels had fixed or moveable masts (or both). Nor do the lexical reconstructions tell us whether PMP speakers made canoes with single or double outriggers (or both). There is another reason why it is sometimes difficult to trace changes in technological details through comparative linguistic evidence: whereas we can safely infer knowledge of certain things by the presence of terms for them, we cannot so readily infer ignorance from the lack of reconstructable terms. Consider the double-hulled canoe. A term for double canoe can be attributed to PPn but not to POc. Can we therefore conclude that such a craft was unknown to POc speakers? Certainly not. Inability to reconstruct a term for a certain referent in Proto X does not prove that the referent itself, and a term for it, were unknown to speakers of Proto X. The term may have been lost in all or most daughter languages and its former existence obscured.

The moral we might draw at this point is a rather obvious one: that for doing culture history several disciplines are, ultimately, better than one. Linguistics and comparative technology need each other, just as both need archaeology and comparative ethnography, to corroborate each other’s evidence on certain questions and to provide testimony on points where the other disciplines are mute. It is, of course, important that each field of study contribute its own independent witness before synthesis is attempted. The challenge then becomes how to combine judiciously the evidence from different disciplines.

Notes⇫