Volume 105 1996 > Volume 105, No. 4 > From sau 'ariki to Hawaiki, by Melenaite Taumoefolau, p 385-410
FROM *SAU 'ARIKI TO HAWAIKI
The Proto-Polynesian Lexicon Project, POLLEX (Biggs and Clark 1995), records the highest-level proto-form of the Maori word hawaiki as Proto-Nuclear Polynesian (PNP) *sawaiki ‘traditional homeland’. 1 No corresponding form is recorded for Proto Tongic (PTO) or Proto Polynesian (PPN). I want to modify this picture by suggesting that the word *sawaiki 2 was derived from a PTO loanword in PNP and that, therefore, the word hawaiki (with all its place-name cognates) is ultimately Tongic in origin.
I will first discuss the formal changes undergone by the original compound word from its bimorphemic PPN form to its monomorphemic modern reflexes including hawaiki. Second, I will describe the connection in meaning between hawaiki and its bimorphemic proto-forms to establish their historical relationship, paying particular attention to the way in which the proto-meanings have become modified until their present state. Finally, I will consider the question of whether a “homeland” for the Eastern Polynesians can be inferred from the data.
THE EVOLUTION OF THE ORIGINAL FORM
Percy Smith speculated on the etymology of Hawaiki:
It may be that an “r” has been deleted, and the word might have been Hawa-riki, which means “little Hawa.” But no Polynesian would, if this had been the case, use the form Hawaiki-nui (the great little Hawa). It seems to me more probable that the name may have been originally, Hawa-ariki or Hawa-the-regal, from ariki, eiki, aka-iki, etc., a high chief, king, firstborn, etc.…(1921:47).
Like Smith and many after him, 3 I had always assumed that the word was a compound, but it was not until my attention was drawn by the Tongan rendering of Hawai'ian Hawai'i, apparently a cognate of hawaiki, that I started thinking seriously about the possible origins of hawaiki. I remember thinking: what if the Tongan form Hauai'i 4 is closer to the proto-form, whatever it was, than Hawai'ian Hawai'i? This possibility would not be entirely far-fetched, given what we now know of the subgrouping of Polynesian languages as well as the broad order of settlement of the Pacific. 5
I considered that, although w has been reconstructed as a PPN phoneme, 6 it seems to occur sometimes in some modern Polynesian languages as an - 386 allophone of u. This use of w can be found in Tongan and Samoan, where sequences of u followed by one of a, e, i or o, provided that the u is unstressed while the following vowel is stressed, are commonly pronounced with the semivowel w. Thus, although Tongan words such as uasi ‘watch’, Uepi a personal name, Uili ‘Willy’, and uoi exclamation of surprise are spelt with an initial u, they are normally pronounced as [wasi], [wepi], [wili] and [woi], respectively. Pronounced thus, each word becomes disyllabic with the stress on the penultimate syllable which consists of the semivowel providing the onset of the syllable and the following vowel providing the nucleus. But since careful pronunciation of such words would bring out the u as distinct from the following vowel, resulting in trisyllabic forms, the semivowel w is not considered to be phonemic in Tongan or Samoan. The w-pronunciation would thus be a phonetic variation on the u-pronunciation.
It sometimes happens that in some words in which u is pronounced as w, further change takes place in which the consonant v replaces u. 7 Thus, the name Uili (from English Willy), commonly pronounced [wili], has a variant Vili, and uasi ‘watch’ has an archaic form vasi attested in the personal name Vasi Taiāmoni ‘diamond watch’. What doublets such as Vili and Uili [wili] show is that sometimes w and v arise out of u, and it is the u that is the original phoneme in such words. The possibility exists, then, that the w in Hawai'i and the v in the cognate Havai'i arose as allophones of the proto-vowel u. 8
By analogy with Tongan Hauai'i, the Tongan rendering of Hawaiki, Hauaiki, might have been closer to the proto-form than Maori Hawaiki. If this was the case, assuming that the proto-form was a compound, where would the morpheme boundary be likely to be in Hauaiki? If the stress system of Tongan reflects the PPN stress system in any important way— and I think there are indications that it may do—any five-syllable word (based on five vowels) such as Hauaiki would be divided into either a 3-2 or a 2-3 stress-group sequence, with the stress in each group on the penultimate vowel. 9 This means the word Hauaiki would be pronounced in one of two ways: either as [Haôa-íki] or as [Háu-aíki]. But which one?
Although, to my knowledge, there has not been any detailed study of the conditions under which the vowel u would be likely to “turn into” w, one important condition would seem to be that the u should be unstressed. 10 This means that the u in Hauaiki (and Hauai'i) would have been unstressed in order to later “turn into” w and v. On this basis, I reject the pronunciation with the stressed u, [Haôa-íki], and opt for the remaining [Háu-aíki]. Since in Tongan, stress-group boundary generally (but not always) corresponds to morpheme boundary in compound words, it seems likely that the two elements of the compound word were hau and aiki rather than haua and iki.- 387
In the matter of reconstructing place names, Koskinen (1973) advises that place-names should be reduced to some form of proto-form to make it easier to look for cognates. It is only a small step from here to work out some hypothetical proto-forms. *Sau and * 'ariki would be likely candidates for PPN.
It turns out that the PPN compound word *sau 'ariki is indeed reconstructible from a number of modern reflexes: Samoan sauali'i, Tokelauan haualiki, Tahitian hauari', Hawai'ian auali'i, and Tongan hou'eiki. 11 Figure 1 below shows these inherited forms.
The Samoan, Tokelauan, Tahitian and Hawai'ian cognates enable us to reconstruct PTO *hau 'aiki, the earlier form of modern Tongan hou 'eiki. The PTO form retains the unassimilated vowels a before u and a before i. The modern reflexes also enable us to reconstruct PNP *sau 'aliki. The question now remains where the form Hawaiki would feature in this framework of inherited forms. It is the Tongic form *hau 'aiki that catches the eye, for supposing the glottal stop were to be dropped for some reason, the resulting hau aiki would be so reminiscent of Hawaiki as to hint, though purely on formal grounds, at a connection. But why would the glottal stop be dropped? It would be dropped if a borrowed form of *hau 'aiki which- 388 - 389
still retained the glottal stop were later inherited by languages in which the original glottal stop has been lost—languages such as Proto Central-Eastern Polynesian (PCE). This possibility supports the postulation that the PTO form *hau 'aiki was borrowed into PNP as the form *sau'aiki. Figure 2 shows the suggested changes undergone by the borrowed word *hauaiki.
The form *sau 'aiki in PNP seems to have been an etymological nativisation of the Tongic word *hau 'aiki. Etymological nativisation differs from ordinary phonological nativisation in that the sound which is substituted for the source language sound is not the host language sound which is phonetically most similar, but one which is perceived to be etymologically related. In the present case, PNP speakers would have been aware of numerous regular correspondences of PNP *s to PTO *h. Thus, since PNP had lost *h, the h in *hau 'aiki was nativised to s. Since PNP retained the PPN glottal stop, the nativised form became *sau 'aiki.
It appears that the process of nativisation is such that once a word such as *sau 'aiki has been nativised,
…it is native in every respect and therefore behaves like a native word, by for instance undergoing synchronically productive phonological rules. It will therefore from that point onward behave like any other native words also for the purposes of linguistic change (Hock 1986:397).
Thus, *sau 'aiki became directly inherited from PNP onward, resulting in the form *sau aiki in the post-Proto Samoic Outlier (PSO) language that was ancestral to modern languages such as Samoan, and in PCE, the language that was ancestral to modern languages such as Maori. At some stage in these languages, the form *sau aiki would have undergone a series of phonetic changes in which the unstressed u became a semivowel and associated itself with the following unstressed a to form [sáwa-íki]. In some languages this pronunciation would later have changed to [sa-wai-ki] with the stress on the middle syllable if the a and i diphthongised. 12
The change from *sau aiki to *sawaiki would be consistent with a shortening of the form from a five-syllable word sa-u-a-i-ki (based on five vowels) to a four-syllable form sawa-iki or a trisyllabic form sa-wai-ki. It is therefore likely that the phonetic changes would have been motivated by morphophonemic factors such as fast speech and would have resulted in the eventual loss of u in the modern reflexes, so that they now all have either w, as in Maori Hawaiki, or v, as in Samoan Savai'i. Although there are several examples of the change from u to w and sometimes to v in Tongan and Samoan—and it is likely that there are parallels in other Polynesian languages—I shall mention only one example from Tongan: tau ua 'i niu - 390 ‘pair of coconuts’ became pronounced as [tawa'i niu] as a result of fast speech. This further developed into the modern phrase tava'i niu.
At some point in PNP there would have existed (as will be seen clearly if Fig. 2 is superimposed on Fig. 1) the doublets *sau 'aliki and *sau 'aiki, the former being the directly inherited form and the latter being the Tongic loan. Just as there was a point at which the glottal stop was dropped from *sau aliki resulting in the modern reflexes, so there was a point at which the glottal stop disappeared from *sau 'aiki, leaving *sau aiki, the form now spelt *Sawaiki.
But the formal cognates so far discussed may not be cognates at all unless it is shown that they are connected in meaning. What Koskinen says about finding place-name cognates applies in this situation: “…before a pair or group of place names similar in form are cognates, the meanings of the terms must be the same as well” (1973:10). We may take for granted the cognacy of the place-name forms; for instance, Savai'i is clearly cognate with Hawaiki. What needs to be determined is whether there is a meaning connection between the place-name cognates as a whole and the PTO form *hau 'aiki, from which the place names are here said to be derived. In what follows, I shall consider the reconstructed meaning of *hau 'aiki and that of Hawaiki with a view to establishing their historical relationship. 13
Towards the meaning of*hau 'aiki
Knowledge of the meanings of the original component morphemes *sau/ hau and * 'ariki/'aiki should help in the reconstruction of the meaning of *hau 'aiki. 14 Assuming that the term was a compound with a likely structure of noun-adjective, we would mainly be interested in the meaning of *sau/ hau as a noun and the meaning of * 'ariki/'aiki as an adjective. If the meaning of the Tongan reflex hau is any guide, then the noun hau may have come from the (intransitive) verb hau, which means ‘to win over (a place or person)’, to be champion at (a game, sport, and so on). As a referential noun, hau would then refer to the person who haus, i.e., the conqueror or champion. This sense of ‘conqueror’ may have been extended to that of ‘ruler’, which is the usual English gloss given to the word. This meaning is present in some Polynesian languages but not in others. In Tahitian, hau means ‘ruler’ (among other meanings), as it is also in Hawai'ian, although the Hawai'ian use is now archaic and present only in some personal names such as haulani ‘royal ruler’. Samoan and Maori are examples of languages which seem to lack this meaning of sau/hau.
The PPN word *'ariki is often glossed as a referential noun ‘chief’ or an - 391 adjective ‘chiefly’. There are uses of the modern reflexes of the word, however, that suggest that the meaning of the adjective was more than just ‘chiefly’. In Tongan, as in some other Polynesian languages, the notion behind tangata 'eiki (or their cognates) is not ‘chiefly man’ but rather ‘elderly or venerable man’. The Samoan idiomatic phrase ali'i o 'āiga is not a reference to the ‘chief of the family’ but the ‘eldest of the children (in the family)’. One somewhat archaic use of the word in Tongan is when it is applied to a pig. If a pig has become 'eiki, it has lived too long to be killed and used as other pigs. It may then be simply presented to a high chief. The honorific word for ‘father’ in Tongan is 'eiki. This use supports the idea of 'eiki as ‘progenitor’ or ‘ancestor’. Koskinen, in his book Ariki the First-Born, gives various meanings of the word in several Polynesian languages and suggests that “in general there seems to have been a very intimate connection between the 'ariki rulers and the senior lineage” (1960:12). He advances the hypothesis that “in its original meaning the word 'ariki refers to the principle of primogeniture and to the first-born”. 15 In modern Tongan, the verb/adjective 'eiki has a relative sense. Even in commoner families, the most senior person is considered to be the most 'eiki.
These uses of *'ariki/'aiki cognates provide an important clue as to the justification behind the notion of ‘chiefliness’. It would appear that chiefliness arises out of seniority. Unlike sau/hau, it has little to do with achievement. Rather, it is simply accumulated through age to the extent that the longer one's line of ancestry is known to go back in time, the more *'ariki ship one and one's ancestors derive. 16 Not surprisingly, the ultimate ancestors were so * 'ariki that they were of divine ancestry or were themselves divinities. The oldest ancestors became perceived as gods. We may say, then, that 'arikiship, now understood as ‘chiefliness’, was originally the quality obtained as a consequence of being sanctified or hallowed by age.
Thus, simply glossing *'aiki (in *hau 'aiki) as ‘chiefly’ would obscure these important senses of the word. As an adjective in *hau 'aiki, it seems to connote the ideas of ‘senior, ancient, long-standing, traditional, original or ancestral’. So rather than ‘chiefly rulers’, a more illuminating gloss would be ‘ancestral rulers’ or ‘long-standing rulers’. The *hau 'aiki were of the highest rank, for they constituted the most senior lineages. Being the most senior lineages, they would be likely to have the largest number of descendants. And so the *hau 'aiki were considered to be ancestral to the people beneath them in the social hierarchy.
A consideration of the modern reflexes of PPN *sau 'ariki (e.g. Samoan sauali'i) may also contribute to an understanding of the meaning of *hau 'aiki. The meanings of the modern reflexes seem to indicate that the chiefly institution the proto-form referred to became either extinct or substantially - 392 modified in all countries except Tonga. In Samoa and Tokelau, the fact that the modern cognates now refer to supernatural beings suggests that the institution in each country became extinct. Samoan sauali'i is defined by Pratt (1984) as ‘a god, the respectful term for an aitu’, which in turn means ghost, spirit or god. The Tokelauan term haualiki refers to a demigod. The gloss of the Hawai'ian cognate by Pukui and Elbert (1986) suggests that the institution had become modified; auali'i is now an adjective meaning ‘chiefly’ or ‘royal’. The Tahitian word hauari'i, glossed by Davies (1978) as ‘kingly government’ and described by Newbury (1967:12) as ‘government by nobles’, seems to have been the institution that most resembled PTO *hau 'aiki. The Tahitian institution did not become extinct until recently. As Newbury (1967) explains it, the hauari'i declined by the 1880s and the hau Falani ‘French rule’ took over.
Part of the meaning of *hau 'aiki may be gleaned from the meaning of modern Tongan hou 'eiki. There are three main senses of the word. The first and perhaps most common is as a collective name for the Tongan aristocracy, which consists of the king, his family and close relatives, and the thirty or so titled aristocrats and their families. This group holds much of the political power in the country. Unlike the case of the Tahitian hauari'i, not even European contact and Christianity could stamp out the ancient institution, although the political power of the hou 'eiki became somewhat modified after contact. The western-type Constitution of 1875 has tended to legalise the traditional power of the hou 'eiki and, in a way, ensured its permanence. Only as late as a decade ago has there been any serious political opposition against the hou 'eiki. Despite strong opposition from pro-democrats today, the hou 'eiki continue to be very powerful and are strongly supported by the majority of Tongans. It may be that this is because the pro-democrats are fighting against not just a class of powerful people but an institution that has been deeply entrenched in the society for hundreds of years.
The second sense is one referring to the nobility excluding the king. It is likely that the word tu 'i has tended to take on a sense very similar to English king. As the king or tu 'i of Tonga, the monarch is separated from the rest of the nobility. The existence of a separate regal vocabulary has also helped to separate the tu'i from the rest of the nobility, which retains the title hou 'eiki.
The third sense is as a singular referential noun which means ‘king’ or ‘chief’. This use is restricted to very high-ranking chiefs such as the king himself or members of his immediate family. This sense seems to be the one given by Martin (1981), who defines the word hou'eiki as 'a title of address to a god, also to a noble …'.
An older use of the term is recorded by Koskinen: “the term hou 'eiki was applied only to persons of the highest ranks, who were known to be of - 393 divine origin” (1960:11). Bott makes a similar point concerning the word 'eiki, as referring to a person of whom, according to Churchward (1959), a collection would make up the hou 'eiki. Bott notes that there was a much more restricted use of the word 'eiki in contrast to its more general application to ‘chiefly people’:
Queen Salote said that when she was a child she was told that the only people who could correctly be addressed or described as 'eiki in this restricted sense were the children of an 'eiki fakanofo of Kauhala 'uta and the daughter of a king. An 'eiki fakanofo means an aristocrat who is formally appointed to a title; Kauhala 'uta are the titles deriving from the Tu'i Tonga line. The “king” in this context means either the Tu'i Tonga or one of the secular kings (hau), the Tu'i Ha 'atakalaua or the Tu'i Kanokupolu. Thus, the children of the Tu'i Tonga and the moheofo (daughter of the Tu'i Ha'atakalaua or Tu'i Kanokupolu) were entitled to be described as 'eiki and so were the children of the Tu'i Tonga Fefine [the Tu'i Tonga's sister; MT] and her husband, who were usually the Tu'i Lakepa or the Tu'i Ha'ateiho, both being aristocrats formally appointed to titles of Kauhala'uta… The children of the Tu'i Tonga and women other than the moheofo were 'eiki in the broad general sense; but not in the second, more restricted sense… (1982:60).
It seems, then, that hou 'eiki, like 'eiki, had an older, restricted sense which is likely to have been characteristic also of *hau 'aiki in earlier times. Only the very highest-ranking people in the land were *hau 'aiki.
Bott also states that “the word 'eiki was used only for men. Women of comparable rank were called ta'ahine (literally ‘girl’)…” (1982:61). She also refers to the hou'eiki as “generally the sons of kings” (p. 160). There is a possibility, then, that the *hau 'aiki were mostly, if not exclusively, male. We do know that it was the males in general who ruled and wielded political power throughout Polynesia. Because commoner families tend to model their lifeways on high-ranking families, it is possible that we have here the seeds of modern Tongan patriarchy.
It may be significant that the word hou 'eiki can be used without the definite article in a way that gives it the namelike quality of semi-proper nouns. Thus, one can say Tapu mo hou 'eiki ‘Let hou 'eiki be exempt’, as opposed to Tapu mo e hou'eikí ‘Let the hou'eiki be exempt’; or ko hou'eiki ‘it is hou'eiki’, as opposed to ko e hou'eikí ‘it is the hou'eiki’. The idiomatic reference to Tonga as fonua 'o hou'eiki ‘land of hou'eiki’, as opposed to fonua 'o e hou 'eiki ‘land of the hou 'eiki’, has the same structure as fonua 'o Tupou ‘land of (king) Tupou’. The significance of this usage without the definite article may be that the word is thus grammatically closer to personal - 394 and local nouns than to common nouns. If *hau 'aiki also had this use, then it would have been possible to speak of fonua 'o *hau 'aiki and understand it to mean 'land of *hau 'aiki, which could have been the antecedent of ‘land of Hawaiki’.
We may conclude, then, that the *hau 'aiki was the topmost of the chiefly classes in Tonga and they were the rulers of the land. Their chiefliness was derived from their descent from the earliest ancestors. The *hau 'aiki were of divine origin. In more recent prehistory it comprised the dynasties of Tu'i Tonga, Tu'i Ha'atakalaua and Tu'i Kanokupolu 17
Towards the meaning of Hawaiki
The place name Hawaiki and its cognates have no meaning in some of the places they are known in. In seeking a meaning for the Hawai'ian cognate Hawai'i, Pukui and Elbert (1986:62) could only say: “Elsewhere in Polynesia, Hawai'i or a cognate is the name of the underworld or of the ancestral home, but in Hawai'i the name has no meaning.” Savage (1962:56) defines the Rarotongan reflex in this way: “'Avaiki (Havaiki or Hawaiki) is the esoteric name of the ancient homeland of the Maori race.” For the name to be without a meaning or to be “esoteric” it is not surprising that it might be of foreign origin, or from a language other than the languages in which the place names occur.
In her perceptive study of Maori Hawaiki traditions, Margaret Orbell says that a new theory of the significance of Hawaiki must involve a reinterpretation of a great many narratives, passages in songs and karakia (‘charm, spell, incantation’) and customs. She believes that
though the Maori formerly believed Hawaiki to be a real place, we must now regard it as a paradisial land similar to other such lands spoken of in religious traditions.… These stories and beliefs are religious, and we can only make sense of them, and understand the range and subtlety of Maori thought, if we approach them from this standpoint (1985:3).
Orbell suggests that the name Hawaiki or cognates of it are given to a supernatural land, and in the two or three cases where the name is given to a real place, it is because the land has distinguished geographical features. Hawai'i in the Hawai'ian group contains the enormous active volcanoes of Kilauea and Mauna Loa. Says Orbell:
And this must certainly be the reason, for in Hawaiian traditions the crater of Kilauea, with its wide lake of molten lava, is the place where the great goddess Pele holds court amongst a throng of lesser gods and goddesses. … The volcano of Kilauea dominates the island of Hawaii and quite often it - 395 devastates large areas there. It is the source of such extraordinary and highly visible power that in the religion of the Hawaiians it was inevitably the most important of their sacred places. As the home of the gods it was, in fact, supernatural … (1985:6).
About another real island with the name, Orbell writes:
One other real island in Eastern Polynesia was called by a name that is a Hawaiki cognate, and it was also regarded as the home of the gods. The ancient name for Ra'iatea in the Society Islands was Havai'i, and the reason must again be the presence of a remarkable geographical feature. The island has a high mountain, an extinct volcano with a deep crater, and this crater was thought to be the entrance to the underworld … (1985:6).
To Orbell, the picture is consistent:
The people who settled Eastern Polynesia brought this name with them from their home, where it belongs to the largest island. In the east, in each of the island groups they discovered, they gave the name to a land with supernatural association—usually a mythical, imaginary place, but in two cases a real one (1985:7).
I differ from Orbell (and others 18) only in that the name did not originate in Savai'i, Samoa. It originated in Tonga where it was the title of the highest-ranking ancestors—those who eventually became perceived as gods in the mythologies. They were the *hau 'aiki, initially known and recognised in the Nuclear Polynesian area as Tongan aristocrats under the name *sau 'aiki. Later, as settlement proceeded eastwards and folk memory blurred, the name of the land from which the *sau 'aiki, perhaps now of the form *sau aiki, came (i.e., Tonga) became forgotten and, via mythology, the title itself replaced the name of the land. Thus, the title of the ancestral rulers became the name of the ancestral land. In the east, the word attained a religious meaning. It referred either to illustrious ancestral gods, reflecting the meaning of the PTO term (see below), or, more commonly now, to the ancestral land.
Since then, Samoan has had the pair of terms: sauali'i and Savai'i. Tahitian has had hauari'i and Havai'i. Elsewhere, there might have been haualiki and Hawaiki. These sets of doublets consist of a PNP-derived form and a PTO-derived form. The PNP-derived word 19 took on the meaning of ‘gods’, ‘kings’ or ‘chiefs’, while the PTO-derived word became a place name. In some places, the name was given to real places, and in others it was given to mythological places. But even the real places, as Orbell says, were not ordinary places. They were perhaps either the centre of the chiefly classes, as possibly Savai'i in Samoa, or the abode of the gods, as possibly Hawai'i - 396 in Hawai'i and Havai'i in the Societies.
If, as postulated here, the word Hawaiki really stemmed from *hau 'aiki, then we would expect, at least in some places in Polynesia, oral traditions that retain something of the original meaning and refer to Hawaiki as ancestors or gods rather than as a land. And indeed, in some places the appropriate Hawaiki cognate is the name of an ancestor or ancestors or the name for persons rather than a place name. Significantly, Lorrin Andrews (1974:152) says of the word Hawai'i: “From time immemorial the people have called themselves ‘Ko Hawaii’ and the islands ‘ka pae ‘āina o Hawaii’”'. Given the meaning of *hau 'aiki as ‘a class of people’, this distinction between the people themselves being Hawai'i and the islands as the islands of Hawai'i makes sense. Perhaps the people think of themselves as the descendants of the ancestral rulers or gods, the offspring of the original people. Of the Tuamotuan phrase Havaiki-nui-a-na-ea, Stimson and Marshall (1964:124) say: it is “a name given to certain ancestors of the people of Raroia”. More recently, Pukui and Elbert (1986) explain the meaning of Hawai'i-loa as the name of a legendary figure believed by some to have discovered Hawai'i.
Bellwood remarks that the word Hawaiki is not found in Western Polynesia, presumably because Western Polynesia is itself Hawaiki (1987:68). But the word is found in Samoa as the name of the main island of Savai'i. Only in Tonga is the place name truly absent. 20 Instead, only the prehistoric *hau 'aiki existed, with which places elsewhere in Polynesia were named. The evolution of the proto-form *hau 'aiki to Hawaiki is matched by the evolution of the proto-meaning from a title for a class of people to the name of a land.
The epithets of Hawaiki
Throughout Eastern Polynesia, from the Cook Islands through to the Marquesas, and from Hawai'i to New Zealand, single epithets or descriptive phrases are idiomatically attached to the name of Hawaiki or its cognate. The usual interpretation of some of these descriptions is that they refer to the size of the homeland. Maori Hawaiki-roa and Hawaiki-nui have been interpreted as ‘long Hawaiki’ and ‘big Hawaiki’. In Rarotongan, 'Avaiki-te-varinga-nui is glossed in Savage's dictionary as ‘the 'Avaiki-of-the-vast-expanse: varinga nui means a large area. The great original land from whence the Polynesians came.’ Andrews (1974:152) records Hawaiiakea as ‘broad or large Hawaii’. Stimson and Marshall (1964:124) gloss Tuamotuan Havaiki-nui-rau-mahora as ‘Great-Havaiki-of-the-wide-oceans’. Such epithets misled Percy Smith into writing:- 397
In some of these epithets of the ancient Father-land, it is clear to me that a continent rather than an island is referred to, and this is the description given to me of Hawaiki-nui … (1921:44).
In the light of the present exposition, however, these epithets are likely to be words of praise for the ancestral rulers rather than the ancestral land. Thus, Hawai'ian Hawai'i-loa and Maori Hawaiki-nui could have been references to ‘the tall ancestral rulers’ and ‘the great ancestral rulers’. The Rarotongan phrase 'Avaiki-te-varinga-nui, which Savage (1962) glosses as ‘the great original land from whence the Polynesians came’ probably originally referred to ‘the great-ancestral-rulers-of-the-vast-expanse’. Similarly, the Tuamotuan phrase of the ‘great-Havaiki-of-the-wide-oceans’ was really a reference to the ‘great-ancestral-rulers-of-the-wide-oceans’. Phrases like Tuamotuan Havaiki tumu, given Stimson and Marshall's (1964:572) gloss of tumu as ‘source, root, cause, origin; the fountainhead’ are likely to have referred literally to the ancestral rulers rather than the ancestral land.
Some descriptive phrases describe localities. For Marquesan, Dordillon (1931) states that Havai'i ta 'a'o ‘Havai'i down below’ refers to hell. In Tuamotuan, Stimson and Marshall (1964:124) record Havaiki-te-a-raro with the gloss ‘the subterranean hemisphere viewed as a land or country inhabited by a certain group of the gods known as tuputupua’. They also give Havaiki-te-a-uta as ‘the home of birds’. These expressions are reminiscent of Tongan hou 'eiki 'o e kauhala 'uta, i.e.‘hou 'eiki of the shoreward side’, and hou 'eiki 'o e kauhala lalo, i.e., ‘hou 'eiki of the seaward side’. 21 The former expression is a reference to the Tu 'i Tonga lineage, the highest-ranking lineage in Tonga, and its associated titles. These hou 'eiki were traditionally associated with a certain locality on Tongatapu that was designated Kauhala 'uta, so called because it was situated inland. The latter expression refers to the other two dynasties of Tonga, the Tu.i Ha 'atakalaua and the Tu 'i Kanokupolu and their associated titles. These hou'eiki were traditionally associated with the seaward locality known as Kauhala lalo. In Tongan the expressions are sometimes contracted simply to Kauhala 'uta and Kauhala lalo and sometimes to just 'Uta and Lalo, and these refer not so much to the places as to the lineages themselves. It is possible that the Eastern Polynesian terms originally referred to certain high-ranking ancestors who occupied certain areas known as Raro and Uta (or their cognates) in much the same way that the Tongan expressions do. In time, perhaps, the Havai'i/Havaiki became only a place name and the expressions came to refer only to localities.- 398
I have used the current subgrouping theory 22, principles of sound changes in Polynesian 23 as well as notions in loan phonology 24 to trace an etymological relationship between the PTO form *hau 'aiki and the modern Maori form Hawaiki and its cognates in Polynesia. It is generally accepted that PPN branched off into the subgroups PTO and PNP. Some of the distinguishing innovations between these two subgroups were the loss of *r in PTO and the loss of *h in PNP. Thus. from PPN *sau 'ariki was derived PTO *hau 'aiki. while PNP derived *sau 'aliki. The phenomenon called “vowel assimilation”. in which a low vowel before a high vowel becomes a mid-vowel, is found in Tongan. Thus, from PTO *hau 'aiki, modern Tongan derived the form hou 'eiki.
If PTO *hau 'aiki was borrowed into PNP, the loss of *h in the latter would be likely to result in the nativisation of h to s, thus deriving the nativised PNP form *sau 'aiki. Characteristic of some languages that developed from PNP was the loss of the glottal stop. In these languages the glottal stop was dropped from *sau 'aiki, leaving *sau aiki. The sound changes of u > w > v then took place in *sau aiki, resulting in the place name *Sawaiki. If a and i (in -wai-) diphthongised, the word became pronounced as Sa-wai-ki with the stress on the diphthong. From this word were derived the modern place-name cognates.
To determine whether the place-name cognates are historically related to PTO *hau 'aiki, the meanings of *hau 'aiki and Hawaiki were reconstructed and compared. It seems likely that, simultaneous with the modification of the proto-form, there has been a corresponding shift in the proto-meaning from the ‘title of the ancestral rulers’ to the ‘name of the ancestral land’.
If these formal and meaning relationships are accepted, then it is the wide-spread presence of the Tongic form in Nuclear and Eastern Polynesian languages that needs now to be explained.
IS A “HOMELAND” FOR EASTERN POLYNESIANS SUGGESTED BY THE DATA?
The notion of a discrete homeland is no longer fashionable among archaeologists, who prefer to think of the “homeland” as a region rather than a single island group. 25 One reason for taking this stance is that sometimes excavated items of material culture turn out to be typical of several neighbouring island groups which seem to have communicated regularly, with the result that it is not always possible to pin down an item to a particular island or archipelago. 26 Another reason is that in a place like New Zealand, it appears that there is linguistic evidence of impact from more than one language, so that it seems likely that the original settlers came from different - 399 island groups. 27 The idea of a “homeland region” seems quite well supported.
Yet, the case described here may be different. The fact that phonetic change is largely regular means that a (reconstructed) word can be shown to belong exclusively to a particular (proto-)language. If, as in the case described in this paper, a word of foreign origin is found extensively in areas where the foreign source language is not spoken, then the presence of that “foreign” word poses a valid question for analysts. And if that word happens to be, as in the case described here, the name of the mythological land of origin remembered in the traditions—and it would be rash to ignore the traditions—then that word poses even more pressing questions, especially if the “foreign” word is more widely attested in the area in question than its directly inherited cognates. In this case, a single island or archipelago “homeland” is indicated. 28
On the question of a single archipelago “homeland”, however, the climate of opinion would seem to favour Samoa. As Bellwood says, “…the Samoan Islands provide the best archaeological and linguistic homeland for the first settlers of Eastern Polynesia, who probably crossed to the Marquesas or Society Islands a century or two before the birth of Christ” (1987:54). Why, then, does the name of the mythological land of origin of Eastern Polynesians have a Tongic name?
Since we cannot ignore the archaeological and linguistic evidence pointing to Samoa or the traditions of Hawaiki, a name that now appears to be Tongic in origin, the question may force us in the direction that there may be different levels of interpretation of “homeland” arising from different courses of events in the earliest period. For instance, an “archaeological and linguistic homeland” would not necessarily be one and the same as a religious and political homeland. 29
In what follows, I shall attempt to reconstruct what little may be known of the *hau 'aiki as a political and religious entity that would be likely to “turn into” a religious and political homeland such as Hawaiki. At the same time, an answer to the question of why the borrowing took place may be more clearly formulated.
Towards a characterisation of *hau 'aiki
If the word *hau 'aiki was indeed borrowed in PNP times, then the fact of the borrowing presupposes that the institution of *hau 'aiki was well known, at least to the speakers of the borrowing language. Not only was the word well known, but it must have been of some significance in order for it to have become adopted and eventually assimilated as a native word. For a class of people to be known thus at this early period, it is likely that they were seafaring and mobile, at least initially. Most likely, the *hau 'aiki was - 400 powerful. If the *hau 'aiki was as powerful and as “international” as the hou 'eiki in the period just prior to regular European contact, then we may have an explanation for the borrowing of the word.
It is generally accepted that in the earlier part of the 2nd millennium A.D. the Tongans ventured forth and created something of a sphere of influence sometimes referred to as the “Tu'i Tonga empire”. 30 Koskinen writes of
a strong political influence which once radiated from Tonga, probably during the later Middle Ages: this hegemony of the Imperial Great Tonga strong in military rulers, termed tu 'itonga, tu 'iha 'atakalaua and tu 'ikanokupolu, has been remembered in the traditions of the neighbouring islands, and as shown during recent excavations, is probably manifested in archaeological material, for example in Samoa (1973:22).
It was this expansion that was probably responsible for the Tongan influence on East 'Uvea, East Futuna, Niuatoputapu and Niuafo'ou.
Bott (1982) partly accounts for this political influence by describing the hou 'eiki custom of sending out relatives as “governors” to outlying islands such as Niuatoputapu, Niuafo'ou and East 'Uvea. If a “governor” became absorbed into the local people thereby losing his rank, another “governor” would be sent from the central court at Tongatapu. This kind of custom indicates that the hou'eiki must have been known internationally and possessed a certain amount of political power over neighbouring islands.
It is possible that this kind of military ambition on the part of the hou'eiki was the continuation of an earlier tradition. Before the Tu'i Tonga dynasty, which is thought to have begun around 950 A.D., Tongan traditions speak of an earlier dynasty about which hardly anything is known (Gifford 1929:49). The reconstruction of terms such as *sau 'ariki for chiefly institutions at PPN, PTO and PNP levels would seem to be consistent with these traditions in implying that there existed chiefly institutions in the earliest period. The *hau 'aiki may have been the general name for the early dynasties. It is interesting that Bott maintains that the political situation in Tonga at the time of first contact had been unchanging for a long time.
…one of the striking things shown by the legends is that the kingdom as it exists today has been united under one government for a very long time. There is no evidence of conquests of Ha'apai or Vava'u, and equally there is no tradition of rebellions in these islands in the early period (1982:95 96).
If the *hau 'aiki came into existence at a very early period in Tongan - 401 prehistory, then this would make the Tu'i Tonga dynasty appear to be an instance—but perhaps a rather salient instance—of *hau 'aiki rule in more recent prehistory. This interpretation of *hau 'aiki would be consistent with that of Tahitian hau ari'i by Andrews and Andrews (1944) as hau ari'i ‘kingdom’, hau ‘rule’, ari'i ‘king’. If the *hau 'aiki was as powerful an entity as its more recent counterpart, it would not be surprising that the word Hawaiki has evolved the way it has.
One of the most striking features of the rule of the hou 'eiki at the time of first contact was their complete control of the common people. It is possible that this has always been the case. According to Tongan traditions, the common people have always been serfs who owned nothing but worked for the hou'eiki all their lives. It was not until European contact, with the introduction of the western form of government and system of laws, that the people were finally freed from their state of servitude to the hou 'eiki. Well-known European writers, such as Cook and Martin, from whom we have derived most of what we know about pre-European Tonga, noted the extraordinary power that the hou 'eiki exercised over the people, which often translated into harsh treatment of the common people. Bott quotes Cook and Anderson on this subject:
He (Cook) considered that the power of chiefs was enormous, and the inferior people were completely subject to the will of the chiefs to whom they belonged. He gives several instances of the disdain the chiefs showed for the lives of their subjects: While we were plying up to the harbour the Natives directed us to, the king kept sailing round us in his Canoe, there were at the same time a great many small Canoes about the Ships; two who could not get out of his way he run quite over with as little concern as if they had been bits of wood (1982:51).
Bott gives Anderson's impression:
We are indeed sorry to observe that they have little obligation to do justice to the inferior sort unless their inclination leads them to it; for we have seen some instances of their treating these with a degree of cruelty which seem'd to end only with their passion, and might to all appearance have been carried so far as even to deprive the offender of life, without hazarding any enquiry or being amenable either to justice or a superior—a sufficient proof of the arbitrary principles and unlimited authority by which these people are govern'd … (pp.51-2).
But what would seem to be the overwhelming power of the aristocracy would not be fully understood without reference to the status of the common people. There is evidence to show that the common people had no alternative - 402 but to accept their lot. Goldman quotes a missionary speaking of the exorbitant power of the hou'eiki over the common people and of the acceptance by the common people of such behaviour:
The power of the high was matched by the meekness of the lowly if this missionary observation about Tui Tonga is to be believed: If he wishes to satisfy his anger or some cruel fancy, he sends a messenger to his victim who, far from fleeing, goes to meet his death. You will see fathers tie the rope around the necks of their children whose death is demanded to prolong the life of the divinity (1970:303).
This same attitude of the commoners surprised Anderson:
…whether it be from that mode of government or some other policy to which we are strangers, it does not appear that any civiliz'd nation have as yet exceeded them in the great order they observe on all occasions and ready compliance with the commands of their chiefs, nor in the harmony that subsists throughout all ranks and unites them as if one man inform'd with and directed by the same principle. Such a behaviour is remarkably obvious on such occasions as requires their chiefs to harangue any large body of them collected together, which is frequently done. The most profound silence and attention is observ'd during the oration, even to a much greater degree than is practis'd by us on the most interesting and serious occasions, and whatever may have been the subject of the speech we have never seen any instance of their shewing signs of displeasure or that seemed to dispute the commands it gave (Bott 1982:48).
Commenting on the situation two centuries later, Bott remarks on the attitude of the hou 'eiki to commoners and vice versa:
Chiefs quite often (in 1958-60) speak of their kāinga in contemptuous terms, both to their faces and when speaking about them, and the higher the rank of the chief, the more contemptuously he can speak…. From the point of view of the commoner, a great chief is one who is strong, whose rank is high, and who is greatly respected by other chiefs. When praising their chiefs, these were the things his kāinga would point to, not to the fact that he treated them kindly. …Under modern conditions this attitude is changing, but one still hears it expressed occasionally. Thus, we were told that a particular chief had been very ferocious to his people, and we expected this remark to be followed by some statement to the effect: he had been a very bad chief indeed. But no, the remark that followed was: “We loved and respected him because he made us great” (pp.52-3).
It seems that the mistreatment of subjects by chiefs was something that was - 403 expected, as Bott explains:
…for a Tongan chief to affect indifference in public for the welfare of his subjects was a mark of etiquette, presumably because it showed that he was so great and so strong that he could afford to incur the displeasure of his people and did not have to rely on their support. Even today one finds the same sort of attitude, though in a very much milder form … (p.52).
If the power of the aristocracy stemmed from the time of the early *hau 'aiki, then it might have been a factor in the borrowing and eventual spread of the term Hawaiki.
There are features of the language and culture which may indicate that the power of the hou 'eiki has persisted for a long time. One such feature is the existence of the so-called “double moralities” called by Helu (1981) the “hou 'eiki morality” and the “commoner morality”. For commoners, great emphasis is placed on such “commoner moralities” as faka'apa'apa ‘respect’fakatōkilalo ‘humility’, and talangofua ‘unquestioning obedience’. The hou 'eiki have a different set of “moralities” which include to'a ‘courage’, fielahi ‘assertiveness’, and pule ‘authority’. There are ways of behaviour that are expected from, and appropriate in, the aristocracy but if manifested in commoners would be considered impertinent. A chief, for instances, can be angry with anyone whereas for a commoner to be angry with a person of some status would be labelled fie'eiki ‘presuming to be aristocratic’. The underlying assumption is that only aristocrats can rightfully be angry with other people. Commoners are supposed to accept their lot without complaint. There are a number of words that similarly show there are ways of behaviour which are accepted only when manifested in chiefly people.
The existence of different levels of vocabulary for commoners and chiefs would seem to be another feature of the language that reflects the longstanding power of the aristocracy. The so-called “self-derogatory vocabulary” which commoners use when speaking of themselves in the presence of chiefs testifies to the great gap between the two classes. But perhaps more significantly, the existence of the so-called “language of abuse”, which is understood to have been used in the past to address serfs and the lowest commoners but is now a language of anger, testifies that abuse of commoners must once have been a common practice. For these genres to have developed as important parts of the language presupposes that the aristocracy must have been a powerful group for a long time.
There are also speech customs that reflect the power of the aristocracy over the common people. For instance, very high-ranking people are able to change the names of those who have the same names as themselves and - 404 substitute rude words. Commoners who receive rude names from aristocrats feel honoured and proud of their new names even if these names are swear words. Such customs show that the power of the aristocracy was such that they could determine what people could say and what they could not say. This kind of language situation may help to explain the borrowing of the word *hau 'aiki as a highly political and, later perhaps, religious word. The fact that the word was taken as far as Hawai'i and New Zealand may be taken as testimony to the esteem with which the *hau 'aiki was once held.
Taken together, such customs as those described above would indicate that the common people regarded the hou 'eiki, in the past and, to a large extent, in the present, with the kind of awe that would be quite similar to those sentiments conducive to religious worship. There was a certain notion of greatness of which virtues such as kindness and principles such as equality were not necessarily a part. The point here is that it was prestigious to be seen to be descended from or related to the *hau 'aiki. In Tonga, this is still the case among many modern Tongans. Later in Eastern Polynesia, being descended from the hau 'aiki was recast in the traditions as coming from Hawaiki.
There is some indication that some early settlers of Eastern Polynesia were commoners. Koskinen (1973) considers that the chiefly Western Polynesian term sa'a ‘clan-like lineage’ was replaced in Eastern Polynesia by the plain word ngaati:
…the absence of [sa'a], together with additional evidence, is in favour of the suggestion that those leaving often belonged to the lower strata of the community: there are reasons to believe that they were often fishermen of those communities that were subject to the caprices of the ocean (1973:42).
The fact that many of the speech genres, such as honorific language, are absent in places like New Zealand as well as the general erosion of female kinship rank, such as the high rank of the father's sister, perhaps originally a chiefly custom, is consistent with commoner settlement. The language and customs for dealing with the hou 'eiki on a daily basis became lost, but the awe was preserved in the traditions and in some of the idiomatic uses of Hawaiki with descriptive phrases of praise. In this light, Rarotongan 'Avaiki-te-varinga-nui ‘Ancestral-rulers-of-the-vast-expanse’ and Tuamotuan Havaiki-nui-rau-mahora ‘Great-ancestral-rulers-of-the-wide-oceans’ can now make sense.
A consideration of the unlimited power of the *hau 'aiki would be consistent with the notion of exile, already discussed by others, such as Sharp (1956). Exile must have come in various forms, as we know from some writers and from Tongan traditions. Sometimes individuals would be - 405 exiled; other times perhaps a man and his family would be evicted from the land of the hou 'eiki; and yet other times perhaps a larger group of people, along with their (lesser) chiefs, would be told to leave. The possibility of exile, however, cannot rule out organised expeditions, as was evident in later military contacts maintained with neighbouring island groups, possibly motivated by the quest for land and, perhaps ultimately, tribute.
But this is not necessarily to summon up a picture of the *hau 'aiki sweeping across the Pacific, conquering and colonising as they went. Rather, this is intended to suggest that some awareness of the *hau 'aiki has always been present in the folk memory of Polynesians, emerging now and then in the mythologies, sometimes as ancestors, as in the Tuamotus and Hawai'i, and other times as an ancestral land, as in the Cooks and New Zealand. It is possible that some of the memories could have been of ancestors from Savai'i or of the land of Savai'i. Yet, because the original Tongic institution of *hau 'aiki, now Tongan hou'eiki, seems to have had a continued existence even to the present, and because the mythologies are replete with a Tongan presence in many places, the possibility is not discounted that the *hau 'aiki or people who were familiar with the *hau 'aiki reached even New Zealand. At this point, the situation remains complex. A number of explanations would “fit” some of the known facts, but the exact details of what actually happened to result finally in the distribution of the place-name cognates remain unclear. For now at least, it seems clear that there are sufficient formal and meaning connections to support in broad outline the hypothesis of the Tongic origins of Hawaiki.
The preservation of the terms
Supposing that the original word *hau 'aiki came into existence some 2,000 years ago, perhaps earlier, after PPN separated into PTO and PNP, how is it that the derived terms, now Tongan hou'eiki and Maori Hawaiki, became preserved and remembered for so long?
Although essentially hou'eiki and Hawaiki are now two different phenomena that are no longer recognised as related, both words have been preserved through the same genre of ceremonial language. In Tonga, the institution of hou'eiki has survived to the present day, powerful as ever, although somewhat modified. So the word is used in formulas at the beginning of formal oratorical speeches: tapu mo hou'eiki ‘hou'eiki is exempt’, fonua 'o hou'eiki ‘land (country) of hou'eiki’, fānau 'a hou'eiki ‘children/ descendants of hou'eiki’, kelekele 'o hou'eiki ‘land (plots) of hou'eiki’. Speakers always acknowledge the hou'eiki. Indeed, it is common to refer to Tonga as fonua 'o hou'eiki ‘the land of hou'eiki’. This idiomatic expression is sometimes the subject of ironic remarks about the not - 406 inconsiderable powers of the hou'eiki: “What else can you expect? This is, after all, the land of [king] Tupou and the hou'eiki.”
In Maori, the word Hawaiki has been preserved in the language of the marae. And so long as the marae stands as symbolic of Maori traditional culture, so long will the term Hawaiki be used as the symbol of Maori origins and identity. 31 References to Hawaiki the homeland have become almost formulaic in the ceremonial speeches of the marae. If there is a word that can persist in the memory of Polynesians in their centuries of wandering, it would have to be a word such as Hawaiki: a word of origin that tells of greatness and divine ancestry; a word of identity stamped with a glorious past; a word which, whatever the particular cognate and its language-specific associations, may be taken to symbolise the journey of Polynesians through time and space.
- 407 Page of endnotes- 408
1 This paper is an expanded version of a paper entitled “Hawaiki” presented at the Second International Conference on Oceanic Linguistics, Suva, July 1995. I wish to thank Ross Clark for helpful comments during the preparation of both versions.
2 An asterisk in front of a form shows it is hypothetical and not attested in modern languages.
3 For instance, Biggs (1993: pers. comm.).
4 In this paper I follow the Tongan convention and represent the glottal stop with an inverted apostrophe.
5 I am referring here only to the broad fact that Western Polynesia was settled before Eastern Polynesia. As Anderson (1995) says, there is yet no secure conclusion on the chronology of the settlement of Eastern Polynesia.
6 See Biggs 1978.
7 Why the phonetic change should be extended to the consonant v in only some words but not others is not clear. It may be that the superimposed sociolinguistic process of conventionalisation has a part to play here.
8 Dyen (1971) and Dahl (1976) discuss the possibility of similar sound changes occurring in Proto Austronesian.
9 My notion of the stress group in Tongan is similar to Schütz's “measure”, which he postulates for Fijian (see Schütz 1978), and his “stress unit” in Hawai'ian (see “Pronunciation of Hawaiian” in Pukui and Elbert 1986).
10 I think this is not an absolute condition, however. The word Niua, the name of an island in Tonga, which has a stressed u in Tongan, probably became Aniwa, the name of a Polynesian outlier in Melanesia in the outlier language of Futuna-Aniwa. My explanation for the change to the semivowel is that the stress system of this language may be divergent in some ways and may therefore permit changes that would not be permitted given the prosodic features of more conservative Polynesian languages such as Tongan.
11 For the purposes of this paper, I write Hawaiki and all its place-name cognates as one word; the modern reflexes of sau 'ariki (e.g. sauali'i) as one word; and all hypothetical forms as two separate words since their morpheme boundaries are still apparent.
12 This final change would not be possible in languages in which, like in Tongan, there are no phonemic diphthongs.
13 Since the Maori form Hawaiki is only one of several place-name cognates all of which are here claimed to be derived from PTO *hau 'aiki, the meanings and distribution of the cognates throughout Polynesia must also be accounted for. It is to be noted that some cognates, such as Hawai'ian Hawai'i, have no known meaning, while others, such as Samoan Savai'i, are explained locally by folk etymologies (see, for example, Pratt 1984:260), which I ignore here.
14 It does not appear that the compound word *sau 'ariki existed beyond PPN although modern Fijian has sau 'chief (Capell 1973) and modern Rotuman has sau ‘king/queen’ (Mary Jensen, pers. comm., 1995) and seems to have borrowed 'ariki 'chief (Biggs and Clark 1995).
15 See Pawley (1982) and Lichtenberk (1986) for more discussions of the meanings and reconstructions of 'ariki.
16 This probably partly accounts for the importance of genealogies in Polynesia, which serve to validate social status and establish identity.
17 The present king of Tonga is said to be the 23rd Tu'i Kanokupolu. The Tu'i Kanokupolu dynasty has significant Samoan origins, as implied in the name.
18 Such as Hale (1846) and Sharp (1956).
19 Along with PTO * hau 'aiki, since all were eventually derived from PPN *sau 'ariki ‘ancestral ruler(s)’.
20 The word Vaihi ‘name of a very distant land’ exists in Tongan, but it is probably a 19th-century borrowing from Tahitian. Also, several tracts in Tonga are known as Havaiki, but this name is probably also a borrowing.
21 See Bott 1982:79, 112, 122; and Gifford 1929:40.
22 See Pawley 1966 and Green 1966.
23 See Clark 1979 and Biggs 1972.
24 See Hock 1986 and Geraghty 1993.
25 See, for instance, Irwin 1992 and Sutton 1994a.
26 See Sutton 1994b
27 See Harlow 1994.
28 Niuean, one of the two daughter languages of PTO, has lost any reflex of PTO *hau 'aiki. Only Tongan has retained it as hou 'eiki.
29 Biggs (1972) points out that a prehistoric settlement can be completely obscured by a later major settlement and would not necessarily leave traces in the language.
30 See Mahina 1992.
31 According to Kawharu (1994), the questions of origin are questions of identity.