Volume 57 1948 > Volume 57, No. 3 > Prefixes and their functions in Oceanic languages (ma, nga), by C. E. Fox, p 227-255
                                                                                             Previous | Next   

- 227

MN, Melanesia or Melanesian languages; PN, Polynesia or Polynesian languages; IN, Indonesia or Indonesian languages; A, Arosi, San Cristoval; G, Gela (Florida); L, Lau, North Mala; M, Mota, Banks Islands.

NOTE.—In this paper the Melanesian g usually represented by the Greek letter gamma is printed in Roman type, as in the Mota m. The glottal stop is represented by q.


PASSIVES can be formed in Oceanic languages by prefixes as well as by suffixes, but the passives so formed are, so to speak, by-products; to form passives in this way is not their chief function or even an important one, and the prefixes differ fundamentally from the suffixes in that there is nothing in the prefixes themselves to give a passive turn or notion to the action. We have seen that the verb in itself is either active or passive. Since the suffixes connect the action with the object and thus make an object prominent, and always make the verb transitive, this in itself inclines to a notional passive, and a passive turn is easily given, and these forms by usage and custom tend to become passive forms. But the prefixes give no such turn for they have nothing to do with the object but show rather a change in the notion of the action as regards the subject. It may be true that originally all these monosyllabic formatives could be put either at the beginning or end of a word-base, giving a turn to the meaning, but with the meaning unchanged by their position before or after; but usage and custom tended to limit the notion of the word-base as regards the object with suffixes, and as regards the subject with prefixes, and that difference between them is so widespread and ancient that it is original IN, MN and PN, i.e., Austronesian.

- 228

The passives made by prefixes are of the following sort in MN, and are equally numerous in IN and PN.

  • A. bwa-hora = ngba-hora, spread.
  • M. ma-ukeg=ngma-ukeg, loosed.
  • A. apweta = ang-peta, split.
  • A. ma-hono, shut.
  • A. ka-ruru, submerged.
  • A. ra-boa, broken.
  • G. ta-vuke, opened.

These are in effect all past participles passive, and can be used as, and are, true passives, although the agent of the action is not expressed when such forms are used.

But we are not here concerned with genitive formatives as used in IN, e.g., di-salin nya, “it was copied by him,” where di= belonging to, and the justaposition of verb and pronoun also gives a genitive sense. Such passives should be considered under genitive formatives or verbal nouns, not with the prefixes. The above MN passives are, however, only a by-product of the functions of the prefixes, and the prefixes must first be considered more generally.


The prefixes can be divided into four groups:

  • (1) ma, nga, and decayed forms of these.
  • (2) ba, pa, and derived forms va, ha, fa, etc.
  • (3) ka, ga, and derived froms: qa.
  • (4) ta, ra, sa; ta being far the most common.

There are many compounds of these such as paka, manga, mara, bara, tava, etc.

They differ from the suffixes in that o and u forms occur; also e; i less frequently, in the final of compounds. The decay is much less than in the case of the suffixes, probably because they are at the beginning of the word, not at the end. But an even greater difference is that the vowel is sometimes omitted, the consonant of the prefix being substituted for that of the word-base, or by coalescing with it causing an alteration in the consonant of the prefix or the word-base or both, e.g., mono for ma bono (cover); nafa for nga tafa (open); gwela for nga bela (platform); kpwore for nga bore (dream); nggava for nga kava (big).

This paper will deal almost entirely with the first group, ma, nga, the most important in the languages whether in IN, MN or PN; and will deal very little with their functions.

- 229

However, a general word must be said about their functions which differ so greatly from those of the suffixes and are so much more numerous. The fundamental function of all the prefixes is to relate the action of the verb in some way to the subject, and thus to define it in various ways. The functions of the suffixes are various: such as making transitives, or causatives or intensifying the action and so on, or making verbs from other parts of speech or other parts of speech from verbs. The latter is also one of the functions of the prefixes. As for their other functions Brandstetter says the prefixes are inchoate, in function not form, and that because their functions are so many, and so shade from one into another, distinguishing between them is difficult. When, he says, they become definite in function they become local in use. This is true, but only partly so, and mainly of the compounds, but the functions (or the main ones) of the simple forms, and even of a few of the compounds, are so ancient and universal that they are common IN, MN and PN, i.e., Austronesian.

They define the action or state as regards the subject in the following ways, or we may say that they answer the following series of questions:

  • (1) Is the subject performing an action? Verbal formative.
  • (2) Is it in a certain state or condition? Conditional.
  • (3) Is it beginning this action or condition or about to do so, or to change it for another? Inceptive.
  • (4) When does the action take place? Temporal.
  • (5) Is something brought about or caused? Causative.
  • (6) Does it happen of itself merely? Spontaneous.
  • (7) Is it mutual action or condition related to another? Reciprocal.
  • (8) Is the action or condition partial or incomplete or of inferior quality? Depreciatory.

The answers to these questions are given in all the languages by the groups of prefixes given above, but the use of the different prefixes will be different in different languages, especially as regards the answers to questions 3, 4 and 8.

- 230

Examples are:

  • 1. Mao.—ranga, rise up; maranga, rise up. Verb formative.
  • 2. M.—wora, break; mawora, broken. Conditional.
  • 3. L.—lea, go; kalea, about to go. Inceptive.
  • 4. M.—ilo, see; meilo, saw. Temporal.
  • 5. M.—tira, stand; vatira, sat up. Causative.
  • 6. G.—voka, separate; tavoka, come apart. Spontaneous.
  • 7. Sam.—sili,ask; fesili, question. Reciprocal.
  • 8. G.—vola, live; hangga vola (hangaka-vola) convalescent. Depreciatory.

There may be different forms, e.g., of the conditional, in one and the same language: 1

  • A. ma-hono, shut up.
  • ka-ruru, submerged.
  • ra-boa, broken.
  • sa-kafu, muddied.
  • a-hita, 2 split.

These are the two most important verb-forming prefixes in our languages and can now be considered. They differ from the suffixes in that they affect the word-base to which they are prefixed, and the suffixes do not. The prefixes may lose their vowel. Probably the first stage is that a is replaced by an indeterminate vowel, written ĕ and called peppit in IN, and then the consonant of the prefix may replace that of the verb word-base, e.g., bono, cover; ma-bono, mĕbono, mono; tulu (sleep); ngatulu, mĕtulu, ngulu (Sam. ngulu, sleep). But with nga there is also often alteration as well as substitution. From tolo (swallow) we get in Mota not ngolo but nolo. Hai(=bai) is the Arosi reciprocal with relationship terms, but in Lau and Sa'a nga is prefixed and we get respectively ngwai and mwai, where original h becomes w, and in Sa'a ng changes to m. We now consider these changes.


Ma may alter to mo as in Mao. mo-roto, sunken inwards, (roto) or to me when it forms the past tense in - 231 Mota: me taur, built; perhaps in the first case from attraction to the vowel of the word-base, in the second from constant use. Both these alterations occur in IN and the second is very common.

When it is prefixed it is very frequently merely a verbal formative and does not alter the meaning:

  • IN. kan, eat; makan, eat.
  • ulug, fall; maulug, fall.
  • MN. too, sit; matoo, sit.
  • uri, live; mauri, be alive.
  • PN. ranga, rise; maranga, rise. 3
  • tika, straight; matika, stand erect.

However, there may be a change, 4 especially when the word-base has a labial as the initial consonant, but also when it is other than a labial, and in these cases the m of the prefix replaces the first consonant of the word-base. This is very common, and important for the study of the languages, 5 and for that reason a number of such changes in IN, PN and MN are given below.

IN. Malay:

  • mati, die; word-base pati, die.
  • malang, cross; word-base pala, cross.
  • maling, thief; word-base bali, steal.
  • mari, come; word-base gari, move, or boli, move.
  • merah, red; word-base bera, glowing.
  • minum, drink; word-base hinu, liquid.
  • muntah, vomit; word-base luta, vomit; puta, put forth.
  • muat, load; word-base luda, load.
  • masin, salt; word-base tasi, sea.
  • mudi, rear; word-base buri, rear.

MN. Mota:

  • mago, dance; word-base lago, step.
  • malamala, girl; word-base kala, child.
  • manga, mouth; word-base wanga, open.
  • maros, want; word-base garo, love.
  • mera, red glow; word-base bera, glowing.
  • mata, eye; word-base pata, inwards.
  • miti, bring close; word-base piti, near.
  • mono, patch; word-base bono, cover.
  • mule, refresh; word-base bula, live.
  • mutu, maimed; word-base putu, cut off.
- 232

PN. Samoan:

  • manga, branch; word-base tanga, fork.
  • mangō, shark; word-base pango, shark.
  • mangu, dried; word-base paku, dry.
  • mala, field; word-base bala, fence.
  • milo, twist; word-base piro, twist.
  • moe, sleep; word-base, bole, dream.
  • mule, despised; word-base bule, fool.
  • mili, rub; word-base biri, rub, smear.
  • moana, sea; word-base hora, spread.

It will be seen that many of the above m forms are nouns, and that is one of the functions of ma which makes nouns from verbs, or adjectives from verbs or vice versa. But the reader will feel doubt about some of the examples, especially those found everywhere in the m form, such as mate, mata and mai, from IN mari) which he is not accustomed to think of as derived forms. The fact that they occur in this form in MN, IN and PN shows, if the list is correct, that the prefixing of ma to these word-bases goes back to the time before the separation of the three branches, and that ma is an Austronesian prefix.

These substitutions of m for the initial consonant take place also in words found in this form in one branch only, such as moana, sea, only PN, developed since the separation. And, of course, they very often occur only in one region, or only in one language, and are quite local, and still more easily recognised. If it were a case of borrowing, PN from IN, or MN from IN, or IN from either of the others, this would not happen, we should get the same m forms everywhere. On the contrary this change is a law all three branches inherit, and it works through their history, applied in different branches, regions, or languages to different words, and so showing far more clearly their interwoven unity.

Mati, die, has this form in IN, but usually in MN and PN the older e form mate. The original word-base is probably seen in the Mao. pae, meaning demolished, overthrown and metathesised in Mota as pea, just as mate is metathesised in M. as meat, low tide (elsewhere always mate, mati, mae, mai). Mate is used for low tide because of the coral reef which is thought of as overthrown, pae, brought to nought, pea, dried (and so called rango in one - 233 language, i.e., the reef at low tide) lifeless, or its life hidden under rocks till the life-giving tide returns. Anyone who knows the reef will understand the native feeling. Where there are no coral reefs the old word is still used. Mate thus means: in a condition of overthrow, brought to nought, and has a far wider meaning consequently than killed or dead; it can be used of a man who is sick and unable to move or unconscious or numb, or of putting out a light 6 or in other ways. IN has the form pati in one language.

Mari, come, is almost always in the more decayed form, mai. I would not say more than that it may come from gari, which is an A. word meaning to move from one place to another and is found elsewhere as ari, ali, alial, and so on. If it is from this, the meaning in the m form has become limited. There is, however, another word, G. boli, which means also to move from one place to another, and in both Arosi and Munggava (Rennell) boi is used for the usual mai. Whichever it comes from, mari is a derived word. IN has the older form, PN and MN the more developed.

Manga is an old m form for wanga, waka to open or gape. In Mao. with the transitive suffix i (mangai, opened 7) it means a mouth, in G. manga without the i. In A. it means the groove in the ridge pole; in M. the gill of a fish; in Sam. the curve of a hook; in Mao. a ditch or water-course. There is, however, another word-base very common in MN, tanga, sanga, to fork, as the legs from the body (A. masanga, twins) or a branch from a tree, or one road from another, and when manga in Mao. means a branch this can be taken as the derivation. The word is also IN.

Mata, an eye, is, according to Brandstetter, original IN and is also original MN and PN, and so Austronesian, but it is doubtful if any of these m words is underived, and there is a word-base in Mota, pata, 8 which means inwards or into; in Arosi mata (maa in A.) means, besides eye, a hole or opening or hollow or gate (where you go in) or the - 234 edge or point (the edge of a knife or axe, the point of a spear, i.e., the part that goes in), a stain (that sinks in), a groove for rubbing fire, a crystal embedded in a rock or the mesh of a net. Some of these might come from eye but not all, and inwards into, seems to be a fundamental meaning to cover them all, in which case mata would be the hollow in the face. With this m form already in use for eye could come later derived meanings such as front (of a man first, then of anything; G. mata, eye, face, front), face, circle, lid, to stare and so on. Mota has both matarag and patarag, to stare and to peer, but used synonymously.

These derivations are open to argument, probably because they are such old words. About most of the others there can be little doubt, a little need be said about them.

M. mango, dance, from lago, step, has a close parallel with Fijian make, dance, leke, leap, and A. mao, dance, rao, move about (the latter Bauro). IN muat, load, may be metathesis for muta. Luda is the widespread MN word for load. PN moe, to sleep, is Fijian modhe, and this appears to be derived from M. qore (ngbore), A. bwore (ngbore) and G. bole, all meaning to dream, but M. and A. have the word with the prefix nga. Moe, we take it, means to be in a condition of dreaming, in dreamland. Mala or mara, an enclosed cultivation, is also MN and evidently derived from the very widespread bara, a fence, found throughout MN and often decayed in PN to ba or pa. Moana from hora (decayed to hoa) meaning spread out, a very widespread word-base in IN, MN and PN. This is confirmed by the G. word for the ocean, which is hora-ra. 9 The A. and L. word is matawa, an m form of tawa, taha, to extend, and so similar in meaning. In A. māi tawa (mata i tawa) is a sheltered landing place. Tawa is a prefix to names of villages in such harbours. The Wa Tawa or A. Tawa are the first immigrants, and though this is etymologically Maori Hawa and IN Saba, it may only mean the People from the Ocean, the great open sea to the north. Thus the G., A. and PN words for the Ocean, “the wide waters,” are all formed in the same manner.

Every language will supply the reader with these m words, often from labials, but by no means always, where - 235 the m of ma has been substituted for the first consonant of the word-base. Maori has many, such as maki as well as mahaki 10 for sick (IN. sakit, MN hagi, tagi) mere (kere), meha (keha), muri, breeze (guri), and so on. It may be that mana meaning spiritual power, is one of these and the m form of the word-base pana, hana, hot. There is some confirmation of this is a similar use in Mala of the word ako as meaning either spiritual power or heat; and in Arosi the word amengo means either the cooling of a red hot stone or the losing of mana by a man, owing to something he has done. It will be remembered that in the Christian story mana came down from heaven as fire, and the conception of mana as a kind of fiery energy (always too “from heaven” in the Melanesian mind) does not seem alien. Mana may mean “in a glowing condition.”

Finally in dealing with these m forms we may cite some interesting ones from the word-base G. vola, to live. The Fijian is bula, Malay orang, a man and orangutan or hutan, a large ape, Mao. ora, so well known to New Zealanders in the phrase kia ora. Ori is either a decayed form of this or a parallel form to uri. If we take ori as from ora the m forms are A. mora, living, real, original, natural, normal, usual, of the country, i.e., indigenous. Both maori and mori also occur with much the same meaning, and haqamori (M. “whakamori”) means yes, truly. From the other word-base uri (IN hudip) we get A. mauri, living, flourishing, and Areare haauri, cause to live.

As an example of the need for caution in deriving words we may cite M. oraora, sportive, playful, one would suppose meaning lively, full of life. But A. parallel is nora, itself from tora (ngtora), move swiftly, restless, Mao. ora, wag, and oreore, fidgety, a more prosaic meaning for the Mota oraora.


We have been considering ma and the way in which it alters the word-base. We now turn to the other nasal prefix nga which is also universal in IN, MN and PN, but has a much greater effect on the following consonant.

Brandstetter considers it one and the same as the article nga, and concludes that verbal formatives are no - 236 different in origin from articles of nouns. We can accept this whether nga comes before the noun or, as in so many of our languages, is suffixed to form nouns, since formatives could originally go before or after the word-base they limit. Brandstetter draws his conclusion with regard to the identity of the noun article nga and the verbal prefix nga from the modern Java ula nguntal, translated the “snake swallows,” which is really, he says, the “snake the swallower,” though we do not so translate it, i.e., the verb is originally thought of and treated as a noun. Articles according to this are really verbal (or noun) formatives become separated by usage and custom, just as we saw verbal suffixes could be separated and become prepositions. We might call what are usually termed verbal particles simply verbal articles. Probably it would be better to do so. As with the suffixes there appear to be more than one crystallisation, the earliest almost any monosyllable, and the second and later more limited and often compound, for example the compound causative paka, vaka, haka, later than the simple pa, va, ha. Nga is one of the very early simple forms. In A. it is added to verbs, nouns and other nouns to make verbal nouns: adaro, beat a gong, adaronga, the beating of a gong; ari, to bend or curve, aringa, the bend of a river. This use of it is very general, common in G. and L. In PN nga is the final syllable of the forms manga, tanga, etc., making verbal nouns. The nga we can take to be identical with the verbal prefix.

IN grammarians are inclined to think the noun was originally plural rather than singular in thought, and that to show singularity an article meaning “one” had to be employed. Plural nouns would therefore, one may suppose, give us the earliest noun article, and the Maori nga with plurals can be considered a survival, nowhere else preserved pure and undefiled of nga formatives with nouns, 11 but no fossil and very much alive. As a fossil it occurs in singulars or plurals such as ngarahu, ngakau, ngao, ngare.

There is no doubt that originally articles could be either prefixed or suffixed. In IN the article i usually precedes the noun but comes after it in Sumatra; the article e usually precedes but follows the noun in Bugis (Celebes). - 237 Mota has two articles o and i, one before, the first an ordinary article and the second giving a general sense o mata i, eyes or eye.


Nga like ma may be simply a verbal formative without change of meaning in the word-base. Since this prefix makes changes in IN and MN, especially in either the prefix or the first consonant of the word to which it is prefixed, or like ma replaces that consonant, examples of this simple use do not look quite so plain as with ma though in reality there is no difference. It must be noted that in IN ma frequently combines with nga to make a compound prefix.


  • Borneo—ngujan, rain; Malay—hujan, rain.
  • Malay—membalut (me-ng-balut), wrap; Malay—balut, wrap.
  • Malay—mengankat (me-ng-ankat), raise; Malay—angkat, raise.


  • M.—qisang (ngpisang), crush; G.—pisa, crush.
  • A.—bwore (ngbore), dream; G.—bole, dream.
  • A.—ngari (nggari), move about; A.—gari, move about.


  • Mao.—nga-hora, spread; hora, spread.
  • Mao.—nga-keke (creak); keke, make confused noise.
  • Sam.—nga-solo, slip; Mao.—horo, slip.

Nga also makes past participles passive, as ma does: Mao. nga-hae, torn, hae, tear (M. sare).


But what is so remarkable about nga is the changes in the word that occur when it is prefixed. They occur in PN and are of the same kind as those in IN and MN, but are not nearly so striking. It almost looks as if the peoples of MN and IN had found this a very difficult sound, and made rather a mess of it, using, as Brandstetter says, a number of “compromises,” and different compromises in different parts of IN, and this is exactly true of MN, the compromises are surprising and are different in different languages.

The chief compromise in IN are as follows:

(1) When ng precedes r, s or t, and combines with one of these, the result is n. This also occurs in MN and PN, but more often ng simply replaces r, s, or t.

- 238

(2) When ng combines with a following labial, the result of the combination is m. Also found in MN and PN but more rarely.

(3) When ng combines with a following g or k the result is that the g or k is lost. In MN and PN such a substitution is common, but not only with g or k, but with other letters also, as was the case with ma.

Brandstetter also gives cases where ng is prefixed to a verb base beginning with a vowel, when the ng unchanged, but as these words once had an initial consonant and ng is an ancient prefix, it is probable that in most of these cases ng was substituted for the consonant once present, and that these are cases of substitution like those in MN and PN, and like the cases of substitution of m in IN. For example, in angkat above, IN has lost the r present in MN and PN where the word-base is ranga, ng was probably substituted for r.

The following examples are given by Brandstetter of the ways in which nga affects word-bases in IN, when prefixed.

  • Philippines—ma-nagni from sagni, ngs to n.
  • Java—ma-nurung from turung, ngt to n.
  • Fagala—ma-mokot from pokot, ngp to m.
  • Celebes—nganro from kanro, ngk to ng.
  • Borneo—ngujan from (h)ujan, ngh to ng.
  • Sumatra—ngapit from (k)apit, ngk to ng.
  • Madagascar—nilu from ilu, ng (= n) added.

The first change caused by this prefix is the same as that caused by m, a substitution of ng for the consonant of the word-base.

  • A. ngahu, strike; word-base rabu, strike.
  • ngado, corner; word-base rado, join.
  • ngari, move about; word-base gari, move about.
  • nguru, fall; word-base turu, down.
  • ngore, prawn; word-base ura (kura), prawn, red.
  • ngusu, spit; word-base pusu, puff.
  • ngoa, 12 broken; word-base hoa (bota), break.
  • G. ngara, blood; word-base tara.
  • ngiha, how many; word-base hisa.
  • ngoti, cut; word-base koti.
  • ngata, strong; word-base kata (Mao. kaha).
- 239
  • L. ngae, dung; word-base tae.
  • ngilo, twist; word-base filo.
  • ngidu, nose; word-base risu.
  • ngalu, stir; word-base kalu.
  • M. ngol, cut off; word-base holo.
  • nganga, come out; word-base ranga, rise.
  • Fiji ngane, 13 male; 14 word-base tane.
  • ngunu, drink; word-base hinu.
  • ngone, 15 child; word-base kane?
  • ngavu, dirty; word-base kafu.
  • ngata, 16 snake; word-base?
  • ngauna, season; word-base tau.
  • ngutu, cut off; word-base putu.
  • ngole, turn head; word-base gole.

In some of these cases ng may have been added to decayed roots as A. nguru from a form uru, for t is dropped in Arosi, but not in all cases. There is some substitution and not only with k and g as in IN.

The second change is like that in IN where the combination of ng with r, s or t leads to the compromise n as in IN. This is fairly common and only a few examples need be given. It is a change in IN, MN and PN.

  • A. nagi, flint; Ulawa ngagi; M. lak, hard, ngl.
  • nuga, loose; luka; ngl.
  • nugu, wrinkle; rugu; nr.
  • niu, 17 coconut; M. ma-tig; ngt.
  • Sa'a niho, tooth; riho; nr.
  • nima, house; rima; nr.
  • Fijian noku, bend; A. loku; ngl.
  • noto, sit; L. toto; ngt.
  • L. nima, hand; lima; ngl.
  • namu, flog; ramu; ngr.
  • nafu, ashes; rafu; ngr.
  • nau, snatch; sau; ngs.
  • M. nolo, swallow; tolo; ngt.
  • nau, leaf; rau; ngr.
- 240
  • nara, 18 blood; tara; ngt.
  • nal, stroll; tal; ngt.
  • manara, wise; rara; ngr.
  • G. mata, plain; rata; ngr.

As usual some of these forms, indeed most of them, are local. But very old is the M. ma instead of nga prefix to the word for coconut. The g of ma-tig is the Melanesian g very close to r, and leading to it. A final u can be inferred in Mota, which closes syllables. The IN niru may well be metathesis, so common in IN, for niru, itself from ngtiru. But niu is common IN, MN and PN, that is Austronesian, and the Mota m form therefore gives us a very old word indeed, possibly not preserved in any other language, except of course the neighbouring ones. Santo has ma-tui.


In considering the way in which the formative ng affects the word-base, this Malay word, adopted into English, but spelt amok in Malay, is interesting. Setting aside the final k as a petrified noun suffix or demonstrative, we get the true word-base amo, which has, we know, lost an initial consonant. It means “to make a furious attack”; as the verb form is meng-amok, probably a substitution of ng for a letter not a labial. From MN we find that the lost consonant was r. The Arosi form is ramo, strong, brave, bold, fierce, a warrior, a killer. On Mala, where the people were fighters, some were pre-eminent, murdering ruffians who went about killing. In Lau ramo means strong, a champion, a hero, to be fierce and violent, in pidgin “him want fight all time.” In Sa'a, Ivens gives “renowned in fighting, a brave”; he might have added a bravado. The word is used everywhere on Mala of man or dog or any fierce fighting animal. In Mota it is the centipede, sulata ramoa, 19 the “fierce worm” who attacks at sight and whom all avoid. In Mao. we find the word as amo, to charge or rush at a person. The Mala ramo was both a hero and a murderous ruffian. A good portrait of a ramo on a larger canvas was Henry Morgan, a fierce, cruel and bloodthirsty buccaneer, but a hero in England where he was knighted - 241 by King Charles, or Hitler, “that wicked man” to Mr. Churchill, but a hero to the Germans.

Now we know that if ng be substituted for the r of ramo we shall get namo and so we have the Malay namok (or nyamok, showing the influence of the ng), meaning a mosquito. It is still ramu 20 in Samoan, but elsewhere namu. Tregear quotes Mangaian: “e te namunamu! The blood-thirsty one!” reduplicated to show he is the little blood-thirsty one, who wants to “fight all time,” the little ramo of the coral island. In Mao. he is namu, a sandfly, wasp, blister; and a “little blister” he is. In M., G., A. and L. he is also namu. So as we follow the word by regular phonetic change we see in turn the Malay running amuck, the Solomon island killer, a mad dog, a savage centipede or a bloodthirsty mosquito.

It is this sort of deep seated unity of the languages, mutually throwing light on each other, not a mere borrowing from each other, which gives the conviction that all three branches stem from one stock.

When we next face this little warrior, out for blood, irritating, insolent (one meaning of ramo) and persistent, we shall easily recognise his etymological kinship with the mad Malay, the Malaita ruffian, the savage centipede—with what we mean by “running amuck.”


We now come to a series of interesting and intricate compromises in MN when ng is prefixed to g or k, to b or p. When nga is prefixed to g or k the g is kept, but the k becomes g. This is seen in Gela and Fiji.

  • G. nggari, child; word-base kari.
  • nggau, branch; word-base kau.
  • nggura, heat; word-base kura.
  • nggalo, scratch; word-base kalo.
  • nggoru, crunch; word-base koru.
  • Fiji nggali, twist; word-base kali.
  • nggalo, swim; word-base kalo.
  • nggara, cave; word-base kara.
  • nggari, scratch; word-base kari.
  • nggovu, 21 clouds; word-base kovu.
- 242

When ng precedes b or p the compromise in Lau is gw and in A. bw, pw, in M. it is kpw.

In A. bwa and pwa are formatives as gwe is in L., i.e., they are compound prefixes, ngaba, ngapa, ngape, and they have a depreciatory force. The compound prefixes are dealt with later.

  • A. bwaa, taro; word-base bata; ngb.
  • bwara, oppose; word-base bara.
  • bwau, head; word-base batu.
  • bwea, platform; word-base bela.
  • bwore, sleep heavily, dream; word-base bore.
  • pweta, smash 22; word-base pota; ngp.
  • pwono, stop up; word-base pono.
  • pwote, with child; word-base pote. 23
  • L. gwau, head; word-base batu; ngb.
  • gwela, shelf; word-base bela.
  • gwero, mushroom; word-base bero.
  • gwai, oil; word-base bati (water).
  • M. kpweta, taro; word-base beta; ngb, ngp.
  • kpwong, night; word-base pongi.
  • kpwir, close; word-base piri.
  • kpwatu, head; word-base patu.


  • kpwaranga, cave; word-base kara.
  • kpwilo, pool; word-base kilo.
  • kpwelu, crooked; word-base kelu.

It will be observed that M. kpw (ngp) sometimes represents k forms (ngk). M. qaranga and qilo = Fiji qara and qilo (F. q = ngk). These will be considered presently.

When ng precedes m, and sometimes when it precedes a labial, the compromise is m in Mota, mw in A. and Sa'a, ngw in Lau. These forms are found commonly like the bw, pw, gw forms as prefixes, i.e., ma, mwa, ngwa, bwa, pwa, gwa and represent the compound prefixes ngma, ngba, ngpwa, comparable with the IN mang, etc.

  • M. mol, orange; word-base moli.
  • mata, snake; word-base mata.
  • moa, first; word-base wota.
  • motu, sever; word-base putu.


  • malu, waves in commotion; word-base kalu, stir up.
  • mao, star; word-base kalo.
- 243
  • mera, mere, child; word-base kare.
  • momo, pool; word-base koho.

Further we find all these compromises in the middle of words, such as lima, hand, ima, house, lama, strike, uma, work a garden; and the same is true of G. and F. ngg, and A. bw and M. kpw. To consider these as infixes would lead us nowhere. We should have to consider M. lenga, walk about (leka), lang, energetic (lak), ringi, power (riki), in the same way.

The intrusion of nga as verbal formative, or noun article, in all the foregoing words, at the beginning of the word-base, appears certain, for the L. forms: ngwane, male, ngwaa, snake, are plainly ngw.

But this ngw in Lau is represented in Sa'a and Arosi by mw which is therefore also ngm.

In Mota the Sa'a and Arosi forms in mw are represented by m, which is therefore also ngm.

But there are exceptions where a word-base with k, not a labial, have also this prefix, where for instance M. kpw = ngk not ngm, or M. m = ngk not ngm. This can only be explained by attraction, the assimilation of sounds. When the compound sounds caused by the prefix ng become customary and familiar, other consonants get attracted to the sound, and it is also put for m, b or k in the middle of a word. This would be especially the case if these compound sounds became familiar through the frequent use of compound prefixes at the beginning of words, and this is actually the case, for in all these languages the prefixes ma, mwa, ngwa, bwa, pwa, ngga are plentiful.

Examples of these very common compound prefixes are:

  • Fiji ngga-kilo, a valley.
  • M. ma-sekeseke, joyful.
  • A. mwa-hiohio, sway.
  • L. ngwa-sinasina, shining.
  • A. bwa-hora, broad.
  • A. pwa-ruru, knee.
  • L. gwe-rodo, night.

Probably the reason for the e form instead of the a form is that it is the most-used prefix in Lau. It can be used before almost any noun, i.e., it is an article in general use, and before verbs and adjectives it makes nouns. Sometimes it alters the meaning gwela, shelf, gwegwela, bier. It - 244 is also a descriptive prefix describing an individual by his qualities and pecularities, as gwe falai, “bald head,” gwe-rumula, a hairy man. It stands for ngpe and in some respects resembles the use of peng in Malay; just as ngma can be compared with mang in IN.

In a previous paper 24 in J.P.S., MG Law is described through which m is represented in other languages by n, g or k; and this is true as far as it goes, but the cause is the prefix ng, not a phonetic law in the usual sense. It is true that child is mera in M., mwera in A., ngwela in L, but mera is for ngmera and mera is itself an m form of kare or else it is an assimilation form. Fiji ngane 25 is the equivalent of mane or ngwane simply because all are forms of tane: the m prefix, the ng prefix and the ngm prefix. M. malo, hidden rock, L. ngwalo is Mao. ngaro, hidden, Areare maro, G. ha-galu, because the word-base karo with the prefixes m and ngw are cases of assimilation probably, and ngaro and maro, ng and m forms by substitution. Uma, umwa, unga all represent uma (but not Mao. huke as suggested) because the m has been assimilated to the Mota m. The island, Mala, can be considered the m form of kala (by substitution). Mwala and Ngwala are for Ngmala, etc. Fiji ngone, child, is simply one (M. anai, IN anak) with the ng formative. What is called the BG Law can also be explained by the prefix ng.


In PN we shall not be troubled by such compromises as we find in MN, nor to the same extent as we find in IN, for the compromises are rare, and when they occur are simple ones, and nga, in the full form of the prefix, is much more common.

As in MN we find substitution of the ng for the first vowel of the word-base.

  • Samoan ngana, glow; word-base hana.
  • ngulu, sleep; word-base turu.
  • ngafa, fathom; word-base tafa or rafa.
- 245
  • Mao. ngoto, deep; word-base roto.
  • ngunu, singe; word-base tunu.
  • ngare, members of family; word-base kare.
  • ngaru, wave; word-base kalu.
  • ngao, trough; word-base kao, bottom.

As an alteration of the consonant from the influence of the formative:

  • Sam. navu, lime; word-base savu.
  • namu, mosquito; word-base ramu.
  • nefu, turbid; word-base refu.
  • nifo, tooth; word-base rifo.
  • nui (?), great; word-base tui.
  • Mao. naku, scratch; word-base raku.
  • noa,feel with hand; word-base rao.
  • nawa, distant; word-base tawa.
  • nehu, dust; word-base rehu.
  • niho, tooth; word-base riho.
  • manaha, open country; word-base mangtaha.

In a PN outlier, Munggava (Rennel) we find the alteration of k to g as in Fiji and Gela. To this we shall return. The alteration of a labial to m no doubt occurs, but the difficulty is to decide whether this is due to the influence of ma or nga since both cause this alteration.

As a simple prefix nga is common in PN.

  • Mao. nga-ora, burst open.
  • nga-hora, spread out.
  • nga-ruru, abundant.
  • nga-hae, torn.
  • nga-oriori, lullaby.
  • Sam. nga-lala, long for.
  • nga-lulu, shake.
  • nga-olo, rattle.
  • nga-epuepu, stirred up.
  • nga-foa, notched.

How are we to decide whether a labial has been lost by the substitution of m, or altered to m by the influence of ng? In some cases it is possible to tell. For example, S. H. Ray in M.I.L. says that M. miti, press together, and mutu, maimed, are from ngpiti and ngputu, but the rule in Mota is for ngp to become q (kpw) or m, and piti and putu with ng would become qiti and qutu; 26 these, therefore, are m, - 246 not ng substitutes. So also the A. and L. toko (toqo), to strike, can give moko or moto (tattoo and strike with fist, M. tut) by substitution of m, but ng prefixed would give noko and noto. The G. guri, a breeze, with ng could only give nguri or nuri; Mao. muri must be an m substitute. 27

But with words like muru, smear (buru), moe, sleep (bole), maho, quiet (G. beto, M. ta-peto), we cannot tell, and may take our choice: “ng or m as the case may be,” to misquote the Anglican Catechism when a child is asked his name.

It is curious that the two formatives most widely occurring and most deeply imbedded in the Oceanic languages, whether in IN, PN or MN should be the nasals m and ng, but it is a remarkable example of the fundamental unity of the languages.

- 247

Munggava and Munggiki are the correct names for Rennel and Bellona, two islands lying roughly fifty miles to the south of the south of the Melanesian chain, and both PN. In this language we have ng in a great many words as a prefix: ngg stands for nkg, and k replaces the r of Maori words, so that we have a change from r to k. In Hawaii we have a change from Mao. t to k; so that this may be considered the PN example of Brandstetter's RGH Law, the MN RGT Law. We explain the k-r change as due to an original “Melanesian g” close to r, and either becoming r or hardening into g and k. Munggavan is one of the most interesting languages of PN. In Vikings of the Sunrise, Sir Peter H. Buck draws a line from Samoa to Munggava and considers the people a backwash from Western Polynesia, but does not explain how they got so far south, nor how and why they made such fundamental changes in their language. Nggava means great and nggiki little; they stand for PN rava and riki, and the islands are “Great Mu” and “Little Mu.” Tangaroa is Tanganggoa with them. Maui remains Maui.

The change from k to g when ng is prefixed is the same as in Gela and Fiji, but occurs in every word which has r in Maori. As with Mota we have the problem as to how the medial ng was introduced, as in hangge for house, and can only offer the same explanation that it occurred through attraction, the sound ngg used initially came to be used also for the same sound in a second consonant.

Here are a few words of their language with the same words in Samoa, from where Sir Peter thinks they came.

M. S. M. S.
wash huhui fufulu sing mako pese
work hekau galue see ina iloa
house hangge fale smoke āu asu
hook gau matau scratch laku felau
speak henggeu ngangana teach ako a'o
man penggea tangata untie huke tatala
male tangata 1 tane seen ngga'a
hill onggo maunga answer hakahoki 2 tali
put tuku tu'u stand tu'u tu
sick masasaki ma'i big nggava lele
1   ngata, male; M. ata.
2   Sam. fa'a causative.
- 248

Bishop Williams who studied a sketch of the grammar and 1,000 of their words thought it nearest to Maori. The nearest Melanesians to them are the Arosi people who have no tradition of intercourse. The Munggavans had no seaworthy canoes, and used rafts. The Arosians had fine canoes, capable of holding one hundred and forty men in the largest (I knew of this one) and knew of the existence of Amoraha and Amoriqi, as the Arosians called them, but they were reputed to be distant islands inhabited by amazons, who propagated their race by the use of banana juice. There are some words similar, like nggava and raha, and both peoples use boi instead of mai. The island is fifty-six miles long with a fine lake which the writer has explored, but the soil is very poor and could never have supported a large population. 28


Another way in which the prefixes differ from the suffixes is that they lead to a greater variety of compounds in several strata or crystallisations, least in MN, most in IN but fairly numerous in PN. Among these compounds some are very old, common to IN, MN and PN, but there are many other later and more local ones. None seem to be as old as the simple forms. Just as the compound suffixes manga, tanga, mina, tina, etc., are later than the monosyllables, so the compound prefixes such as paka, manga, bari, are later than pa, ma or ba (causative, conditional, reciprocal). The last are often found only as fossils, where the compound forms are alive. Manga is common in IN, but only in fossil form in MN and PN, if even that, though ngama may be the same prefix. Manga seems a late development. It is of course only by fossils that unwritten languages can be read, just as the story of the rocks can only be read through their embedded fossils.

The causative prefix paka occurs in IN, MN and PN, but Brandstetter does not think it “original IN.” It is found in G. as vaga, A. haqa, L. faa, M. vaga, Fiji vaka; and in PN as Mao. whaka, Munggava haka, Sam. fa'a, Tahiti faa or haa, Haw. haa (later hoo), Tonga faka. Brandstetter says it is secondary, later than pa, though found in the Celebes, Philippines, Java and Sumatra. Pa is frequent in MN.

- 249

S. H. Ray writes: “Alongside pa, the chief causative formative, there is the less widely distributed secondary form paka in MN but with some restrictions in use and meaning.” He goes on to argue that therefore (because of these restrictions) MN received it later from IN than it did pa. So also, then, must IN have received it later than pa from where? In PN this secondary and later form is universal and so could not have given pa to MN, where it is common.

The meaning that could be drawn from these facts is that PN has had a longer development, and so has developed farther, than IN and MN. Pa is only fossil in PN. This was also the case with the suffixes. Something seems to have retarded growth and development in IN and MN. It is the same with decay; PN has decayed more than IN and MN. Development and decay have advanced farther in PN than in the other two, as we saw was the case also with the Passives. MN especially is at an earlier stage of Austronesian. This would appear to agree with the hypothesis that Austronesian was the native language of those who now speak PN, but a foreign one to those who now speak IN and MN. We should then expect in IN and PN a retarding of development, and less decay.

Malay has the compound prefixes meng and peng, and these may be the same as MN ngam and ngep; though which is the transposed form 29 there is nothing to tell us (unless we beg the question and declare MN derived from IN). PN here seems less developed and keeps to the simple forms nga and ma, i.e., is more primitive for once than IN and MN. The other compound paka shows IN and MN more primitive and PN more developed. The third general compound bari, the reciprocal shows MN most developed and PN least, and most primitive (although more decayed).

But it is remarkable how IN and MN make what Brandstetter calls “compromises” with the sound ng, and to some extent with m. PN does not do so to anything like the same extent; it either substitutes these sounds for the consonant of the word-base, or prefixes them in full form nga, ma. Not so IN and MN. For example, when - 250 we have the form pao, to shoot or sprout, Mao. simply replaces the p with ng and makes ngao, but Arosi loses the ng altogether and replaces the p with two labials, pwao. IN has nyamok (ng before r), but Maori simply namu. The compromises with ng in MN are very complicated and numerous, and to a less extent in IN. In both IN and MN these are also different compromises in different areas. Thus in IN and MN we get fewer words beginning with the prefix nga, where PN, Maori or Samoan, has a good many. All this gives at least the appearance of IN and MN receiving a foreign tongue and making rather a mess of some of the sounds. You get the same appearance when metathesis runs riot in IN, to which Brandstetter refers again and again. It does not run riot in MN as the “compromises” do, but it is very considerable. Metathesis is what sometimes happens when people are hearing, but not hearing too well, a foreign language. When Tongans settled on Oakea about a hundred years ago they introduced into the nearby language of Mota only one word: siopa, cloth, which still remains, but is a metathesis for siapo. Of course this is not the main cause of metathesis, which is probably simply slang. “Slang,” says Fowler in Modern English Usage, “is the diction that results from the favourite game, among the young and lively, of playing with words, and renaming things and actions; some invent new words or mutilate the old, for the pleasure and novelty; and others catch up such words for the pleasure of being in the fashion; many (new) words perish; a few establish themselves.” Yet people trying to learn a new language can cause metathesis. 30

Nevertheless, development is not always greatest in PN. Consider another of these widespread compounds, what is usually called the reciprocal bari. Malay has ber. MN has various forms. M. vari, A. hari, L. fai, G. vei, Fiji vei. But though Samoan has fe (the older, because not the compound form) Bishop Williams does not give whe as a reciprocal in Maori, perhaps because it is fossil, not living, in that language. But obviously it is the IN, MN and PN reciprocal in words like whe-ako, whe-ao (A. ao, - 251 rule), whe-au, whe-inu and others, while in wheiwhei, quarrel, we have apparently the MN and IN compound. Moreover, I would suggest that it is the reciprocal whe, in a fossil form, in several other words. One universal function of the reciprocal is to form a plural, either plural verb as in Samoan, or noun as very commonly everywhere. Consider these three words in Maori, whenua, wheke and whetu (land, octopus, star). May not whenua mean a “collection” of nua, whatever nua means. 31 Whenua is common to IN, MN and PN from Malay benua, continent, to (MN) Lau finua, island, the world, the weather, A. hanua or henua, the whole island, San Cristoval (Hanuatoo), village, Bauro binua, M. vanua, island or village, Fiji vanua, land, region, place, to the PN forms fenua, fonua, honua, henua, which may mean the known world. May not the meaning be fundamentally a “collection” of houses (village), places (region or island), islands (known world) and the whe be the reciprocal in its earliest crystallisation? As to wheke, octopus, this name for it is only PN and not MN or IN so far as I know, and so this is used as an argument that PN is distinct from the other two. The word-base is best seen in Lau gari (octopus), A. is quria. Gari is apparently a word-base (karu, kari, garu, gari, kau gau) meaning to grasp, and is used for things that grasp (hand, claw, tentacle), or creatures (e.g., crab) that do so. May not whe here be the reciprocal and ke=gari, i.e., a tentacle, so that we might translate whe ke as “multiple tentacled”; (picture an octopus in movement). As to whetu we are on sure ground. The G. is veitugu, and vei is the G. reciprocal. M. is vi-tu, A. he-u, and in L. though another word (bubulu, shining), is used, star is febubulu. Malay is bi-tang. Must not whe here mean “company,” in G. the “company” of tugu (speakers, witnesses). I suggest then that whe, he, vi, 32 bi above are a first crystallisation of the reciprocal in its simple form; bi older than ber in Malay, he older than hei, hari in A., and unrecognized for what they are.

- 252

But in the reciprocal MN has developed much farther than PN, and IN also more than PN. Here MN seems to take the honours. 33


The present writer does not think the chain of small islands, running S.E. to N.W. parallel to the Melanesian chain, but about 100 miles north of it, are old stepping-stones of the Polynesian trek into Polynesia. They have had no influence either on the people or languages of Melanesia. But it will be necessary to study their languages to come to any definite opinion about them. Some of these people certainly came from the east, from Tonga and Samoa. But some seem to have come from the north; and possibly during the long period, when the Polynesians were moving east into Polynesia, some parties of them moved south to discover Munggava and other islands. 34 An important language to study would be Pileni. The Rev. H. N. Drummond, who spoke the language well, considered it an important Polynesian language. It became long ago the lingua franca of the Santa Cruz group and is understood by the pre-Melanesian speaking peoples of Santa Cruz, Nifilole and other islands, as well as by, of course, the Polynesian speakers of Matema and Nukapu (where Bishop Patterson was killed by Polynesians). Ndai, given in Sir Peter Buck's list, is Melanesian. But none of his arguments against the Polynesians going through Melanesia to Polynesia holds against a far earlier trek of proto-Polynesians who went through Melanesia as far as Fiji, and on further; and such a trek (if they did not go that way into Polynesia) is required by both linguistics and ethnological facts—for example by the linguistic fact that Melanesian languages of a large area are Austronesian through and through, fundamentally one in structure, grammar, and even vocabulary, with the languages of Polynesia and Indonesia. This argues a large number of immigrants and (because there are at least two strata in the languages) more than one wave of people, of a superior and dominating culture.

- 253

That the Polynesians were once in Melanesia, through the length of it along the line of the Trades, in large numbers, unifying it, giving it a common Austronesian tongue, deeply influencing the pre-Melanesian inhabitants, mingling with them and intermarrying with them, seems beyond doubt. They held sway and were the masters of Melanesia, at least from the central Solomons to Fiji, and then were swamped by force of numbers, leaving the great gift of the one language, and a blending of their thought and characteristics with those of the pre-Melanesians, and giving them that Golden Age of which Arosi traditions tell, very long ago, in the far past, when peace, and free travel and intercourse and trade flourished among the Melanesian islands.


1. Matrineal Society. It is true this was the basis of Melanesian society, but it is not all the truth. There is also patrineal in MN. All Mala, which has the largest population (50,000), is patrilineal. There the chiefs hand on from father to son, and they can recount their genealogies for over thirty generations. The eldest son is chief, the second priest, and knowledge of senior and junior branches carefully preserved. The same was true of the chiefs of Arosi (the Araha) who handed on power and chieftainship from father to son. The same is true of other large areas. Various other things of course went from father to son, such as ownership of, and ceremonies with, the sacred sharks. In Arosi there were a number of people called Matai, 35 the skilled craftsmen, canoe builders and so on, who handed on from father to son. It is a question how much the sister's son position is matrilineal. It was very important in Mota where a man's sister's son could take any of his uncle's goods, his uncle was his ma-rau. In Arosi the same was true, where each was mau to the other, and where it gave the nephew a position in other villages that was valued. Parallel to this, and very like it in privileges, was the position of ma-rahu. A man became ma-rahu to another (exchanging names) and thus got privilege among the people of his ma-rahu. A chief of Wango became ma- - 254 rahu to a man in Santa Ana, and even when Wango and Santa Ana were at war he was safe there; he was also privileged to take any of the property of his ma-rahu. In Gela the word is mavu. This custom was also practised in the patrilineal island of Mala, and Arosi men became ma-rahu to men on Mala and then, even if the two peoples were at war, the ma-rahu could go from one to the other.

2. Hau and Sau. These are two Arosi words with much the same meaning. Sau in Arosi is an honorific prefix to the names of heroes of old time, and famous familiar sharks, and the same word is in South Mala where it is used in the same way, and with names of chiefs. But Mala sau should be au in Arosi 36 and probably the word sau came from Mala, between which and Arosi there was regular intercourse. It appears to be the Mao. hau, famous, illustrious, and is in PN.

Hau, also hou, is quite another word, h standing for a labial. It means exalted; haqahou is to exalt a man, and hou-ra is a kingdom. T is lost in A. and the M. is patu (with ng: qatu) and there it means head, literally and metaphorically, head cook, or Archdeacon (qat priest), or Pope (qat Loglue), one at the head of work or organisation, the principal, chief person. It is Maui's M. name. In Mala hou is: to acclaim as famous; and the noun form means glory, honour, fame. It is PN. Tahiti fatu, Hawaii haku, Marq. fatu, all given as meaning lord or master. 37

3. Kava. Kava was made and drunk ceremonially in Arosi long before the advent of white men. The kava ceremonial drinking was during burial rites. The plant was called qawaqawa and there was another variety of it called baqe kakawa but not so good for making the drink. The custom was common on Mota and the plant called gea, possibly the same word. In Fiji it has only a descriptive name: yanggona, i.e., bitter (M. gona).

4. Kumara. Neither the name nor the plant were known in MN. It is a recent introduction from New Zealand via Norfolk island and Melanesian teachers. Even now it is foreign and not thought to be proper food for a man. Names are sometimes given to it, as in Fiji: kawai ni - 255 vavalangi, foreign yam, or in L. kai rogi, yam with flower like a rogi, native “morning glory,” convolvulus. This absence of name and thing may be thought to show the Polynesians were never in Melanesia, but actually it confirms the belief that they got the kumara much later, after they came to Polynesia; from Peru.

5. Maui Stories. The writer grew up as a boy in New Zealand, much among Maoris, and with many Maori friends, from whom he first heard these incomparable stories. They are the common heritage of all New Zealand boys, whatever the colour of their skins, since the Vikings of the Sunrise settled down amicably with the Sea Rovers of the Sunset, the one having found America from the west, the other from the east, after which both went south to become New Zealanders. He next heard these stories as a young man on Mota, from the children, different in many ways; yet Patu was Maui again without any doubt. For the third time he heard them fresh from Aosi, a Melanesian lady of Arosi of high rank, into whose family he had been adopted, and who was full of stories. Maui was Waro this time, and again the stories were different, but it was the same youngest of the brothers, much cleverer than they, mischievous and doing incredible feats, and always getting the better of his brothers. Patu and Waro did not pull up islands, perhaps because the Polynesians were not yet doing so in Polynesia; but Maui must be long before that time. And probably the stories about him grew and changed, only the essentials remaining.

1   The conditional prefixes form passives when the verb to which they are prefixed is a transitive verb; also ta which often gives a sense of spontaneity, often does not, and then forms a passive if the verb is transitive. ma, nga; pa, ba; ka, ga; ta, ra, sa, all form past participles passive in various languages, like the participles formed from transitive suffixes. The causative forms a notional passive, and is often used as a passive.
2   =tahita.
3   Note transitive suffix making past participle passive in maranga-i raised up.
4   Complications are caused in IN when ma or mu is used as an infix, but are not dealt with here.
5   It is very useful in comparing words, e.g., G. kabu, rest. Stop is mapu in Samoan and mapu-sag in M. This may be substitution.
6   I have heard a European say: “How quaint of them to say they ‘kill’ a man when they only knock him down, or ‘kill’ a light.”
7   Or an old noun article.
8   The Mao. waha, opening, mouth, Tahitian vaha, would be the same if the h here represented t as it generally does. The reduplicated first syllable vava is mouth in Malagasy and speak in Mota. U. wawa, mouth, and IN baba. But the IN form bawa seems to show that this Mao. h represents a labial.
9   cf. the East Coast Maori use of tahora for “the open sea.”—Eds.
10   Munggava masasaki, sick.
11   M. makes plurals with nga reduplicated: ngang, but it comes after the noun.
12   cf., Mao. ngota, broken.
13   The ng form where IN has the m form mane.
14   cf. Maori tungane, brother or male cousin of a female.—Eds.
15   The word-base is uncertain. The base without the initial consonant is found in M. anai, IN anak and Fijian one.
16   The m form is mata.
17   Malay nyior, where the y shows the influence of the ng on the t as does nyamok, mosquito (from ramo), or nyala, shine (from lala). Tiru or tigu is the word-base of niu.
18   A word may show a substitution in one language, an alteration in another: G. ngara, blood, M. nara, A. manaranara; all from original tara.
19   M. sulata, worm, Malay ulat, PN ulo.
20   The change from o to u means decay, a word in very common use.
21   The Mota forms of the last three are kpwara-nga, kar, kov.
22   Bauro pota.
23   G. boke.
24   “Phonetic Laws in Oceanic Languages,” pp. 71-76, Vol. 56.
25   The author of the New Fijian Dictionary is in playful mood when he gives the first meaning of ngane as sister (b.s.). This is parallel to the use of vavina in G. for brother (w.s.), my brother vavina-nggu.
26   Or possibly miti and mutu. M. motu, severed, is ng prefix. The change from r to n may not always be due to the prefix ng, but to ordinary phonetic law. The change is common internally in Maori, e.g., aneane, areare; monemone, moremore, etc.—Eds.
27   The Samoan word ngafa means a fathom, and there does not seem to be an obvious Samoan word from which to derive it, though ava, wide apart, to open, seems the most likely, but there are two word-bases, both PN, and both used to make “fathom.” Here are some of the forms:
  • A. taha, to open, reach, extend to.
  • taha-nga, fathom (nga as suffix).
  • tawa, open, matawa, ocean.
  • tewa, long, awa, abide.
  • Mala taba, taha, fathom.
  • M. taua, open.
  • Malay tabah, increase.
  • tyapai, extend to (note tr. suffix).
  • A. raha, great, haqaraha, increase, araha, chief.
  • Mala laha, great, alaha, chief.
  • M. lava, great.
  • Fiji raba, wide, probably rawata, reach.
  • G. lava, in manalava, one who talks too much.
  • M. labah, spacious, lapang, wide.
  • Mao. whaka-rawa, increase, (raha, open, extended.—Eds.)
  • Munggava nggava, great=rava.
  • O FORMS:
  • Munggava nggoha, fathom=roha.
  • Mao. roha, extended, roa, long.
  • M. rova, fathom.
  • There are also m forms:
  • Sam. mamafa, heavy.
  • Mao. taimaha, heavy.
  • Tah. aha, heavy.
  • G. and M. mava, heavy.
These, however, are not “to be large,” and so heavy, but are said to be from O. Java wwat, heavy, L. baba, babasi, weigh on, and so must be from substitution of m, not from ngb (L. ngwaba, M. mava).
28   The islands should be called by their native names.
29   This however is not a case of metathesis, but reversed order of the prefixes, IN putting ma first and MN nga. Neither is derived from the other.
30   For example, the Manginangina Block in North Auckland is often called Maninganinga by Europeans.—Eds.
31   It is tempting to take it as the MN muta, island, but there are difficulties.
32   Vi is not given in the New Fijian Dictionary but is evidently a reciprocal, for it occurs in a tale in Kinship in Fiji, Capell and Lester, p. 305: vi talatala, translated “talking together.”
33   The development of the reciprocal in Arosi is richer than in any Oceanic language known to the writer.
34   If so the Manggavans passed through peoples already speaking Austronesian, brought into Melanesia by a far earlier movement.
35   M. is matai, excellent, “o matai carpenter,” clever carpenter; G. matagi, alert, clever.
36   A. has ao, rule, and aonga, ruler. L. has aoria, ruler or high priest.
37   The Fiji etymological equivalent could be vasu.