Volume 75 1966 > Volume 75, No. 1 > Polynesian languages: a subgrouping based on shared innovations in morphology, by Andrew Pawley, p 39 - 64
POLYNESIAN LANGUAGES: A SUBGROUPING BASED ON SHARED INNOVATIONS IN MORPHOLOGY 1
1. The claim is made here on grounds of comparative morphology that all the well-described Polynesian (PN) languages spoken within the Polynesian Triangle 2, apart from Tongan, Niue, and possibly Uvean, belong in a subgroup of Polynesian. This subgroup, called here Nuclear Polynesian (NP), also includes at least one Polynesian Outlier 3.
Tongan and Niue are regarded as comprising a second major subgroup, co-ordinate with Nuclear Polynesian, and called here Tongic (TO). 4- 40
An essentially similar subgrouping has been proposed by Elbert, primarily on grounds of comparative phonology. 5
Further sub-classification of NP languages into an Eastern Polynesian (EP) subgroup and a Samoic (SM) subgroup is proposed here on morphological grounds detailed in 4.2 and 4.3.
2. The following hypotheses, conventions and terms to be used here require clarification.
2.1.1 Morphological evidence for sub-relationship between certain PN languages is provided by the presence in these languages of uniquely shared features which can be shown to reflect innovations of the subgroup ancestor. Demonstration of innovation by a sub-proto-language presupposes correct reconstruction of the proto-language. A particular reconstruction in the proto-language is assumed to be correct in at least the following circumstances:
The innovations postulated as the basis for the subgrouping claims made above all involve reconstructions of type (2), where external witnesses can be seen to agree with one set of internal witnesses, permitting the conclusion that this set of internal witnesses reflect the proto-language while the remaining internal witnesses exhibit innovations. An item is reconstructed for Proto-Polynesian (PPN), for example, only if (1) it is reflected by all the chief witnesses (i.e. by both TO and NP languages) or if (2) its reflection in at least one PN language is cognate with an item in at least one non-Polynesian language, such as Fijian. TO and NP languages disagree, for example, as to the shape of the second person dual marker in the nucleus slot; Tongan and Niue forms reflect an earlier *kim(o)urua while NP languages reflect an earlier *ko(u)lua. Fijian (Bauan), with kemudrau ‘2nd pers. dual, nuclear’ provides decisive evidence for reconstructing PPN *kim(o)urua. The conclusion here is - 41 that the languages reflecting *ko(u)lua form a subgroup, and that the form *ko(u)lua was innovated by the subgroup ancestor after it had diverged from the language(s) ancestral to Tongan and Niue. (See 3.2.2 for fuller discussion of this example).
Internal PN evidence is weighted according to the assumptions of sub-relationship stated in 2.1.2. to 2.1.6 below. Certain languages, e.g. Eastern Polynesian languages, are assumed to have been until recently a single language, and therefore in reconstruction of PPN are regarded as the equivalent of a single witness.
2.1.2 The name Tongic (TO) is given to a putative subgroup with two members, Tongan (TON) and Niue (NIU). The parent language of this subgroup is called Proto-Tongic (PTO).
2.1.3 The name Nuclear Polynesian (NP) is given to a putative subgroup consisting of Samoan (SAM), Futunan (EFU), Ellice (ECE), Tokelauan (TOK), 6 Tikopian (TIK), Easter Island (EAS), Marquesan (MQA), Mangarevan (MVA), Tahitian (TAH), Hawaiian (HAW), Rarotongan (RAR), Tuamotuan (TUA), and New Zealand Maori (MAO). The parent language of this subgroup is called Proto-Nuclear Polynesian (PNP).
2.1.4 Within NP a further division is made into Samoic (SM) and Eastern Polynesian (EP). The SM subgroup includes SAM, FUT, ECE, TOK, TIK while the EP subgroup includes MQA, MVA, TAH, RAR, TUA, MAO, HAW. 7 The parent languages of SM and EP are called here, respectively, Proto-Samoic (PSM) and Proto-Eastern Polynesian (PEP).
2.1.5 Apart from Tikopian (included in SM) data is quoted from the following unclassified Outliers: Mae (MAE), Nukuoro (NUK), Kapingamarangi (KAP), Futuna-Aniwa (WFU), Sikaiana (SIK) and Pileni (PIL). Sub-classification of these Outliers is not attempted here for several reasons. In most cases the descriptive data are not detailed enough to permit full comparative treatment. And in several of these languages, spoken by small populations, there are indications of borrowing of uncertain extent from several as yet undefined PN and non-Polynesian sources.
It is notable, however, that all six of these Outliers possess at least some of the features noted in 3.1 as characterizing NP as opposed to TO languages, including features (1), (4) and (5).
2.1.6 Uvean (EUV) is not included in any subgroup. Evidence of extensive borrowing in Uvean requires that it be excluded from a genetic sub-classification until sources and relationships of the borrowed and inherited components have been defined. 8- 42
2.1.7 Material is cited from a number of non-Polynesian languages. Most weight is given to data from Fijian (FIJ) and several languages spoken in S.E.Solomons: Sa'a, Ulawa, Nggela, Bugotu, Vaturanga, Kwara'ae, Lau and Fagani. It is assumed, following Grace and Biggs, 9 that these languages belong with PN in a subgroup of Austronesian (AN), called here Eastern Oceanic (EO) and derived from Proto-Eastern Oceanic (PEO).
2.1.8 Phoneme correspondences between PPN, Proto-Austronesian (PAN), PEO, and PN sub-proto-languages are shown in Table 1. Reconstruction of PAN phonemes follows Dyen's orthography except that ŋ is used for his N. Reconstruction of PEO phonemes follows Biggs's orthography. 10 Reconstruction of PPN follows Elbert except that a is used for his a, æ, and o. Reconstructions of Polynesian sub-protolanguages are mine. For all languages the orthography of the sources is used except that in all cases glottal stop is written ?.
TABLE 1 PPN consonants and their correspondences in PAN, PEO, PTO, PNP, PSM and PEP
2.1.9 The morphological material to be discussed consists of a number of morphological markers. In PN languages such markers are, for the most part, either single morphs or sequences of morphs which belong to small substitution classes and which carry grammatical meaning, e.g. MAO te ‘definite article singular’, which belongs in a two-member class with ŋaa ‘definite article plural’, nei ‘near speaker’, which substitutes for - 43 two other markers also indicating position in time or space relative to speaker and addressee, and tee-nei ‘this’, which substitutes for a small number of demonstratives each of which can be analysed as consisting of a definite article plus a position marker. 11
Morphological markers contrast with bases, morphemes which belong to large open substitution classes and which carry lexical meaning, e.g. MAO fare ‘house’, which belongs with several thousand other bases in the N (or Noun) class. 12
In any PN language most markers can be sub-classified in the first place as nuclear or peripheral, according to whether they occur in the phrase nucleus as affixes to bases, or whether they occur in the periphery of the phrase. Markers can be sub-classified in the second place as preposed or postposed, according to their position relative to the nucleus (in the case of perpheral markers) or relative to the base to which they are affixed (in the case of nuclear markers). Further sub-classification can be made according to the co-occurrence privileges of markers relative to each other and to base and nucleus sub-classes. 13 E.g. MAO te and ŋaa form a paradigmatic set of definite articles within the class of nominal preposed peripheral markers, characterized (for instance) by their occurrence preceding nuclei of class N, A or G, but not preceding nuclei of class L, V or P.
2.1.10 The terms cognate and reflex are used of markers shared by collaterally and lineally related languages, respectively, which fulfil the usual standards of correspondence. A reflex which corresponds in all features (formal, semantic and distributional), with the earlier stage, is said to reflect or replace the earlier stage regularly or without change. A marker which corresponds in some features to a marker of an earlier stage, but which differs in one or more features, is said to reflect or replace the marker of the earlier stage with change or innovation. E.g. PPN *eni ‘this, these’ is replaced by PNP *teenei ‘this’ and *(ee)nei ‘these’, with innovations exhibited in meaning and distribution as well as in form (see 3.2.3).
3. Nuclear Polynesian and Tongic.
3.1 Certain morphological features shared by NP languages exclusive-of TO and non-Polynesian languages are noted here. Exclusively shared features which can be established as reflecting innovations of PNP are cited in (1) to (6); exclusively shared features which cannot be shown to reflect innovations of PNP (though they may in fact do so) are cited in (7) to (14).
The following is a list of PNP innovations. These are discussed in detail in 3.2.
Exclusively shared features cited in (7) to (14) below cannot be established as reflecting innovations of PNP. In each case they are exhibited by markers whose PPN equivalents cannot be reconstructed with precision or certainty. Reconstruction of PPN is difficult in these cases either because cognates of the NP markers are not found in TO or non-Polynesian languages (making it uncertain whether the marker existed at all in PPN) or, where cognate (or structurally equivalent) markers exist in TO, because no cognates are found in non-Polynesian languages (making it uncertain whether the differences exhibited by TO and NP markers is the result of an innovation in PNP, in PTO or both).
PTO *fee ‘time-space interrogative (when?, where?, which?)’ is reflected by TON, NIU fee.
Three Outliers have markers corresponding to TON and NIU: KAP, NUK hee, MAE fee. EUV fea, SIK hea, on the other hand, correspond to NP cognates in exhibiting final -a. TON and NIU final -e can be explained as following the normal pattern of assimilation by which PPN *a becomes PTO *e near either *e or *i. 14 This kind of assimilation occurs in NUK and KAP in a few items (e.g. NUK, KAP mee ‘thing’, replacing PPN *me?a).
PPN *fea rather than *fee is thus suggested by the internal evidence as the more probable reconstruction. External evidence in this case is not decisive. PEO *pei ‘where’ can be reconstructed from the following: FIJ, Nggela, Bugotu vei, Ulawa, Maŋo hei, Kwara'ae fa?i, Efate we, Volow, Gog, Motlav ve, Sa'a tei, Vuras a-ve Mota, Oba vea, Merlav via, Tasiko a-pe ‘where?’. The regular replacement in PPN would be *fei, rather than *fea or *fee.- 45
PNP *laa, *teelaa, and *(ee)laa do not appear to have cognates in TO or in other EO languages. TON, NIU ia ‘that, those (distant from speaker or addressee)’ correspond in function to PNP *teelaa and *(ee)laa (cf. 3.2.3), but probably reflect PEO *ia ‘3rd pers. singular, 3rd pers. demonstrative’.
3.2.1 The claims made in (1) to (6) of 3.1 concerning innovations of PNP will now be discussed in detail.
220.127.116.11 NP languages share forms for ‘2nd person dual’ and for ‘2nd person plural’ which differ in certain features from those common to TON, NIU and most other EO languages; these uniquely shared features of NP languages are accordingly assumed to reflect innovations of PNP.- 46
The reconstructions in question are cited below.
Some discussion of the distribution and morphemic structure of person-markers in PN languages is necessary before these reconstructions can be justified.
18.104.22.168 In many PN languages, e.g. SAM, TON, EUV, MAE, EFU and PIL, and in FIJ (Bauan dialect), at least three sets of person-markers can be distinguished on the basis of formal, semantic and distributional differences. The formal diffierences exhibited in FIJ, TON and SAM between the nuclear, preposed subject, and preposed possessor sets are shown in Table 2.
Person-markers in Tongan, Samoan and Fijian (Bauan).
1st dual exc.
1st trial exc.
1st pl. exc.
1st dual inc.
1st trial inc.
1st pl. inc.
In these three languages and in the other PN languages cited above the nuclear person-markers fill the nucleus slot in the phrase. 16 Occurrence in the nucleus slot marks a person-marker as carrying the meaning component ‘focus’ or ‘emphasis’ (other markers signal whether it refers to actor, possessor, goal, or some other relation).
In contrast to the nuclear person-markers two sets of person-markers occur in FIJ, TON and SAM in the preposed periphery of the phrase. One of these sets marks preposed subject, the other marks preposed possessor.
It is assumed that a similar distinction existed in PPN between nuclear, preposed subject and preposed possessive person-markers. It seems probable, for instance, that in PPN at least three distinct forms occurred carrying the meaning ‘second person singular’: *koe filled the nucleus slot, *ke the preposed subject slot, and *-u/-o the preposed possessor slot (cf. TON and SAM forms in Table 2).
The PPN nuclear person-markers *kim(o)urua and *kim(o)uto(l)u can each be analysed as consisting of a prefix *ki- ‘focus’, a root *m(o)u- ‘2nd person non-singular’ and a suffix*-rua ‘dual’ or *-to(l)u ‘plural’. The corresponding PPN preposed possessive person-markers *m(o)urua and *m(o)uto(l)u differ only in the absence of *ki-.
It is less clear what forms should be reconstructed as filling the preposed possessor slot (see 22.214.171.124).
126.96.36.199 *ki- ‘focus’ is reconstructed in PPN *kim(o)urua and *kim(o)uto(l)u on the basis of the correspondence of TON ki- in kimoua - 48 ‘2nd pers. dual, nuclear’ and kimoutolu ‘2nd pers. pl., nuclear’, with FIJ ke- in kemudrau, kemudou, kemunii (see Table 2 for glosses), with PNP *k- in *ko(u)lua and *ko(u)tou (see 188.8.131.52). PPN *ki- is reconstructed as a prefix to all dual and plural person-markers in the nucleus slot; it is reflected regularly by TON ki- in all cases (see Table 2); it is reflected regularly by SAM ?i-, ECE, TOK, PIL, MAE, 17 WFU ki- and by NUK gi- as a prefix to all first person and third person dual and plural person-markers in the nucleus slot (see Table 2 for SAM, and see (1) of 3.3 for further discussion).
184.108.40.206 *m(o)u- ‘2nd pers. non-singular’ is reconstructed in PPN *kim(o)urua and *kim(o)uto(l)u on the basis of:
220.127.116.11 PPN *m(o)u- is also reconstructed as the root in the corresponding markers in the preposed possessor slot, namely *mo(u)rua and *mo(u) to (l)u, on the basis of:
18.104.22.168 PPN *-rua ‘dual’ is reconstructed in the PPN dual person-markers *kimo(u)rua and *mo(u)rua cited above, on the basis of:
The correspondence of PNP *l with zero in TON and NIU indicates that PPN *-rua rather than *-lua should be reconstructed. The same correspondence is exhibited by the free form numerals, TON, NIU ua ‘two’, SAM, ECE, TOK, FUT, HAW lua ‘two’.- 49
22.214.171.124 PPN *-to(l)u ‘plural’ is reconstructed in PPN *kim(o)uto(l)u and *m(o)uto(l)u on the basis of:
126.96.36.199 PNP *ko(u)lua ‘2nd pers. dual’ and *ko(u)tou ‘2nd pers. pl.’ are reconstructed as occurring in the nucleus slot on the basis of the following reflexes: SAM ?oulua, EFU, ECE, EUV, TOK, PIL koulua, MAO, TUA, RAR, KAP koorua, TIK, EAS, MAE korua, WFU a-kolua, SIK kolua, HAW ?oolua ‘2nd pers. dual’, SAM ?outou, EFU, ECE, EUV, SIK, TIK, TUA, MAO, TOK koutou, HAW ?oukou, RAR kootou, KAP kootou/k(h)outou, WFU a-kautau, MAE kotou ‘2nd pers. plural’.
In SAM, EFU, ECE, TIK, MAE, EUV and PIL the forms cited above occur in the nucleus slot, while partially different forms mark second person dual and plural in preposed slots. In EP languages the nuclear 2nd person markers also fill the preposed possessor slot.
188.8.131.52 PNP *o(u)lua and *o(u)tou are reconstructed as replacing PPN *m(o)urua and *m(o)uto(l)u in the preposed possessor slot, on the basis of the following:
ECE lu, EUV, ulua, SAM, lua, PIL ulua, WFU rua, NUK olu ‘2nd pers. dual, preposed possessive’; ECE outou, EUV utou, SAM tou, TIK otou, PIL utou/tou, WFU rautau, NUK odou 2nd pers. pl., preposed possessive’. (See also the arguments presented in 184.108.40.206).
220.127.116.11 In the preposed subject slot PNP *(kou)lua ‘2nd pers. dual, preposed subject’ and *(kou)tou ‘2nd pers. pl., preposed subject’ are reconstructed as replacing, respectively, PPN *(mou)rua and *(mou)to(l)u. These reconstructions are made, tentatively, on the basis of the following:
It can be seen that, because of the diversity of forms within PN, precise reconstruction of PPN and PNP in this case is not possible. In SAM, EUV, and PIL (and in FIJ) second person non-singular in the preposed subject slot is marked by zero preceding a number marker. In MAE and EFU it is marked by ko-; in TON it is marked by mo-. Comparable external evidence (except for FIJ) is lacking.
18.104.22.168 As noted above, correspondences between TON and NIU and non-Polynesian languages, especially FIJ, indicate that PPN - 50 *kim(o)u- ‘2nd pers. non-singular nuclear’ and *m(o)u- ‘2nd pers. non-singular, preposed possessive’ should be reconstructed. Agreements between NP languages, on the other hand indicate that PNP *ko(u)- should be reconstructed as replacing PPN *kim(o)u- and *m(o)u-, respectively. The change could have resulted from the loss *m from the PPN root, or could reflect the innovation of a completely new root *ko(u)- or *o(u)-. One possible explanation of the development of the PNP forms is the following: 19
At some stage shortly before the break-up of PPN (let us call this stage pre-PPN) the form *-mu or *-mo occurred in the preposed possessor slot marking second person singular. Pre-PPN *-mu/-mo is attested by cognates in almost all closely related languages e.g. FIJ, Nggela, Bugotu, Vaturanga -mu, Oba mu -m, Efate -ma ‘2nd person singular preposed possessive’. In PPN, however, *-u/-o replaced *-mu/-mo, with loss of *m. This is attested by TON, NIU, MAO, TOK -u/-o, SAM, KAP, EUV, TIK -u ‘2nd pers. sing., preposed possessive’.
Now if, at the break-up of PPN, *mu or *mo still existed as remembered or stylistic variants of *-u or *-o ‘2nd pers. sing., preposed possessive’, then the stage would have been set for the analogous loss of *m from *mou- in -kimo(u)rua and *mo(u)rua, etc. At any rate, for this or for some other reason, this further loss is presumed to have occurred in PNP (but not in PTO). Loss of *m in the PNP markers would have given rise to forms *kio(u)lua and *kio(u)tou in the nucleus, and *o(u)lua and *o(u)tou in the preposed possessor slot. It can be supposed that *kio(u)lua and *kio(u)tou assimilated to *ko(u)lua and *ko(u)tou, which are the forms reflected by all NP languages in the nucleus slot.
As noted in 22.214.171.124 above, *o(u)lua and *o(u)tou are reflected by many but not all NP languages in the preposed possessor slot. Following the break-up of PNP it can be argued that in certain languages (notably EP) reflexes of *ko(u)lua and *ko(u)tou came to be used instead of - 51 *o(u)lua and *o(u)tou in the preposed possessor slot as well as in the nucleus slot. Typically, this situation is reflected by those PN languages which do not distinguish a set of preposed subject person-markers; in these languages the pattern pressure to maintain distinct series of forms as preposed and nuclear person-markers would have been diminished.
3.2.3 PPN demonstratives *eni ‘this, these’, *ena ‘that, those (near addressee)’ are reconstructed from:
PEO *eni or *ini ‘this’ and *ena or *ina ‘that’ can be reconstructed on the basis of reflexes in Nggela, Vaturanga, Bugotu, Sa'a, Ulawa, Cakaudrave and Fagani. It can be assumed that PPN reflected the PEO forms as *eni and *ena, respectively, on the basis of the TON, NIU and EUV reflexes cited in (a) above.
In TON and NIU the demonstratives occur either in the nucleus slot, as TON ko eni ‘this one, these ones’ or post-posed, as TON ko e fale eni ‘this house, these houses’. Demonstratives in the EO languages cited in (c) above appear to have more or less the same distribution; thus it seems probable that the PPN demonstratives should also be reconstructed as occurring either in the nucleus slot or postposed to the nucleus.
It is assumed tentatively that PPN *eni and *ena should be reconstructed as not distinguishing number. Few EO languages distinguish number in the demonstratives (as far as can be discerned from the descriptive data available) and none appear to do so in a way comparable to NP languages (see below and 3.2.7) ; TON, NIU (and EUV) demonstratives resemble those of most EO languages in not distinguishing number in the demonstratives and can be assumed to reflect PPN in this. 20
If these assumptions regarding PPN demonstratives are correct the PNP reflexes exhibit a number of innovations. TON, NIU and EUV - 52 demonstratives are not readily analysable; demonstratives in NP languages can be analysed as consisting of an article plus a position marker, e.g. MAO tee-nei ‘this’ where tee- is a variant of te ‘definite article singular’ and -nei equates with the position marker nei‘near speaker’ and ee-nei ‘these’ where ee- can be regarded as a variant of ŋaa ‘definite article plural’. In SM languages the definite article plural in the demonstratives (and elsewhere) is marked by zero in the article slot e.g. TOK ko tee-nei mea ‘this thing’, ko nei mea ‘these things’ (cf. 3.2.7). It must be supposed that the differences between the EP and the SM plural demonstratives is the result of different treatment in PEP and PSM of the singular demonstratives. In PSM *teenei and *teenaa were treated as *tee- and *tee-naa, and the plurals were formed as *ø-nei and *ø-naa, with ø-making definite article plural here as in all other environments (cf. 3.2.6). In PEP, however, the singular demonstratives were treated as *t-eenei and *t-eenaa with plurals *eenei and *eenaa, perhaps on the analogy of the possessive pronoun where *t- and *ø- mark singular and plural respectively (see 3.2.7).
It is notable that in those NP languages where vowel length data is supplied, the initial element in the singular demonstrative is tee- (or corresponding form) with long vowel, though in each case the definite article singular in these languages is te (or corresponding form) with short vowel. It may be that in PNP the singular demonstratives were formed by the addition of *te ‘definite article singular’ to the PPN demonstrative *eni and *ena, resulting in PNP *teenei and *teenaa. Alternatively, they can be explained as a combination of *te ‘definite article’ and *nei or *naa, with *nei reflecting PPN *eni with metathesis.
In NP languages demonstratives can occur in the nucleus slot as MAO ko teenei, SAM ?o leenei ‘this one’, or preposed to the nucleus as MAO ko teenei fare, SAM ?o leenei fale ‘this house’. This preposed occurrence probably represents an innovation of PNP (cf. discussion above of distribution of PPN demonstratives).
3.2.4 PPN *ni ‘near speaker in time and space’ is reconstructed as postposed position-marker on the basis of the following reflexes:
In all PN languages the functional replacement of PPN *ni occurs as a member of a paradigmatic set of postposed markers indicating position in time or space. At least one other such marker can be reconstructed for PPN, namely, *na (reflected by TON, NIU, MAE, na, SAM, RAR, MAO naa) ‘near addressee in time or space’, as in TON ?i Togatapu ni ‘here in Tongatapu’, MAO kei Niu Tiireni nei ‘here in New Zealand’, TON ho nima na, MAO too riŋa naa ‘your hand there’. (Cf. (14) in 3.1).
It appears likely that PNP *nei ‘near speaker in time or space’ replaces PPN *ni in function and distribution, but formally reflects PPN - 53 *eni ‘demonstrative, near speaker’ (with metathesis), rather than PPN *ni.
TON and EUV eni, NIU nei/nai ‘demonstrative, near speaker’ substitutes for ni ‘near speaker’ in some postposed environments, carrying similar though contrastive meaning. This was probably the case in PPN as well (cf. remarks on the distribution of the PPN demonstratives in 3.2.3).
nei/nai and na occur in NIU as members of the demonstrative set, where they replace PPN, PTO *eni and *ena (3.2.3.). NIU nei can be explained as representing metathesis of PPN *eni (independent of the metathesis posited for PNP above). NIU reflects PPN *ni and *na, post-posed position-markers, by enclitic ni and na, as in TON.
PNP *nei is reconstructed as having an additional privilege of occurrence not exhibited by PPN and PTO *ni, namely, occurrence preposed to the nucleus (1) following possessive pronouns, on the basis of e.g. MAO ko tooku nei fare, SAM ?o lo?u nei lava fale ‘my own house’ (2) following articles as the initial morpheme in demonstrative markers (see 3.2.3 and 3.2.7).
3.2.5 PPN *ha ‘indefinite article’ is reconstructed from:
The indefinite articles found in NP languages do not reflect PPN *ha. PSM *se ‘indefinite article singular’ and *ni ‘indefinite article plural’ are reconstructed from the following:
PEP *he ‘indefinite article’ is reconstructed from MAO, TUA, HAW he, RAR, MVA ?e, TAH e ‘indefinite article’.
It is notable that SM and certain Outlier languages distinguish a singular and a plural indefinite article, but that TON, NIU and EP languages do not. In TON, NIU and EP the indefinite article can refer to a single object as TON ha tohi ‘a letter’, a number of countable objects, as TON ha tohi ?e nima ‘five letters’, or to an amount of a mass (non-countable) noun, as TON ha vai ‘some water, any water’. In SM languages the first two of these functions are divided between a singular and a plural indefinite article. In some languages an additional article exists marking quantity of a mass noun, contrasting with the other indefinite articles as in the following: SM se tusi ‘a letter’, ni tusi ‘some letters’, sina suka ‘some sugar’.
It is possible that the distinction between a singular and a plural indefinite article represents an innovation of PSM rather than of PNP. It - 54 can be argued, however, that the PSM articles, *se and *ni both reflect PNP equivalents. This is suggested by the fact that *se and *ni are also reflected by several Outliers: MAE se, NUK, SIK he ‘indefinite article singular’, NUK, PIL, SIK, KAP ni ‘indefinite article plural’. TON ni?ihi ‘to be several’ (a base) is a probable cognate of PNP *ni. Cf. also EUV hina ‘indefinite article singular’, ni-/ni?ihi ‘indefinite article plural’ he ‘indefinite article’, MAE a ‘indefinite article plural’.
TON and NIU ha ‘indefinite article’ reflect PEO *(n) sa ‘one’. PNP *se and *ni, however, appear to have no cognates in non-Polynesian languages. The simplest explanation of these facts seems to be that which reconstructs PPN *ha ‘indefinite article’, reflected regularly by TON and NIU, and which assumes that the PNP indefinite articles *se and *ni represent innovations.
3.2.6 PPN *taha ‘one, a certain one’ is reconstructed from the following:
Initial *ta- in PPN *taha is attested by *ta- in reflexes of both PTO *taha and PNP *tasi. TO and NP languages disagree, however, as to the form of the final element (TON and NIU final -ha, NP languages reflect final *-si).
Final *-ha in PPN *taha ‘one’ is reconstructed on the basis of TON, NIU taha, Willaumez tara and Nakanai sasa ‘one’, and the arguments presented below.
PPN *ha- ‘one’ is reconstructed as a variant of *taha occurring prefixed in the numerals for ‘ten’ and ‘one hundred’ on the basis of the following evidence:
(d) PEO *(n)sa-ŋapulu, PPN *ha-ŋafulu ‘a unit of ten’, PEO *(n)sa-ŋarau, PPN *ha-ŋarau ‘one hundred’: Sa'a e ta ‘one’, ta-ŋarau ‘one hundred’, Vaturanga ke-sa ‘one’, sa-ŋavulu ‘ten’, sa-ŋatu ‘one hundred’, Ilocano may-sa ‘one’, sa-puloh ‘ten’, sa-ratus ‘one hundred’; TON, NIU, EUV hogofulu ‘ten’, te au ‘one hundred’; SAM a-gafulu, TIK a-gafuru, 23 EFU kau-a-gafulu, MAE ŋafulu, TAH ?ahuru, MQA ?ogohu-u, MVA rogo?uru, MAO ŋahuru ‘ten’, SAM a-galau ‘a hundred’, lua-galau ‘two hundred’, TON, NIU ua-geau ‘two hundred’.
It must be supposed that PEO had the marker *(n) sa ‘one’ as a productive morpheme occurring both as a free form (as noted in 3.2.5) and prefixed in the numerals for ten and one hundred. This situation is reflected by several EO languages e.g. Sa'a where ta occurs as the free - 55 form numeral ‘one’ and as a prefix in the words for ten and one hundred (see (d) above). It can be assumed that, similarly, *ha occurred in pre-PPN as the free form numeral ‘one’ and *ha- occurred as a prefix in the words for ten and one hundred (see (d) above, and cf. 3.2.5).
Later, in PPN, *ha was for some reason replaced as the productive morpheme for one by *taha; *ha- was retained in the PPN numerals *ha-ŋafulu ‘ten’ and *ha-ŋarau ‘one hundred’, and *ha was retained as a free form marking ‘indefinite article’ (see 3.2.5). PNP *tasi is assumed to be an innovated form replacing PPN *taha; final *-si in *tasi can perhaps be related to the innovation of PNP *se replacing PPN *ha ‘indefinite article (see 3.2.5).
3.2.7 NP languages share certain formal and structural features exclusively of TON and NIU in those markers of nominal constructions which, in the grammars of PN languages, are loosely called ‘articles’.
All EO languages possess several construction-markers (generally preposed) which mark the phrase of which they are a member as a nominal, locational, personal or verbal phrase, e.g. in FIJ, na marks a phrase as nominal, i.e. non-verbal, non-locational, non-personal, etc), as na vosa the language(s)', ko marks a phrase as personal (i.e. non-nominal, non-verbal, etc.) as ko viti ‘Fiji’, sa marks a phrase as verbal, as sa lako ‘go’, etc. 24
Equivalent (not always cognate) markers exist in PN languages, As in EO languages, these markers are mutually exclusive, occur preposed to the nucleus of the phrase, and distinguish nominal, locational, personal, and verbal phrases. They may be called nominal, locative, personal and verbal articles. In TON there are four nominal articles, e/he ‘definite’, ha ‘indefinite’, ti?i ‘diminutive definite’, and ti?a ‘diminutive indefinite’ and these contrast with a personal article ?a, a locative article (represented by zero in the article slot), and several verbal articles (tense-aspect markers), as exemplified by the phrases ki he fale ‘to the house(s)’, ki ha fale ‘to a/some house(s)’, ki ?a pesi ‘to Bessy’, ki Toga ‘to Tonga’, na?e ?alu ‘went’.
All PN language appear to distinguish at least two nominal articles, a definite and an indefinite, although in many cases they distinguish more than two. (SAM, for instance, has seven nominal articles, EUV has six). 25 Most NP languages, as noted below, make a further distinction between singular and plural definite articles. Neither TON, NIU nor other EO languages distinguish number in the articles, and it seems probable that such a distinction represents an innovation of PNP.
PNP *te ‘definite article singular’ is reconstructed from MAO, RAR, MQA, MVA, TUA, TAH, EAS, ECE, TOK, TIK, te, HAW ke - ka, SAM, FUT le. Cf. also EUV, SIK te, MAE re, KAP ti, NUK de, WFU te /ta /ti. All these languages exhibit a non-isolable variant in the article - 56 slot in preposed possessive pronouns (cf. 3.2.3), consisting of the initial consonant of the free form (as MAO t-, SAM l-), and permitting reconstruction of PNP *t-, in this environment as a variant of *te.
PNP *ŋaa or *na ‘definite article plural’ is reconstructed from MAO, TIK, EAS ŋaa, RAR, ŋaa, HAW, TAH, MVA na ‘definite article, limited plural (a few)’. Cf. also SIK na, WFU, MAE ŋaa ‘definite article plural’, SAM naai ‘diminutive article plural’. In these languages the singular and plural articles are mutually exclusive and substitute for each other, as MAO te fare ‘the house’, ŋaa fare ‘the houses’. A zero variant of the free form definite article plural fills the article slot in preposed possessive pronouns in all NP languages, where it substitutes for the variant of the definite article singular noted above, as MAO tooku fare ‘my house’, ooku fare ‘my houses’.
Possible cognates of PNP *ŋaa or *na occur in TON, NIU and non-Polynesian languages. TON gaahi ‘unlimited plural’, NIU na‘plural’ occur following nominal articles as one of a set of several markers indicating kinds of plurality as TON ha gaahi me?a ‘some things’; (these number-marking post-articles are paralleled in many NP languages, e.g. RAR, TUA ŋaa, NUK ŋa ‘two or three’, as RAR te ŋaa metua ‘the (two) parents). na occurs in many EO and Oceanic languages as a preposed nominal construction marker, and FIJ na (see above) can be assumed to reflect PEO *na ‘nominal article’. There is little evidence other than this and the occurrence of na rather than ŋaa as the definite article plural in TAH and MVA to suggest that *na occurred in PPN, as distinct from *ŋaa.
The agreements between TON, MAE, NUK, WFU and NP languages, however, suggest the following as the history of PNP *ŋaa. PPN *ŋaa(hi) is reconstructed as a post-article, carrying the meaning ‘two or three’ or ‘a few’. It is reflected in this function by TON gaahi, NUK, RAR, TUA ŋaa. SAM naai ‘diminutive article plural’, as ni naai mea ‘some few things’, is a possible reflex. PNP inherited *ŋaahi as ŋaa in this same function, but at the same time came to use ŋaa in a second function, as a plural indefinite article substituting for PNP *te ‘definite article singular’. ŋaa in this second function is reflected by TIK, WFU, MAE, HAW, RAR, SIK, EAS and MAO.
NIU e, and TON e/he (cf. fn. 28) ‘definite article’ cannot be established as having any cognates in NP languages. PNP *te exhibits only a partial formal correspondence. It does seem probable, however, that PPN had only one definite article, and this can be tentatively reconstructed as having the shape *(t)e. 26 That PPN did not distinguish a singular and a plural definite is suggested by the fact that neither TON nor NIU makes such a distinction, that most EO languages have only a single nominal article, and that while NP languages possess both a singular and a plural definite article, plurality in these languages is frequently marked by an alternative technique in which a post-article indicating one of several kinds of plurality follows the singular definite article, as SAM le ?au talavou ‘young people’, MAO te huŋa mate ‘the dead’, RAR te au mea - 57 ‘the things’. This is the technique used in TON and NIU, though indication of number is optional in these two languages.
SAM, ECE, EFU and TOK mark definite article plural by zero contrasting with the occurrence of other articles in the article slot, as SAM ?o le fale ‘the house’, ?o fale ‘the houses’, lo?u fale ‘my house’, o?u fale ‘my houses’. The occurrence of zero as the mark of the definite article plural in all environments (as opposed to the article slot in preposed possessive pronouns—see below) can be regarded as representing an innovation of PSM.
It may be noted that in NP languages (but not in TON and NIU, nor in other EO languages) singular and plural number are distinguished not only in the articles but in the demonstratives (3.2.3), the preposed possessive pronouns, and in certain other markers, by prefixing the nominal articles to the following morphs. For example, PNP *tee-fea ‘which?’ (singular) and *(ee)fea ‘which (plural)’: MAO, TOK teehea, ECE teefea, SAM leefea, EFU lefea, WFU tefe ‘which? (singular)’, MAO eehea WFU efe, SAM, ECE, EFU ø-fea, TOK ø-hea ‘which? (plural)’, as MAO teehea fare ‘which house?’, eehea fare ‘which houses?’. TON and NIU, like certain other EO languages, have a monomorphemic interrogative which, when it follows a nominal phrase, can be translated ‘which?’ (without number distinction), as TON, NIU ko e fale fee? ‘which house(s) ?’.
As noted above, in NP languages the initial morpheme in the demonstratives, possessive pronouns, etc. marking number, can be equated with the nominal articles, as for example the following SAM preposed possessive markers: lo?u ‘1st pers. singular possessive, definite singular’, o?u ‘1st pers. possesive, definite plural’, o?u ‘1st pers. sing. possessive indefinite singular’, nio?u 1st pers. singular possessive indefinite plural’, sio?u ‘1st pers. singular possessive, diminutive singular’, etc. where l-, ø-, s-, ni-, si- can be regarded as variants of the free form articles le, ø, se, ni, si (cf. fn. 25) is less clearly the case in TON and NIU, where the demonstratives and fee ‘which?’, for instance, appear not to be analysable in this way.
In any case, the manner of distinguishing singular and plural number in NP languages in nominal articles, demonstratives and preposed possessive pronouns is not paralleled in TO and EO languages, and is assumed to reflect an innovation of PNP.
3.3 The following morphological features can be established as innovations of PTO, 27 or are shared by TON and NIU exclusively of NP and non-Polynesian languages:
(1) TON and NIU share an intrusive -u- in three plural person markers filling the nucleus slot: PTO *kitautolu ‘1st person inclusive plural’ (TON kitautolu, NIU tautolu); *kimautolu (TON kimautolu, NIU mautolu) ‘1st person exclusive plural’, and *kilautolu (TON kinautolu, NIU lautolu) ‘3rd person plural’ are reconstructed as filling the nucleus slot.- 58
These PTO forms reflect PPN *kitato(l)u ‘1st person inclusive plural’, *kimato(l)u ‘1st person exclusive plural’ and *kilato(l)u ‘3rd person plural’ as nuclear person-markers with innovation of *-u-. These PPN reconstructions are made on the basis of the TON and NIU markers and the following:
(2) PTO *-tolu ‘plural suffix to person-markers’ corresponds to PNP *-tou, and to FIJ (Bauan) -tou/-dou, Nakanai -tou/-teu, and Raga, Mota, Gog -tol ‘trial’, and reflects either PPN *-tou or *-tolu (see 126.96.36.199).
(3) Several cases of grammatical markers in which PPN *a has become PTO *o or *e (cf. fn. 14). For example:
(4) PTO *heni ‘here’, *hena ‘there (near addressee)’, *hee ‘specified location’ are reconstructed from corresponding TON, EUV heni, NIU hinei ‘here’, TON, EUV hena, NIU hina ‘there’, TON, EUV, NIU *hee ‘specified location’ are reconstructed from corresponding TON, EUV heni, NIU hinei ‘here’, TON, EUV hena, NIU hina ‘there’, TON, EUV, NIU hee ‘specified location’.
These location markers differ from the PTO demonstratives only in the presence of initial *h-. In morphemic analysis this initial *h- can be - 59 interpreted either as a location marking morpheme preceding the demonstratives *-eni, etc., or as a morpho-phonemic variable in the morpheme *heni/eni, etc. (*h- being manifested when the morpheme occurs in locative environments). 28
The corresponding PPN location-markers cannot be reconstructed with certainty from the internal PN evidence. PNP *konei ‘here’, *konaa ‘there (near addressee)’, *koo ‘yonder’ can be reconstructed from the following: MAO konei, EFU kunei, SAM ?inei, KAP kinei, TOK koonei ‘here’, MAO konaa, EFU kuna, SAM ?inaa, TIK kona, TOK koonaa, NIU kunaa ‘there (near addressee)’ MAO, TOK koo, SAM ?oo ‘yonder’, WFU kuu ‘here’.
(5) PTO *ee ‘emphatic demonstrative this, these, that, those’ is reconstructed from corresponding TON, NIU, EUV ee, as TON ko e fale ee ‘this, these house(s)’ as a member of the class of demonstratives also containing *eni and *ena (see 3.2.3). The semantic correspondence of *ee is with PNP *teeia ‘emphatic demonstrative singular’ (as reflected by SAM lea, EFU leia, RAR teeia, TAH teeie, HAW keeia) and *(ee) ia ‘emphatic demonstrative plural’ (reflected by SAM, EFU ø-ia, RAR eeia). Formal correspondence, however, is probably with PNP *ee ‘antecedental pronominal marker (SAM, EFU lee, ee as SAM ?o lee ?ua alu ‘the one who has gone’, ?o ee ?ua oo ‘those who have gone’).
TON and NIU also share certain features cited in (1) to (6), and in (8), (10) and (15) of 3.1.1 exclusively of NP languages. If the claims made there are correct (1) to (6) consist of cases in which PTO markers replace PPN without change, while in other cases the PPN reconstruction is uncertain.
4.3 Morphological and phonological evidence for the SM subgroup posited in 2.1.4 is less impressive than the evidence for EP. This may be due in part to the fact that, of the 5 languages included in SM, only SAM has been described in detail. The following morphological features are shared by Samoic languages exclusively of Tongic, EPN and non-Polynesian languages, and are assumed to represent innovations of PSM.
EP languages reflect PEP *he ‘indefinite article’, and, like TON and NIU, do not distinguish a singular and a plural indefinite article. It is possible that such a distinction in SM languages represents an innovation of PSM, and that PNP *se should be reconstructed as marking ‘indefinite article’ rather than ‘indefinite article singular’ (but cf. 3.2.5).
1 I am indebted to Professor Bruce Biggs and Mr. D. S. Walsh for valuable criticisms of an earlier draft of this paper.
2 The apices of the Polynesian Triangle are Hawaii, New Zealand, and Easter Island.
3 The term Nuclear Polynesian was suggested to me in conversation by Professor S. H. Elbert.
4 Dyen, 1965: 34 uses the term Tongic to refer to a hesion within his West Polynesian Cluster comprising Tongan, Uvean, and Niue, and based on his lexicostatistical comparisons.
5 Elbert 1953, p. 169, Table 4. He concluded that, within the 16 Polynesian languages he classified, Tongan, Niue, Uvean and Futunan form a major subgroup, co-ordinate with a second subgroup which includes all the remaining well described languages spoken within the Polynesian Triangle. Although in his paper Elbert is not completely explicit on this point it is clear that this part of his family tree is based on his comparative study of Polynesian phonology, and not on his lexico-statistical results. He notes (pp. 163-164) that the 12 languages he derives from P-Sa-O-E share several innovations in their treatment on PPN *?, *h, *s, *r, and *l (see PNP reflexes 3.2.2, Table 1, of the present paper). The phonological grounds for the subgroup he derives from P-To (consisting of Tongan, Uvean, Niue and Futunan), are weaker. Of these languages only Tongan and Niue can be clearly shown to share any phonological innovations (see fn. 27).
In another part of his paper Elbert concluded that (contrary to his phonological study) his lexico-statistical results suggest “that West and East Polynesian had become distinguished before the differences developed among languages in either area” (p. 158). In fact, however, his lexico-statistical results support the subgrouping proposed in his family tree at least as strongly as they do an East-West division. For instance, Tongan shares 45 - 58 per cent, and Niue 47 - 56 per cent of cognates with the eight Eastern Polynesian languages in his sample; but Samoan (with 52 - 67 per cent), Ellice (61 - 67 per cent) and Tikopia (62 - 71 per cent) share consistently 10 per cent more cognates with these same Eastern languages than does either Tongan or Niue. Futunan, in lexicon as well as in grammar and phonology, seems to group with Nuclear Polynesian (and Samoic, see 2.1.4) rather than with Tongic; Futunan ? sometimes corresponding with Tongan ? can be explained as a borrowing. The position of Uvean is not clear (see 2.6.1 and fn. (8)).
6 Data for SAM, ECE and TOK was obtained in Auckland from native-speaking informants.
7 I have not sub-classified EAS beyond its inclusion in NP, partly because descriptive material for EAS was not available for checking during the later stages of preparation of this paper. Elbert includes EAS in his Eastern Polyneasian. The recent discovery that EAS ? reflects PPN ? * ? suggests that the separation of EAS pre-dates the break-up of the other Eastern Polynesian languages.
8 The fact that Uvean (EUV) reflects PPN *?, *r and *h inconsistently, sometimes agreeing with SM, sometimes with TO, points to borrowing. There are a number of doublets in EUV, e.g. hifo, ifo ‘downwards’, hake, ake ‘upwards’, lua ‘two’ (free form), ua- ‘two’ (in compounds, e.g. uafulu ‘twenty’)'. EUV possesses some morphological features otherwise unique to TON and NIU, e.g. the demonstratives eni, ena, ee (see 3.2.3), hogofulu ‘ten’, te au ‘one hundred’ (see (3) in 3.3), the locatives heni, hena, hee (see 3.3 (4)). On the other hand, it has many features which are otherwise shared by NP languages exclusively of TON and NIU, here regarded as reflecting innovations of PNP, and cited in 3.2.
In Elbert's study Uvean (EUV) shares 86 per cent of basic vocabulary with TON, 72 per cent with NIU, and 70 - 78 per cent with SAM, ECE, TIK and EFU. In grammatical structure it shares more features of restricted distribution with NP and especially with SM languages than with either TON or NIU. There are at least three possible explanations of these facts: (a) that EUV is a SM language which has been extensively influenced by borrowing from TON (b) that EUV is a TO language which has been even more extensively influenced by a SM language or languages (c) that the language ancestral to EUV split off early from the languages ancestral to TO and SM, but remained in contact with and was influenced by their descendants.
9 Grace 1955, 1959, 1961, Biggs 1966.
10 Biggs 1966. Table 1, p. 384.
11 Some markers are discontinuous, e.g. MAO e . . . ana ‘progressive tense-aspect’ (see 4.2 (10)), and some markers are not morphemes at all in the usual sense but are realized in structural relations between units such as order. E.g. in SAM, the absence of a morpheme in the article slot in nominal phrases is the mark of the definite article plural (cf. 3.2.6). Some morphemes classified as bases by Biggs 1961, and Carroll 1965, can be regarded as markers in the present sense because they belong to small closed substitution classes, e.g. certain bases with locative and temporal meaning such as MAO runa ‘above’, inanahi ‘yesterday’, and the nuclear person-markers.
12 MAO base classes are described in Biggs 1961, pp. 23-27.
13 Such a descriptive model is suggested in Biggs 1960, and is elaborated in Biggs 1961, Pawley 1961 and n.d., and Carroll 1965.
14 Cf. Elbert 1953, p. 153-4 who, however, sets up two extra PPN vowel phonemes, *œ and *O, in addition to *a, to account for correspondences such as TON a = SAM a, TON e = SAM a, TON o = SAM a.
15 As noted in 3.2.7, phrases in PN languages can be classified as nominal, verbal, locational, personal, etc. according to the presence or absence of certain preposed construction-markers.
16 The nucleus slot is normally filled by a base or bases, and is bounded by the adjacent peripheral markers. For example, in SAM, the nuclear person-markers have a distribution similar to that of personal name bases and ai ‘who’.
17 MAE ki- appears in non-singular person-markers in texts cited in Capel 1962 (p. 52), although Capell does not list it in his grammatical section.
18 Grace 1959, pp. 44-5 points out that only FIJ and PN of Austronesian languages possess a prefix to non-singular person-marker roots corresponding to ki- or ke-. PAN *ki(n)ta ‘1st person inclusive plural’,*kami ‘1st person exclusive plural’, *kamu ‘2nd person plural’, *siDa ‘3rd person plural’ apparently became Proto-FIJ-PN *ki(n)ta-, *kima-, *kimu, and *(k)ira- as a result of a change by which the initial CV of each root was treated as a prefix of the shape *ki- or *ke-, with metathesis of *kami giving *kima- (cf (1) in 3.3).
19 Several other explanations have been considered but found less economical (i.e. require the postulation of more changes) than the one proposed here. Grace (personal communication) has proposed an explanation of the development in many PN languages of forms such as koulua and koutou which (rephrased in our terms) is as follows: pre-PPN reflected Pan *kamu ‘2nd person plural’ as *kamu- in *kamurua ‘2nd person dual’ and *kamuto(l)u ‘2nd person plural’. Later, for some reason, *m was lost from the root *kamu-, giving rise to the forms *kaurua and *kauto(l)u, which, by assimilation, became *kourua and *kouto(l)u.
An alternative theory derives the root kou- in forms such as koulua and koutou from Proto-Austronesian *kaw ‘2nd person singular’, with *kou- in these languages replacing the reflex of PAN *kamu- and taking on the new meaning ‘2nd person non-singular’.
In both of these explanations it is necessary to suppose that the innovation of *kou- replacing *kamu- as the second person non-singular marker occurred not in PPN, but in PNP, and to suppose that PPN in fact reflected *kamu- in the nucleus slot (while reflecting *-mu- or some such shape in the preposed possessive slot). This is necessary in order to account for the TON and NIU reflexes. Following the first of the two above theories, it must be assumed that PPN had *kamurua and *kamuto(l)u as the nuclear variants, with the root *kamu- becoming TON *kimou- and NIU mu- (the innovation of TON ki- being on the analogy of the prefix ki- before other dual and plural person-markers, (cf. (1) in 3.3 and fn. 18).
Such a theory takes no account of the close correspondence between TON kimou- and FIJ kemu- ‘2nd person non-singular. nuclear’, and between TON mo- and NIU mu-, and FIJ mu- ‘2nd person non-singular, preposed possessive’. Rather than explain these correspondences as convergent developments, it is simpler to reconstruct Proto-FIP-PN *kimu- and *mu- (or similar shapes), as the 2nd person non-singular, nuclear and preposed possessive markers, respectively, and to derive the TON, NIU and FIJ reflexes from these with a minimum of changes.
In these explanations we are still left to account for forms in SM and several other PN languages reflecting *o(u)lua and *o(u)tou in the preposed possessor slot (see 188.8.131.52 and 184.108.40.206).
20 Some Oceanic languages distinguish singular and plural in the demonstratives by a preposed plural-marker (not also functioning as a nominal article), as Bugotu iaeni ‘this’, ira-eni ‘these’, ianeni ‘that’, ira-neni ‘those’ (where ira- marks plural), or by a postposed plural marker, as Mota iloke ‘this’, iloke nan ‘these’.
21 EFU and EUV ni- occur in the article slot in possessive pronouns, as EFU nioku fale ‘some of my houses’.
22 The significance of the Nakanai and Willaumez cognates cannot be judged until deatiled descriptions of these languages become available, but Goodenough, 1961, gives some evidence suggesting that these languages are closely related to PN and Fijian. The correspondences with PPN *taha appear to be regular.
23 Williams 1926-27 lists TIK gafuru ‘ten’ but the form agafuru appears several times in texts quoted in Firth 1961.
24 Comparable preposed construction-markers occur in many other Austronesian subgroups, e.g. Tagalog ang ‘nominal (i.e. non-personal) focus’, si ‘personal focus’, sa ‘nominal location’, kay ‘personal location’, ng ‘nominal non-focus’, ni ‘personal non-focus’, as ang Maynila ‘Manila’, sa Maynila ‘at/to Manila’, si Hose ‘Joseph’, kay Hose ‘at/to Joseph’, etc.
25 Namely, SAM le EUV te ‘definite article, singular’, SAM, EUV Ø ‘definite article plural’, SAM se EUV hina ‘indefinite article singular’; SAM ni, EUV ni/ni?ihi ‘indefinite article plural’; SAM si ‘diminutive article singular’, naai ‘diminutive article plural’, EUV ti/ti?i, diminutive article’; SAM sina ‘quantitative article’, EUV he ‘indefinite article’.
26 Gilbertese te ‘definite article’ can be accounted a borrowing from a PN source.
27 TON and NIU share the following several phonological features exclusively of NP languages; PPN *r is reflected by ø, PPN *l by l; PPN *h and s are both reflected by h; PPN *a becomes TON, NIU o near o or u, PPN *a becomes TON, NIU e near e or i (cf. fn. 14).
28 The TON definite article e/he exhibits comparable variation. The variant he occurs in certain ‘locative’ environments, e.g. after ki ‘motion towards position’, as ki he fale ‘to the house(s),’ where e occurs elsewhere. Cf. Churchward, C. M. p. 151.
29 Elbert 1953, p. 154 points out that EP languages (and some Outliers) united PPN *f and *s in some or all positions. Haudricourt 1964, p. 389 notes that all EP languages (except EAS) reflect PPN *f . . . f by v . . . h (or corresponding phonemes), e.g. PPN *fafine, MAO wahine ‘woman’, PPN *fafie, MAO wahie, PPN *fafo, MAO waho ‘outside’, etc.
30 PEP *noo, *naa is probably derived from combination of an earlier *ni ‘possession’ with *aa ‘dominant possession’ and *oo ‘subordinate possession’, respectively. This is suggested by the occurrence in NUK and WFU of forms such as NUK nioogu, WFU nioku ‘mine’, paralleling PEP *nooku, etc, and by the existence of ni as a free form marking possessive or modifier relation in FIJ and several other EO languages.