Chapter 2.8 Talking about space: terms of location and direction

Malcolm Ross

1. Introduction⇫

Talking about space is a part of talking about the environment as a whole. We include talking about space in this volume because some of the terms reconstructed in §2 are also used to denote parts of the landscape and seascape reconstructed in Chapters 3 and 4. However, much of the terminology reconstructed in this chapter was also used to talk about space in relation to manufactured objects (vol. 1), to flora and fauna (vols. 3 and 4) and to human beings (vol. 5). Many languages have complex terminologies for talking about space, and the length of this chapter bears witness to the fact that Proto Oceanic was no exception.

Following Levinson (1996) and Hyslop (2001), we distinguish four kinds of semantic system employed in talking about space. These are listed here with some commentary on their expression in Oceanic languages.

1. A system of geographic directions based on a division of the environment that normally has a vertical (‘up’/‘down’) axis and a horizontal axis. On the horizontal axis European languages have the cardinal directions ‘north’, ‘south’, ‘east’ and ‘west’. In Oceanic languages there are usually two sets of geographic directions, one used on or near land, the other used at sea (cf. Hill 1997, François 2003, 2004a). Typically each has two subsystems.
   1. The land-based subsystems are (cf. Ozanne-Rivierre 1997):
      1. one with an inland/seaward axis, sometimes with a transverse axis pointing left and right along the coast (each axis is typically denoted by a local noun in a local construction; p.232);
      2. one based on a river valley with an up/down axis (often using the vertical terms) and a transverse axis with one directionally neutral (‘across the valley’) term (each axis is typically expressed by a directional verb or other directional morpheme).
   2. The sea-based subsystems both refer to a northwest–southeast axis,
      1. one using the terms for the northwest storm wind and the southeast trade wind (POc *apaRat and *raki respectively; Ch. 5, §4.2);
      2. the other applying the river valley subsystem with its up/down axis (and perhaps its transverse axis) metaphorically to the sea, such that ‘down’ is towards the northwest and ‘up’ towards the southeast (François 2003, 2004a). If we wanted to be particular, incidentally, we could label a number of geographic expressions ‘deictic-geographic’. To say that something is seawards, for example, is to place it in relation to the speaker. If the speaker were closer to the sea, the same referent might be ‘inland’.
2. An intrinsic system specifies the location of an object in relation to a reference object. European languages often use prepositions for this purpose (in, on, under, over, beside). A few Oceanic languages also use adpositions, but in the majority a local construction is used. This is often an adpositional phrase containing a relational local noun, i.e. a noun that refers to a part of the reference object or to a location in relation to the reference object (‘inside’, ‘upper surface’, ‘top’, ‘underside’, ‘side’).
3. Relative locations entail both the position of a reference person, often the speaker, and the position of a reference object. Relative locations look like intrinsic locations, but the latter do not entail a reference person (Leech 1969:167–168). For example, ‘in front of the house’ is an intrinsic location because a house has a ‘front’. For an Oceanic speaker, this is the side with the ladder, balcony and door. But ‘in front of the tree’ is a relative location because a tree has no intrinsic front. When an English speaker uses this expression, s/he treats the tree as if it were facing her/him, and so the part of the tree that ‘faces’ the speaker is treated as its front. Thus the ‘front’ changes with a change in the speaker’s position. Similarly, ‘turn left’ entails a reference person—the addressee— and a reference object. The reference object is the addressee’s body, and the direction of ‘left’ varies according to the addressee’s location and which way s/he is facing. Relative locations are not used in Oceanic languages, however. Oceanic speakers use terms like ‘front’ and ‘left’ intrinsically. One does not talk about the ‘front’ of a tree, and ‘left’ is strictly an intrinsic part of the speaker’s body, not a direction (Ozanne-Rivierre 1997). Instead, one uses the geographic system: ‘seawards of the tree’, ‘go left-along-the-coast’ (cf. Hill 1997).
4. A deictic system is based on location relative to the speaker or to some other person and often also relative to the addressee. Deictic expressions are tied to the context of the individual speech act; they do not refer to fixed points in space. All languages probably have demonstratives of some kind (‘this’/‘that’, ‘here’/‘there’, ‘voici’/‘voilà’ etc). Many Oceanic languages in addition have deictic directional morphemes in their verb phrases (or sometimes in locative phrases) which indicate whether direction associated with the event is towards the speaker and/or, in some languages, towards the addressee.¹

Relationships between widespread Oceanic categories that probably existed in POc and systems of talking about space are summarised below.

1. Some local nouns (§2) are used to express the directions of the vertical, inland/seaward and coastal axes of the geographic system. Overlapping with these semantically are local nouns that express locations and directions in the local environment that are part of the shared experience of speaker and addressee (‘home’, ‘bush’, ‘garden’ etc). There are also relational local nouns that express locations (‘inside’ etc) in the intrinsic system.
2. Directional morphemes (§3) in verb phrases and in adjuncts are used in Oceanic languages to express geographic (vertical and valley-based) and deictic directions. In single-verb predicates and in serial verb constructions these morphemes are verbs. Elsewhere they are morphemes grammaticised from verbs. Grammaticisation had probably already occurred in some cases in Proto Oceanic (Ross 2003).²
3. Demonstratives are used as static deictics. The reconstruction of demonstratives is not treated in this chapter.

We can illustrate the typical Oceanic system by summarising the system in Longgu (SES) as described by Hill (1992, 1997, 2002). Local nouns marking the axes of the geographic system are asi ‘seawards’, loŋa ‘inland’, alaʔa ‘east’, toli ‘west’, vuʔa ‘down’, laŋi ‘up’. Longgu has terms meaning ‘left’ and ‘right’, but their referents are the arms and the sides of the body, and they do not form part of the system of spatial orientation and are not local nouns. Where English would use ‘left’ or ‘right’, a Longgu speaker refers to aʔae asi ‘the seawards leg’ or aʔae longa-i ‘the inland leg’. ‘Nouns such as komu ‘village’ and luma ‘house’ function as place [local] nouns when their referent is the village or house considered to be the “home” location’ (Hill 1997:103) and occur in different constructions from when they are used as common nouns (p.233). One of these constructions is with the preposition i or with no preposition. Relational local nouns express intrinsic locations in relation to objects. For example, ubu- ‘inside’ is treated as possessum in relation to the possessor pera ‘basket’:

• • Longgu (SES)

    ‘The banana is inside the basket.’

    vugi	ubu-na	pera-i
    banana	inside-P:3SG	basket-SG

Other relational local nouns are buri- ‘behind’, naʔova- ‘front’, gege- ‘side’, orova- ‘underneath’, vavo- ‘top surface, space above’, levua- ‘middle, centre’. Demonstratives are also a type of local noun, occurring with the preposition i to indicate a location. Unlike many Oceanic languages, which have a person-oriented deictic system (§3.4), Longgu deictics distinguish four degrees of deictic distance from the speaker: nene ‘this, here’, nina ‘that, there’, ninaina ‘that yonder, yonder’ and nihou ‘that/there far away’.

Like many Oceanic languages, Longgu has postverbal particles that distinguish motion towards and away from the speaker — mai ‘hither’ and hou ‘thither’ — and are hence deictic. Some languages have a third particle indicating motion towards the addressee (§3.4), and others also have particles expressing the geographic directions ‘up (valley)’, ‘down (valley)’ and ‘across (valley)’ (§.3.3).

There is no one-to-one relationship between the geographic, intrinsic, and deictic systems and the morphosyntactic categories used to express them. For this reason this chapter is organised on the basis of the morphosyntactic categories as they occurred in POc, rather than on the basis of the semantic systems outlined above.

2. Local nouns⇫

In POc and indeed throughout the history of many Oceanic subgroups there were three subclasses of noun: personal, common and local. These subclasses are defined by the constructions in which they occur. Since some nouns occur in a number of constructions, they are assigned to more than one subclass.

1. Personal: personal proper names and kin terms used of known individuals. In POc a personal noun phrase construction consisted of the personal article *i or *e plus a personal noun,³ reflected in Taiof (MM) e Maras ‘Maras (proper name)’, e cina-nai (ART mother-P:3SG) ‘his/her mother’.⁴ In Longgu (SES) personal nouns occur without an article, but POc *i is reflected in the ‘pronominal article’ in i gira ‘they’ and other free pronouns.
2. Common: human nouns and non-human animates when not used of known individuals, as well as inanimates and abstract nouns. A common noun with a specific referent was in most cases preceded by the common article *a or *na, as in POc *a/na Rumaq ‘a/the house’.⁵ The construction is reflected in Taiof a numa ‘a/the house’, a patu-re (ART head-P:3PL) ‘their heads’; Longgu a komu-i (ART village-SG) ‘the village’⁶. If such a noun was used in an adjunct construction it was preceded by the sequence *i ta-, as in POc *i ta-ña Rumaq (PREP *ta-P:3SG house) ‘at a/the house’.⁷ In this construction *ta- was a monovalent semantically empty noun. In most Oceanic languages where this latter construction is reflected, however, *i has dropped out, leaving *ta- as a preposition.⁸ Hence Taiof ta-na patu-na tober (PREP-P:3SG head-P:3SG hill) ‘on top of the hill’, Longgu ta-na iola-i (PREP-P:3SG canoe-SG).
3. Local: nouns used with reference to a specific location, a time, or an intrinsically located part of something. The POc local construction consisted of the preposition *i plus a local noun, with no intervening article. Local nouns include:
   1. proper placenames;
   2. nouns denoting locations, including
      1. nouns denoting familiar places like ‘home’, ‘(own) village’, ‘(own) garden’, ‘bush’, ‘beach’ etc.;
      2. nouns denoting geographic directions, ‘down below’, ‘up above’;
   3. temporal nouns;
   4. monovalent relational nouns, e.g. nouns denoting intrinsic parts, like ‘inside’, ‘upper surface or space above’, ‘lower surface’, ‘space beneath’ and so on; In Taiof a local noun may form a locative expression without a preposition (i.e. *i has been lost), like koma-na matan below.⁹ A locative expression with a common noun must be formed with a preposition.

• • Taiof (MM)

    ‘Maras is in the ditch.’

    E	Maras	to	noŋos	no-n	koma=na	matan.
    ART	Maras	REAL	dwell	IMPF-P:3SG	inside=ART	ditch

In Longgu local and common nouns are preceded by different paradigms of relators and prepositions (the term ‘relator’ is explained on p.268):

	location	extent	direction
	‘at’	‘as far as’	‘to, towards’
with a local noun	i	mi	vu
with a common noun	ta-	mi ta-	vu ta-

Thus we find:

• • Longgu (SES)

    ‘s/he went towards her/his (home) village’ (local noun)

    e	la	vu	komu
    S:3SG	go	R	village

  • Longgu (SES)

    ‘s/he went to her/his (canoe)’ (common noun)

    e	la	vu	ta-na	iola	ŋaia
    S:3SG	go	R	PREP-P:3SG	canoe	D:3SG

When a local noun indicating a geographic direction follows a verb it may occur without a preposition or relator, e.g. lae asi ‘go seawards’.

A noun like *tama-gu (father-my) used with the personal article served as a name (‘Dad’, ‘Papa’); used without an article it served as a common human noun (‘my father(s)’). Similarly, in a local construction a noun like *Rumaq ‘house’ behaved grammatically and semantically like a placename, so that *i Rumaq meant ‘at home’ (p.241), but in the general adjunct construction *i ta-ña Rumaq ‘at a/the house’ *Rumaq was a common noun.

Inspection of the grammars of Oceanic languages shows that many have retained the distinction between local and common constructions, although a majority express it with morphemes other than reflexes of *i and *ta- (for elaboration, see Ross 2004); see also the local prepositions listed under POc *la[-] (p.289). This inspection also shows that common nouns can be readily co-opted into the local construction. Thus it is common to find the common nouns for ‘house’, ‘garden’, ‘village’, ‘bush’ and ‘beach’ also occurring in the local construction, but this does not mean that the POc etyma they reflect necessarily occurred in the POc local construction.

As the Taiof examples above show, the division into common and local nouns cut across the division into zero-valency and monovalent nouns (vol.1,32).

Zero-valency local nouns denoted familiar places like ‘home’, ‘own village’, ‘own garden’, ‘bush’, ‘beach’, i.e. locations whose reference would be clear to the addressee without further specification or whose exact reference was irrelevant (like the English construction with at and no article in at home, at school, at hospital, at work). Also belonging to the zero-valency local category were nouns denoting regions, either in relation to, say, the island home of the interlocutors or in relation to the speaker. These nouns denote geographic directions such as ‘up above’, ‘down below’, ‘inland’, ‘at sea’ and so on. There is no sharp distinction between these and nouns for familiar places: ‘inland’ and ‘bush’, for example, are often synonymous in Oceanic languages. This is not surprising. Palmer (2001) points out that the terms for a culture’s geographic directions are commonly grammaticised from perceptually salient phenomena of the landscape. These nouns have sometimes been labelled ‘absolute’ local nouns in the literature, to distinguish them from relational local nouns, but I prefer the label ‘geographic’ as some of their uses are deictic, referring to a location in relation to the speaker and addressee. Familiar-place and geographic local nouns are reconstructed in §2.2.

Monovalent local nouns, often labelled ‘relational’ nouns in the literature, referred to parts of objects. These are reconstructed in §2.3.

In POc the zero-valency/monovalent distinction was apparently not as sharp as it is in many modern Oceanic languages (Lynch, Ross & Crowley 2002:78–79), but there were semantically driven tendencies in the behaviour of nouns. When a noun was viewed as semantically inalienable, like the inside of an object, it was monovalent (i.e. directly possessed, with a possessor suffix), but the same noun could also have zero valency if used in a context where inalienability was irrelevant. For this reason, the relational nouns reconstructed in §2.3 have both monovalent and zero-valency forms, and we find cases where some reflexes of a relational noun are monovalent, others zero-valency.

The major local nouns reconstructed below are listed here with simplified glosses, in two groups, relational and familiar-place/geographic. Where a noun is also reconstructed as a common noun, its common-noun gloss is also given, and where a local adverb in *qa- is reconstructable this is also shown.

Familiar-place and geographic local nouns:

	as common noun	as local noun	adverb
*qutan	‘bushland, hinterland’	‘inland’	*qa-qutan
*loŋa	…	‘inland’	—
*laur	—	‘seawards’	*qa-laur
*tasik	‘sea, salt water’	‘at sea’	—
*Rumaq	‘house’	‘home’	*qa-Rumaq
*tanoq	‘earth, soil’	‘down below’	*qa-tanoq
*atas	‘high country, uplands’	‘up above’	—
*laŋit	‘sky, weather’	‘up top, high up’	—
*laka	…	‘up above’	*qa-laka
*liwaŋ	‘open space’	‘middle’	—

Relational local nouns:

	as common noun	as local noun	adverb
*lalo-, *lo-, *lalom	—	‘inside’	—
*papa-, *pa-, *papak, *pak	—	‘underside’	—
*papo-, *po-, *papo, *po	—	‘upper surface’	—
*qulu-, *qulu	‘head, (head) hair’	‘top’	PNCV *qa-qulu
*[pʷa]pʷaRa-, *[pʷa]pʷaRa	‘cheek’	‘side’	—
*qaro-, *qarop	‘face’ ?	‘front’	—
*muqa-, *muqa	—	‘front’	PNCV *qa-muqa
*mata-, *mata	‘eye’	‘front’	—
*nako-, *nako	‘face’	‘front’	—
*muri-	‘back’	‘back’	—

‘—’ indicates that the item is not reconstructable, ‘…’ that the evidence is insufficient to decide whether it is reconstructable.

Heine (1989) observes an implicational relationship ‘under’ > ‘on’/‘in’ > ‘front’ > ‘back’ such that if any of these relational meanings is derived historically from a body-part term, so will be the meanings to the right of it. This is supported by the listing above, as ‘under’ and ‘on’/‘in’ are not derived from body-part nouns, but ‘front’ and ‘back’ are. Heine argues that this is a scale of increasing deictivity from left to right, but Bowden (1992:53) argues that ‘front’ and ‘back’ are based on body-part terms simply because entities in the landscape are not perceived as having intrinsic fronts and backs, i.e. Oceanic languages do not have a relative spatial system (p.230).

2.1. The preposition *i and the prefix *qa-⇫

The preposition *i, which occurred before local nouns, is widely reflected in Oceanic languages, but there are also many languages in which a local noun occurs without a preposition.

PAn		*i	[PREP] ‘g:locative’ (ACD)
POc		*i	‘g:locative’
Adm	Mussau	e-	(fossilised prefix on locative nouns)
Adm	Penchal	i	[PREP] (locative, temporal, allative, instrumental)
Adm	Lou	e	[PREP] (locative, allative)
NNG	Manam	e-	(locative prefix, e.g. e-lau ‘seawards’)
PT	Molima	i-	(locative prefix)
PT	Sudest	e	[PREP] (locative)
MM	Tigak	e	(PREP w PLC; locative)
MM	Kara	i	[PREP] (locative)
MM	Nalik	i-	(prefix on locative demonstratives)
MM	Tabar	i	(PREP w PLC, N LOC; locative)
MM	Lihir	i	(PREP w PLC, N LOC; locative)
MM	Sursurunga	i	[PREP] (locative, temporal)
MM	Ramoaaina	i-	(prefix on locative demonstratives)
MM	Halia	i	[PREP] (locative)
MM	Taiof	i	(PREP w predicate N LOC; locative)
SES	Bugotu	i	[PREP] (locative)
SES	Gela	i	(PREP w N LOC; locative)
SES	Lengo	i	[PREP] (locative)
SES	Longgu	i	(PREP w N LOC; locative)
SES	Lau	i	(PREP w PLC, N LOC; locative)
SES	Kwaio	i	[PREP] (locative)
SES	Kwara’ae	i	[PREP] (locative)
SES	Sa’a	i	[PREP] (locative)
SES	Arosi	i	[PREP] (locative)
NCV	Mota	i	[PREP] (locative)
NCV	Merlav	i	[PREP] (locative)
NCV	Nguna	e-	[PREP] (locative)
SV	Kwamera	i-	(prefix on locative nouns)
SV	Lenakel	i-	(prefix on locative nouns)
SV	Anejom̃	i-	(prefix on locative nouns)
NCal	Iaai	e-	(prefix on locative adverbs)
Fij	Wayan	i	[PREP] (locative)
Fij	Bauan	e	[PREP] (locative)
Pn	Tongan	i, ʔi	[PREP] (locative)
Pn	Niuafo’ou	ʔi	[PREP] (locative)
Pn	East Futunan	i	[PREP] (locative)
Pn	East Uvean	ʔi	[PREP] (locative)
Pn	Samoan	i	[PREP] (locative)
Pn	Pileni	i	[PREP] (locative)
Pn	Marquesan	ʔi	[PREP] (locative)

Tongan, Niuafo’ou and East Uvean ʔi are phonologically problematic, as they reflect POc *qi, the form of the genitive preposition governing a non-specific inalienable possessor (Pawley 1972, Clark 1976, Hooper 1985, Ross 1998b, 2001b). However, as Clark (1976) and Blust (ACD) note, the introduction of a historically secondary glottal stop also occurs in some other Tongan grammatical formatives.

POc *i occurred with local nouns. Its exact distribution is not clear, but it is likely that *i was omissible if the phrase it initiated was licensed by the verb.

There are a number of languages in which the reflex of *i is the general locative preposition. I take these to be cases where an extension of meaning has occurred, as there are ample and widely distributed languages where *i is attested only with local nouns.

As mentioned above, a number of local nouns also formed POc local adverbs with the prefix *qa-. Jauncey (1997) notes for Tamambo that the prefix a- means ‘location in/at a place’, and must be prefixed to a noun signifying a place in relation to the speaker, i.e. a local noun. The cognate set is listed below. This is the *qa- which Pawley (1972:82, 114) found in Southeast Solomonic and Northern Vanuatu languages as a formative of temporal adverbs (Ch. 9, p.324).

POc		*qa-
NNG	Bariai	ga-	(local adverb formative)
NNG	Manam	a-	(local adverb formative; fossil)
NNG	Kairiru	qa-	(local adverb formative; fossil)
MM	Vitu	ɣe-	(local adverb formative)
MM	Siar	a-	(local adverb formative)
MM	Tolai	a-	(local adverb formative)
SES	Gela	ɣa-	(local adverb formative; fossil)
NCV	Mota	a-	(local adverb formative)
NCV	Ambae	a-	(local adverb formative)
NCV	Raga	a-	(local adverb formative)
NCV	Port Sandwich	a-	(local adverb formative)
NCV	Tamambo	a-	(local adverb formative)
NCV	Paamese	a-	(local adverb formative; fossil)

There also seems to have been an alternant *ŋa-, reflected in Poeng (NNG) ŋa-, Nakanai (MM) ga-, Longgu (SES) ŋa- and Samoan, Tikopia (Pn) ŋa-, but it is not strongly attested.

2.2. Familiar-place and geographic local nouns⇫

The nouns reconstructed in this section are zero-valency local nouns denoting regions, either in relation to the island home of the interlocutors or in relation to the speaker. They have meanings like ‘inland, bush’, ‘seaward, beach’, ‘down below’ and ‘up above’. Some of these glosses are adverbial in English. This is because a zero-valency local noun preceded by *i often performed the task of an English adverbial. For example, POc *i tanoq evidently expressed something like ‘down there’.

Some POc zero-valency local nouns, at least, were also used as common nouns, and this difference sometimes also entailed a difference in meaning. Thus *i tanoq meant ‘down there’ but *tanoq meant ‘earth, soil’ (vol.1,119); *i qutan meant, among other things, ‘inland’, but *qutan meant ‘bushland’ (vol.1,118).

As far as possible, the cognate sets below are limited to reflexes of local-noun uses. However, the glosses of the reconstructions distinguish between common-noun (N) and local-noun (N LOC) meanings.

2.2.1. ‘Inland, bush’⇫

As a common noun POc *qutan meant ‘bushland, hinterland’ (vol.1,118; this volume, Ch. 3, §5.1). As a local noun, however, *qutan denoted the direction of the bush, namely ‘inland’. By extension, this has come to mean ‘upwards’ in a number of languages, by virtue of the fact that the inland region is significantly higher than the coast on many Pacific islands.

PAn		*quCaN	‘scrubland, bush’ (ACD)
PMP		*qutan	‘small wild herbaceous plants; scrubland, bush’ (ACD; Dempwolff 1938)
POc		*qutan	[(N) ] ‘bushland, hinterland’
POc		*qutan	[(N) ] ‘bushland, hinterland’
POc		*qa-qutan	[(ADV) ] ‘in the bush, inland’
NNG	Manam	-uta	‘inland’ (root of adverbials)
NNG	Manam	(a)uta	[N] ‘inland’
NCV	Mota	uta	‘bush, forest, unoccupied land; the inland country’
NCV	Ambae	(a)ute	[ADV] ‘up in the bush’
NCV	Tamambo	(a)uta	[ADV] ‘inland direction’
NCV	Paamese	ut	[N LOC] ‘ashore’
NCV	Lewo	ura	[N LOC] ‘ashore’
NCV	Nguna	uta	‘inland’
SV	Lenakel	-ət, -it	[DIR] ‘upwards’
SV	Kwamera	(p)uta	[N LOC] ‘up, upwards, on top of’
Mic	Kosraean	wʌt	[N] ‘area inland or towards the mountains’
Mic	Marshallese	(e)ɒ̄c	[N] ‘interior of an island’
Mic	Mokilese	(e)wic	[N] ‘inland’
Fij	Rotuman	ufa	‘land (from the sea); interior (from the coast)’
Pn	Tongan	ʔuta	[N LOC] ‘inland (from shore); shore, land (from sea)’
Pn	Niuafo’ou	(ŋā)ʔuta	[N LOC] ‘upland’
Pn	Samoan	uta	[ADV] ‘on shore, inland’
Pn	Samoan	i uta	[N LOC] ‘on the side facing the land’ (as opposed to i tai ‘on the side facing the sea’)
Pn	Samoan	(ŋā)ʔuta	[ADV] ‘to shore, in an inland direction’
Pn	Pileni	(ɣa)uta	[N LOC] ‘shore, village location on shore, inland’
Pn	Tikopia	(ŋa)uta	‘inland, landwards’
Pn	Hawaiian	uka	‘inland (from shore); shore, land (from sea)’
Pn	Marquesan	uta	[N LOC] ‘towards the mountain’

The cognate set below has fewer members that the one above, but appears to reflect a local noun with similar meanings. Polynesian reflexes display a vowel change and a change in meaning from ‘inland’ to ‘top, space above, up top’, a change presumably mediated by the fact that on a high island (as opposed to atoll) the inland of the island was also its ‘top’.

POc		*loŋa	[(N) ] ‘inland’; [(N LOC) ] ‘inland’
POc		*loŋa	[(N) ] ‘inland’; [(N LOC) ] ‘inland’
NNG	Bariai	loŋa	‘bush people’
SES	Bugotu	loŋa, (i)loŋa	‘landwards, from sea’
SES	Gela	loŋa, (i)loŋa	‘landwards, inland’
SES	Longgu	loŋa	[N LOC] ‘direction towards bush’
SES	Kwaio	(i ka)loŋa	‘in the forest’
Mic	Marshallese	-lʌŋ	[DIR] ‘upward’
Mic	Puluwatese	-loŋ	[DIR] ‘inland’
Mic	Woleaian	roŋ	[DIR] ‘inland’
PPn		*luŋa	‘top, space above, up top’ (-u- for expected *-o-)
Pn	Niuean	luŋa	[N LOC] ‘above, upon, top’
Pn	Niuafo’ou	(ʔo)luŋa	[N LOC] ‘up’
Pn	Samoan	luŋa	[N LOC] ‘top, space above, up top’
Pn	Pileni	luŋa	[N LOC] ‘top, up, east’

2.2.2. ‘Seaward, towards the beach, at sea’⇫

The principal POc local noun meaning ‘seawards’ was *laur, and it was evidently the antonym of *qutan and *loŋa above. It reflects PMP *lahud ‘downriver, towards the sea’, and it is likely that it was inherited into POc primarily, perhaps exclusively, as a local noun denoting a direction (for common-noun reflexes, see p.95). In this regard it differed from POc *tasik ‘sea, salt water’ and POc *laman ‘deep sea beyond the reef’, reconstructed in Chapter 4, which were common nouns.

POc *tasik is quite often reflected as a local noun and presumably functioned as both a local and a common noun in POc. Occasional reflexes have meanings similar to those of POc *laur, but most mean ‘at sea’, and this was presumably its POc meaning.

It is unlikely that POc *laman normally occurred as a local noun, as only two local noun reflexes have been found (Mussau [Adm] lamana ‘beach’, Tigak [MM] laman ‘down there, at the beach’).

PMP		*lahud	‘downriver, towards the sea’ (Blust 1997; Dempwolff 1938)
POc		*laur	[(N LOC) ] ‘sea, seawards’
POc		*qa-laur	[(ADV) ] ‘in a seaward direction’
Yap	Yapese	lʔāy	‘seaward’
NNG	Manam	-lau	‘seawards’ (root of adverbials)
NNG	Manam	(i)lau	[N] ‘seawards’
MM	Nakanai	(go)lau	‘go toward the sea’
MM	Meramera	-lau	‘beach’ (root of adverbials)
MM	Kokota	rauru	‘seaward’
SES	Bugotu	lau, (i)lau	‘seawards’ (as opposed to i-loŋa ‘landwards, inland, towards land’)
SES	Gela	lau, (i)lau	‘seawards, shorewards from a speaker inland, (river) bank’ (as opposed to i-loŋa ‘landwards, inland, towards land’)
SES	Lau	lau	‘north; open sea to the north’
NCV	Mota	lau	‘the beach, as approached from the land’
NCV	Ambae	(a)lau	[ADV] ‘down by the sea’
NCV	Tamambo	(a)lau	[ADV] ‘seawards, shore direction’
NCV	Paamese	(a)lau	‘seawards’

PMP		*tasik	‘sea’ (Dempwolff 1938)
POc		*tasik	[(N) ] ‘sea, salt water’; [(N LOC) ] ‘at sea’
POc		*tasik	[(N) ] ‘sea, salt water’; [(N LOC) ] ‘at sea’
MM	Lihir	(i) tes	‘at sea’
MM	Ramoaaina	(nə)tai	‘on the sea, out to sea’ (not ‘seawards’)
MM	Tolai	(na)ta	‘on the sea, out to sea’ (not ‘seawards’)
SES	Longgu	asi	[N LOC] ‘seawards’
SES	Kwaio	(i) asi	‘at the coast’
SV	Lenakel	(i)rhe	[ADV] ‘at/to sea’
Fij	Wayan	taði	[N LOC] ‘coast, seashore, sea, from the perspective of the land’
Pn	Niuafo’ou	(ŋā)tai	[N LOC] ‘outer sea’
Pn	Samoan	tai	[ADV] ‘at sea’ (i tai ‘on the side facing the sea’, as opposed to i uta ‘on the side facing the land’)
Pn	Samoan	(ŋā)tai	[ADV] ‘to sea’
Pn	Marquesan	tai	[N LOC] ‘sea’

2.2.3. Directions along the coastline⇫

The items reconstructed above for ‘inland’ and ‘seawards’ were orientations at an angle to the coastline. Also recorded for a few languages are items referring to the two directions along the coastline. Possible PWOc candidates for these meanings are given below. Although the data are fragmentary, the fact that both members of the pair are recorded in widely separated languages supports their reconstruction. The Nalik glosses ‘south-east’ and ‘north-west’ are equivalent respectively to ‘to one’s left when facing the sea’ and ‘to one’s right when facing the sea’ if one is on the west coast of New Ireland.

PWOc		*pa	[(N LOC) ] ‘to one’s left when facing the sea’
PWOc		*qa-pa	[(ADV) ] ‘to one’s left when facing the sea’
NNG	Manam	(a)wa	[N] ‘to one’s left when facing the sea’
MM	Vitu	(ɣe)va	[ADV] ‘downwards’
MM	Nalik	pa	[N LOC] ‘south-east’

PWOc		*ta	[(N LOC) ] ‘to one’s right when facing the sea’
PWOc		*qa-ta	[(ADV) ] ‘to one’s right when facing the sea’
NNG	Manam	(a)ta	[N] ‘to one’s right when facing the sea’
MM	Vitu	(ɣe)ta	[ADV] ‘upwards’
MM	Nalik	ta	[N LOC] ‘north-west’

Nakanai (MM) has the non-cognate pair of roots (used in directional and local verbs) -ale ‘to one’s left when facing the sea’ and -muli ‘to one’s right when facing the sea’.

2.2.4. ‘At home’⇫

The cognate set attesting POc *Rumaq ‘house’ was presented in vol.1,48. It is reasonably clear that it also functioned as a local noun in the phrase *i Rumaq ‘at home’, and apparently also in the adverb *qa-Rumaq.

PAn		*Rumaq	‘dwelling house’ (Blust 1987)
POc		*Rumaq	[(N)] ‘house’; [(N LOC) ] ‘home’
POc		*Rumaq	[(N)] ‘house’; [(N LOC) ] ‘home’
POc		*qa-Rumaq	[(ADV) ] ‘at home’
PT	Saliba	numa	[N LOC] ‘home’
MM	Nakanai	(go)luma	‘go to nearest hamlet, usually one’s home’
MM	Meramera	-luma	‘home’ (root of adverbials)
MM	Ramoaaina	(nə)ruma	‘at home’
MM	Taiof	numa	[N LOC] ‘home’
NCV	Tamambo	(a)imo	[ADV] ‘at home’
NCV	Paamese	(tela)im	[N LOC] ‘home’
NCV	Lewo	umʷa	[N LOC] ‘home’
SV	Lenakel	īmʷa	‘at home, homewards’ (cf. nimʷa ‘house’)

2.2.5. ‘Down below’⇫

POc *tanoq is reconstructed as a common noun meaning ‘earth, soil, ground; land’ in vol.1,119 and in this volume, p.41. However, there is well distributed evidence that as a local noun it meant ‘down below’. This is not surprising when one considers that POc speakers must generally have lived in stilt houses (vol. 1, Ch. 3) for whose inhabitants the ground was indeed ‘down below’.

PMP		*taneq	‘earth, land’ (Dempwolff 1938)
POc		*tanoq	[(N)] ‘earth, soil, ground; land’; [(N LOC) ] ‘earth, soil, ground; land’
POc		*tanoq	[(N)] ‘earth, soil, ground; land’; [(N LOC) ] ‘earth, soil, ground; land’
POc		*qa-tanoq	[(ADV) ] ‘down on the ground, down below’
Adm	Loniu	tan	[ADV] ‘down’
NNG	Takia	tan (na)	[N + POSTPOSITION] ‘on the ground, down below’
PT	Gumawana	tono	‘down’
MM	Harua	tano	‘down there’
MM	Nakanai	(go)talo	‘go down’
MM	Meramera	tano(do)	‘down there’ (-do DEM)
NCV	Tamambo	(a)tano	‘down on the ground, downwards’
NCV	Paamese	dan	‘down, below’
NCV	Lewo	tano	‘underneath, downwards’
SV	Kwamera	təna	‘earth, ground; land, island, country’
NCal	Iaai	kɔnɔ	‘earth, ground’
NCal	Nengone	ten	‘under’

The meanings above overlap with the adverb reflexes of POc *sipo ‘go downward’, (ADV) ‘down below’, but *tanoq, a noun, and *sipo a verb (and perhaps adverb), belonged to different word classes (§3.3.1).

2.2.6. ‘Up above’⇫

A few of the reflexes of POc *atas ‘top, space above’ are monovalent relational nouns. However, the vast majority of reflexes are geographic, not relational, nouns, and it seems that POc *atas was also a geographic noun. It also seems that it was not a common noun (in this respect it resembles POc *laur, p.239).

The items listed under ‘cf. also’ below reflect a Proto North Bougainville form *yasa, which has replaced *yatasa. Possibly *yasa is derived from expected *yatasa by idiosyncratic deletion of the middle syllable.

PAn		*aCas	‘high, tall’ (ACD)
POc		*atas	[(N) ] ‘top; space above’; [(N LOC) ] ‘up top’
POc		*atas	[(N) ] ‘top; space above’; [(N LOC) ] ‘up top’
NNG	Ali	yat	‘on top’
NNG	Tumleo	yot	‘on top’
PT	Are	yata	‘on top’
PT	Gapapaiwa	yata	‘on top’
PT	Sinaugoro	iata(na-i)	‘on top of it’ (N-P:3SG-POSTP)
PT	Motu	lata-	‘summit, top’
PT	Motu	ata(i)	‘on top’ (N-POSTP)
MM	Bali	ɣata	‘up (there)’
MM	Nakanai	(go)ata	‘go upwards’
MM	Meramera	uata	‘upwards’
MM	Lavongai	(la)kat	‘top’
MM	Nalik	uata	‘top’
MM	Sursurunga	(u-ram)iet	‘upwards’
MM	Sursurunga	ieti	‘top’
MM	Tangga	(l)iat	‘up (there)’
MM	Tangga	(ua)yat	‘upwards’
Mic	Kiribati	(i)eta	‘up, on high, above, top, upper, heavens’
Mic	Marshallese	ec	‘upper, eastern’
Mic	Chuukese	ās	‘upper part, top, summit, eastern side’
Mic	Chuukese	asa-	‘upper part’
Mic	Woleaian	yat	‘up, top’
Fij	Wayan	ata	[N LOC] ‘top, above; interior of a mountainous island, up the hill, inland’

cf. also:

MM	Solos	yas	‘top’
MM	Solos	(i)yas	‘up (there)’
MM	Petats	(i)yas	‘up (there)’
MM	Petats	yas	‘topside; upwards’
MM	Halia	(i)yasa	‘up (there)’
MM	Halia	(pal)yasa	‘upwards’
MM	Selau	(i)yasa	‘up (there)’
MM	Taiof	yas	‘up (there); upwards’

POc *laŋit ‘sky, weather’ is reconstructed as a common noun in Chapter 5, but the reflexes below suggest that it was also used as a POc local noun.

PAn		*laŋiC	‘sky’ (Dempwolff 1938)
POc		*laŋit	[N] ‘sky, weather’; [N LOC] ‘up top, high up’
Yap	Yapese	lǣŋ	‘up’
SES	Lau	(i) laŋi-	‘up, above’
SES	Kwaio	laŋi-	‘space above’
NCV	Mota	laŋ	‘upwards, heavenwards’
NCV	Nguna	(e)laŋi	‘up, high, above, top’
Mic	Marshallese	lʌŋ	‘up, above’

The meanings above overlap with the adverb reflexes of POc *sake ‘go upward’, (ADV) ‘upwards, up top’, but *atas and *laŋit were nouns, *sake a verb (and perhaps adverb, p.277).

It is less clear whether POc *laka ‘up above’ was a noun or a verb. In Takia its reflex is a zero-valency noun, in Mapos Buang and Kiriwina a local adverb. These could be derived from either a noun or a directional verb. Monovalent noun reflexes occur in the two New Ireland (MM) languages Lihir and Siar, but in other New Ireland languages (Lavongai, Tigak and Kara) the reflexes are verbs. However, the Southeast Solomonic reflexes reflect the derived adverb *qa-laka ‘up there, up above’ (which was then used as a verb in some SES languages). Since *qa- is far more readily reconstructable as a prefix to nouns than to verbs, I assume that *laka was a noun.

POc		*laka	[(N LOC) ] ‘up above’
POc		*qa-laka	[(ADV) ] ‘in an upward direction’
NNG	Takia	lak (na)	‘high up’ (na local postposition)
NNG	Mapos Buang	raq	‘up, above’
PT	Kilivila	lake(va)	‘top, in sky’
MM	Lavongai	(saŋ)lak	‘(sun) rise’
MM	Tigak	lak	‘(sun) rise’
MM	East Kara	lak	‘(sun) rise’
MM	Lihir	laka-	‘top surface, space above’
MM	Siar	laka-	‘top surface, space above’
SES	Gela	(ɣa)laɣa	‘up’
SES	Talise	(ɣa)laɣa	‘go up’
SES	Birao	(ha)laha	‘go up’

There was also a POc verb *laka, which meant ‘walk’ and apparently had no directional meaning.

2.2.7. ‘In the middle, between’⇫

There is well distributed evidence that POc *liwaŋ, *liwa-/*liwaŋa- was a local noun meaning ‘open space, space between, middle’, and that it had at least one relational alternant. The form we would expect to find reflected in most languages with relational forms is POc *liwa-. However, we also find reflexes of *liwaŋa-. This may be the result of local developments, and this in turn may suggest that there was no relational form in POc.

The forms listed below under ‘cf. also’ are similar in form to those listed here. However, the fact that they share a formal irregularity — they seem to reflect POc *lua — and a different meaning — ‘outside’ — suggests that POc *lua ‘outside’ may have been a separate etymon, and also a local noun.

PMP		*liwaŋ	‘open space’ (ACD: Proto Western Malayo-Polynesian)
POc		*liwaŋ, *liwa-, *liwaŋa-	[(N) ] ‘open space, space between, middle’
POc		*liwaŋ	[(N LOC) ] ‘in the middle’
Adm	Loniu	(lɔhɔ)luwa-	‘middle’
NNG	Mangap	lwo-	‘torso; middle’
NNG	Manam	luʔaŋa-	‘space in middle’
NNG	Misim	livuŋ	‘front’
PT	Minaveha	niwani-	‘midst, among’
PT	Sudest	luɣawo-luɣawo-	‘middle’ (metathesis of **luwaɣo-)
MM	Ramoaaina	(nə)liwən	‘between’
MM	Tolai	(na)livuan	‘in the middle’
MM	Tolai	livuan	‘(be) in the middle’
SES	Longgu	levua-	‘middle, centre’
NCV	Ambae	livuge-	‘middle’
NCV	Tamambo	livuɣa-	‘middle part of s.t.’
NCV	Paamese	luhi, luhu	‘middle’
SV	Sye	(i)lvuɣ(teve)-	‘between, in the middle of’
Mic	Woleaian	rɨwan-	‘between, among’
Fij	Bauan	liwa	[N] ‘ocean far from land’; [ADV] ‘far from habitation’
Fij	Bauan	(mā)liwa	[N] ‘space between, interstice’

cf. also:

MM	Minigir	(na)lua	‘outside’
MM	Tolai	(na)lua	‘outside’
SV	Lenakel	(i)lua	‘outside’
SV	Kwamera	(i)rua	‘outside’

2.3. Relational local nouns⇫

The function of a POc monovalent relational local noun preceded by *i was similar to that of an English preposition, as in these Tabar (MM) examples, where the relational noun paki- ‘underneath’ performs a function similar to that of the English preposition under.¹⁰

• 1. Tabar (MM)

     ‘under the house’ (more literally ’at the house__e__s underneath’)

     i	paki-na	mi	vanua
     PREP	underneath-P:3SG	ART	house

  2. Tabar (MM)

     ‘under me’

     i	paki-gu
     PREP	underneath-P:1SG

In these Lengo (SES) expressions the relational noun muri- ‘back’ performs a function similar to that of the English preposition behind.

• • Lengo (SES)

    ‘behind the house’ (more literally ’at the house__e__s back’)

    i	muri-e	na	vae
    PREP	back-CSTR	ART	house

  • Lengo (SES)

    ‘behind me’

    i	muri-gu
    PREP	back-P:1SG

Hence we can with reasonable confidence make POc reconstructions such as:

• • POc

    ‘inside the house’ (more literally ’at the house__e__s inside’)

    *i	lalo-ña	Rumaq
    PREP	inside-P:3SG	house

  • POc

    ‘underneath the house’ (more literally ’at the house__e__s underneath’, gabʷari- ‘the area underneath a raised house’; vol. 1, p.51)

    *i	gabʷari-ña	Rumaq
    PREP	underneath-P:3SG	house

  • POc

    ‘behind me’

    *i	muri-gu
    PREP	back-P:1SG

Many Oceanic languages have relational nouns with the meanings reconstructed below: ‘inside’ (§2.3.1), ‘underneath, lower surface, space below’ (§2.3.2), ‘top, upper surface, space above’ (§2.3.3), ‘side’ (§2.3.4), ‘outside’ (§2.3.5), ‘front, time before’ (§2.3.6), ‘back, space behind, time after’ (§2.3.7). Although the nearest semantic equivalents of Oceanic relational nouns are English prepositions, I have used nouns and noun phrases in the titles of these subsections in an attempt to replicate the meanings of the reconstructed Oceanic terms.¹¹

In their monovalent form, relational local nouns are reconstructed below like other monovalent nouns, i.e. without their final consonant, on the assumption that it was lost before a possessor suffix: for example, *lalom ‘inside’ became *lalo-, *papak ‘underneath’ became *papa-. However, as I note in Lynch et al. (2002, Ch. 4), there is some evidence from Tanna languages (SV) that POc retained the final consonant in this context, so that, e.g., POc *lalo-ña in the reconstructed example above may have been (optionally?) *lalom-ña.

2.3.1. ‘Inside’⇫

The most widely reflected POc term for ‘inside’ is *lalo-/*lalom. This reflects PMP *Daləm with assimilation of the initial liquid to the intervocalic liquid: the expected POc form is **ralo-/**ralom. Reconstruction of unsuffixed *lalom is supported by just one reflex, Mussau e-lom-e.¹²

In Polynesian languages reflexes of *lalo-/*lalom denote the region underneath something. Blust (1997) suggests that this meaning change comes from the use of *lalom in relation to a planar surface, the sea, rather than a three-dimensional container.

PMP		*Daləm	‘inside’
POc		*lalo-, *lalom	[(N, N LOC) ] ‘inside’
Adm	Mussau	(e)lom(e)	[ADV] ‘inside’
NNG	Gitua	lolo-	‘inside’
NNG	Mangap	lele-	‘inside, in’
NNG	Kakuna	lolo-	‘inside’
NNG	Bam	liluo	‘room’
NNG	Kairiru	lal	‘inside’
NNG	Ulau-Suain	lulua-	‘room’
NNG	Ali	lal	‘room’
NNG	Numbami	(tae)lalo	‘intestines’
NNG	Numbami	(weni)lalo	‘forest’
NNG	Yabem	(ŋa)lelom	‘inside’
NNG	Kela	raro	‘inside’
PT	Motu	lalo-	‘inside, within’
PT	Mekeo	alo-	‘inside’
MM	Bola	lilo	‘inside’
MM	Meramera	lilo	‘inside’
MM	Notsi	lolo	‘inside’
MM	Lihir	lilie	‘inside’
MM	Sursurunga	lali	‘underside’
MM	Ramoaaina	lolo	‘intestines’
SES	Gela	lalo	‘deep, profound’
SES	Talise	lalo-na	‘in’
SES	Lau	(i)lalo	‘inside, in’
SES	Sa’a	lalo	‘inside’
SES	Arosi	raro	‘in’
NCV	Mota	lolo-	‘the inner part; a hollow; the inward part of man, heart, affections’
NCV	Raga	lolo-	‘inside, middle; body, stomach’
NCV	Uripiv	lolo-	‘inside’
NCV	Port Sandwich	nalö-n	‘inside; seat of feelings’
SV	Kwamera	reri-	‘internal portion, insides, heart, mind, feeling, emotion’
SV	Anejom̃	lele-	‘inside; heart, seat of feelings’
NCal	Tîrî	ñɯwɯ-	‘inside’
Mic	Kiribati	(i)nano-n	‘inside, in’
Mic	Kosraean	lʷal	‘deep’
Mic	Marshallese	i-lɔwa	‘inside, in’
Mic	Marshallese	lalᵚ	‘down, bottom, below, earth, world’
Mic	Ponapean	lɔlɛ	‘inside (it)’
Mic	Chuukese	ɾɾɔɾ	‘inside of’
Mic	Puluwatese	llɔn	‘in it’
Mic	Woleaian	raro	‘inside’
PPn		*lalo	‘region underneath’
Pn	Tongan	lalo	‘below, under’
Pn	Samoan	lalo	‘under, down, below’
Pn	Pileni	lalo	‘bottom, down, west’

There is evidence that POc *lalo- had two short forms, *lo- and *la-. A number of their reflexes occur as prepositions and may have been conflated with reflexes of POc *lako/*la ‘go (to); away from speaker’. They are listed together with a discussion of this conflation in §3.4.5. Listed below are those reflexes of the short forms which are not prepositions; most are local nouns. Significantly, there is a difference between the distributions of the two short forms. Reflexes of *lo- occur quite commonly as local nouns, and a number of them have a fossilised prefix reflecting the POc local preposition *i. Reflexes of *la- have a stronger tendency to occur as prepositions (p.288), and may reflect the short form of *lako rather than of *lalo-.

There is, of course, also a possibility that *lalo- has undergone haplology to form *lo- more than once in the history of Oceanic languages, but reflexes of *lo- below and in §3.4.5 are widespread enough to warrant its reconstruction in POc.

POc		*lo-	[(N LOC) ] ‘inside’
Adm	Titan	lo(n-um)	‘floor, inside of a house’ (um ‘house’)
NNG	Malasanga	lo-	‘inside’
NNG	Sio	(i)lo	‘inside’
NNG	Tami	lo	‘inside’
NNG	Mengen	lo-	‘inside’
NNG	Roinji	lo	‘inside’
NNG	Manam	(i)lo-	‘inside, in’
NNG	Bing	lo	‘inside’
NNG	Takia	(i)lo-	‘inside, in’
MM	Nakanai	-(i)lo	‘inside’
MM	Siar	lo	‘inside’
SES	Gela	lo-	‘inside’
NCV	Lonwolwol	lo-	‘inside; heart, feelings’

POc		*la-	[(N LOC) ] ‘inside’; [PREP] ‘in’
Yap	Yapese	lā-n	‘inside’
MM	Tigak	la-	‘inside’
MM	East Kara	la	‘inside’
Mic	Woleaian	ra-n	‘inside’

POc *loto- ‘space within a concave object’ is not well supported. It has become the default relational noun for ‘inside’ in Polynesian languages (where POc *lalo- is reflected with the meaning ‘underneath’; see above) and is also reflected in Wayan Fijian, so it can be reconstructed for Proto Central Pacific. Its reconstruction in POc rests on a single Admiralties reflex, Loniu lɔtiyɛ-, with -i- for expected -o-.

POc		*loto	[(N, N LOC) ] ‘space within a concave object’
Adm	Loniu	lɔtiyɛ-	‘inside’
Fij	Wayan	loto-	‘bottom, lowest part (e.g. of kava bowl)’
PPn		*loto	‘inside’
Pn	Tongan	loto	‘inside; hole or depression in coral reef or sea bed’
Pn	Samoan	loto	‘deep hole in lagoon; (house) interior’
Pn	Tahitian	roto	‘pool, lake, lagoon; inside’
Pn	Hawaiian	loko	‘pond, lake, pool; inside, interior; internal organs, as tripe’

In many Oceanic languages the word for ‘inside’ is the reflex of a POc body-part term. Two of these may have had the secondary meaning ‘inside’: POc *bʷal(o,a)-, *bʷal(o,a)k seems to have denoted the belly, POc *tinaqe- the intestines.

POc		*bʷal(o,a)-, *bʷal(o,a)k	[(N) ] ‘belly; hollow space’; [N LOC] ‘inside’
Adm	Nyindrou	bolo-n	‘inside, in’
Adm	Titan	pólo-n	[PREP] ‘among, inside’
NNG	Kairiru	balai	‘inside’
MM	Vitu	polok	‘inside’
MM	Sursurunga	polgo	‘inside’
MM	Tolai	(ta ra) bala-na	‘inside, in’ (ta PREP, ra ART, bala- ‘belly, interior’)
NCV	Raga	bʷala	‘shell’
NCV	Raga	bʷala(lolo)	‘middle’
NCV	Lonwolwol	bwele-n	‘hollow vessel, empty shell’
NCV	Paamese	vale(-ŋe-ne)	‘hollow part of something, cave’
NCV	Namakir	bwele-n	‘belly’
NCV	Nguna	(na-)pwele	‘stomach, belly, abdomen, waist, genital region’
NCV	Nguna	(na)-pwala( u-na)	‘among, middle, inside’

PMP		*tinaqi	‘small intestine’ (Blust 1981b)
POc		*tinaqe-	[(N)] ‘intestines; ?? (N LOC) inside’
Adm	Drehet	kxine	‘inner part, inside’
PT	Tawala	(u) hine-na	‘inside, in’
PT	Iduna	hinage-ne	‘inside’
PT	Gapapaiwa	sine	‘inside’
PT	Sudest	tine	‘inside’

2.3.2. ‘Underneath, lower surface, space below’⇫

The most widely reflected POc term for ‘underneath, underside’ is *papa-, *papak. This reflects PMP *babaq, which Blust reconstructs as referring to the underside or lower surface of something (the change from PMP *-q to POc *-k is unexplained). In a number of Oceanic languages, its meaning also includes the space beneath something, e.g. a house (see the NNG reflexes below), and it is probable that this extension of meaning had already occurred in POc. Its zero-valency forms are the source of local adverbs meaning ‘below, down there’ in a number of languages.

Scattered reflexes also suggest the reconstruction of monosyllabic forms without the first (reduplicated) syllable. An innovative monovalent form *pʷake- is reflected in Meso-Melanesian languages, apparently by the addition of *-e to the monosyllabic form *pak.

No reflexes occur in Central Pacific languages. In Fijian, *papa-, *papak has been ousted by reflexes of POc *ruku- ‘underneath’ (see below), in Polynesian languages by reflexes of POc *lalom ‘inside’ (p.247).

PMP		*babaq	‘lower surface, bottom, underside’ (ACD)
POc		*papa-, *pa-, *papak, *pak	[(N, N LOC) ] ‘underneath, lower surface, bottom, underside’
Adm	Mussau	pak(e)	‘underside’
Adm	Loniu	paʔaha-	‘underside’ (metathesis of *pahaʔa- < *papaqa- with unexplained final *-a-)
Adm	Drehet	pehe(kxa-)	‘underside’
NNG	Lukep	pa(rumu)	‘area under house’ (< POc *pak qi Rumaq ‘underneath of house’)
NNG	Dami	pa(rume)	‘under’ (< POc *pak qi Rumaq)
NNG	Bing	papa(rum)	‘under (a house)’ (< POc *papak qi Rumaq)
PT	Are	baba-	‘beneath’
PT	Gapapaiwa	vava-	‘beneath’
PT	Tawala	baba-	‘base, underneath, bottom; reason’
PT	Mekeo	papu-	‘under’
SES	Talise	vava-	‘below’
SES	Birao	vava-	‘below’
NCV	Nokuku	veva-n	‘underside’
NCV	Kiai	vova-na	‘underside’
NCV	Uripiv	(mel)ve-n	‘the underneath of it, the shade of it’ (*malu ‘shadow’)
NCV	Lonwolwol	fa-n	‘underneath’
NCV	Paamese	hehe-ne	‘underneath’
NCV	Nguna	na-ve(ruku)	‘underneath’
Mic	Kiribati	ā-	‘underside, underneath, bottom’
Mic	Ponapean	pā-	‘underneath’
Mic	Mokilese	pā-	‘underneath’
Mic	Chuukese	fā-	‘underneath’
Mic	Woleaian	fā-	‘underneath’
NCal	Cèmuhî	hāhî-n	‘underneath’
PMM		*pake-	‘underneath, underside’
MM	Bali	va-vake(ni)	‘down (there)’
MM	Tigak	pak(a-)	‘underside’
MM	East Kara	pa-	‘underside; down there’
MM	Notsi	pai-	‘underside’
MM	Tabar	paki-	‘underside’
MM	Lihir	pakie-	‘underside’
MM	Lihir	pek-	‘down (there)’
MM	Tangga	(ua-i)fafi	‘downwards’
MM	Konomala	fəi-	‘underside’
MM	Tolai	(na)vavai	‘under’
MM	Taiof	fai-	‘underside’
MM	Teop	pa-	‘underside’
MM	Roviana	(pana)peka	‘below’ (vowel metathesis)
MM	Vangunu	(pana)peka	‘below’ (vowel metathesis)
MM	Kia	peka	‘below’ (vowel metathesis)
MM	Laghu	peka	‘below’ (vowel metathesis)

Three other POc terms can be reconstructed with a meaning related to ‘underneath’ or with a denotation which has given rise to it in daughter languages.

Several reflexes of the first, POc *ruku-, are concatenated with a reflex of *pa- ‘underside’, the short form of *papa-. These seem to be compounds, implying that the meaning of *ruku- was perhaps more specific than that of *papa-. The latter was evidently the generic term for ‘underneath’. Perhaps *ruku- denoted the undersurface of something.

The second term, POc *gabʷari- meant ‘the area underneath a raised house’ (vol.1,51) and has come to mean ‘underneath’ in some languages by extension. POc *puqu-, puqun had the relational meaning ‘base, foundation’ when used in association with an object, as well as the more abstract meaning ‘origin, source, reason’.

POc		*ruku-	[(N, N LOC) ] ‘underneath, undersurface (?)’
MM	Bulu	luku(va)-	‘underside’ (-va < POc *pa- ‘underside’)
MM	Meramera	luʔu(va)-	‘underside’ (-va < POc *pa- ‘underside’)
MM	Nakanai	(lau)lu(va)-	‘underside’ (-va < POc *pa- ‘underside’)
MM	Nalik	ru	‘down (there)’
SES	Gela	(ru)ruɣu	‘below’
SES	Lau	rū	‘inside of roof’
SES	Kahua	ruɣu(ha)-	‘below’ (-ha < POc *pa- ‘underside’)
NCV	Tamambo	ruhu-ruhu	‘underneath part of s.t.’
NCV	Nguna	(na-ve)ruku	‘underneath’ (ve- < POc *pa- ‘underside’)
Fij	Wayan	ruku	‘underneath, under, below, space underneath’
Fij	Bauan	ruku-	‘space underneath’

POc		*gabʷari-	[(N, N LOC) ] ‘the area underneath a raised house’ (vol. 1, p.51)
Adm	Titan	kapʷaliŋ	‘area underneath a house’
NNG	Mapos Buang	ɣbi(ne)	‘underneath’
NNG	Mangga	kabi(ni)	‘underneath’
NNG	Patep	ŋbi-	‘underneath’
PT	Gumawana	gabula	‘underneath’
PT	Tawala	gaboli-	‘area underneath a house’
PT	Dobu	gabura	‘area underneath a house’
PT	Duau	gabule-	‘area underneath a house’
PT	Misima	gabúla	‘area underneath a house; underneath’
PT	Sinaugoro	gabule-	‘underneath’

PMP		*puqun	‘beginning, cause, origin, source, basis’ (ACD)
POc		*puqu-, *puqun	[(N, N LOC) ] ‘base, foundation, origin, source, reason’
Adm	Loniu	puʔu-	‘bottom, underside’
NNG	Tami	pu-	‘base, origin’
NNG	Mangga	kabi(ni)	‘underneath’
NNG	Takia	fu-n	‘bottoms’
NNG	Yabem	m̀	‘origin’13
NNG	Bukawa	(ŋa)pu	‘underside’
NNG	Mangga	vu	‘underside’
NNG	Wampar	fo(n)	‘origin’
NNG	Labu	(a)ho	‘base; bottom; reason’
NNG	Middle Watut	fogo	‘origin’
NNG	Wampur	hugu-n	‘trunk’
NNG	Adzera	fugu-n	‘tap-root; base’
PT	Bwaidoga	vu-vu-	‘cause, origin, foundation of anything; (tree) root’
MM	Tolai	vu-	‘beginning, cause, origin, source, basis, root, foundations’
SES	Longgu	vuʔa	‘below, down; a time before’
Fij	Wayan	-vū	‘base, bottom; origin, source, cause; taproot, tuber’

2.3.3. ‘Top, upper surface, space above’⇫

The basic POc term for ‘top, upper surface, space above’ is *papo[-], *po[-]. Blust (ACD, 1997) writes that PMP *babaw ‘upper surface, top’ is the antonym of PMP *babaq ‘lower surface, bottom’, and the same is true of their POc reflexes: POc *papo[-], *po[-] is the antonym of POc *papa-, *papak (p.249).

The unsuffixed forms are a source of local adverbs meaning ‘above, up there’ in a number of languages, although here POc *qulu[-] below is a close competitor.

The Kiribati (Mic) reflex of POc *papo[-] also has the meaning ‘outside’, and this is the sole sense of the Nemi (NCal) and Polynesian reflexes. Blust (ACD) suggests that (as with *lalom; p.246) this is the result of applying the term to the planar surface of the sea. In relation to the sea, *papo[-] was its surface and the space above it. This is beyond the land, hence ‘outside’ it.

PMP		*babaw	‘upper surface, top; above; highlands’ (ACD)
POc		*papo[-], *po[-]	[(N, N LOC) ] ‘upper surface, top’
Adm	Mussau	po(na)	‘top’
NNG	Tami	[ka]popo-	‘top (of s.t.)’
NNG	Tami	po	‘above’
NNG	Takia	fo	[POSTPOSITION] ‘on’
NNG	Numbami	wao-	‘above’
NNG	Yabem	aɔ̀	‘upwards’
NNG	Yabem	(ŋ)aɔ̀	‘upper surface’
NNG	Kela	baba	‘topside’
NNG	Mapos Buang	vavu	‘up top’
NNG	Mapos Buang	vavu(ne)	‘upwards’
NNG	Wampar	we(ŋ)	‘topside’
NNG	Yalu	waʋ(g)	‘topside’
NNG	Adzera	wagu(ŋ)	‘topside’
MM	Notsi	papa-	‘topside’
MM	Tabar	popo-	‘topside’
MM	Tangga	fo-	‘topside’
SES	Baegu	fafo(luma)	‘thatch’ (luma ‘house, building’)
SES	Lau	fafo-	‘top’
SES	Longgu	vavo-	‘top surface, space above’
SES	’Are’are	haho-	‘topside’
SES	Sa’a	haho-	‘above’
SES	Arosi	haho-	‘topside’
NCV	Mota	vawo	‘above, upon’
NCal	Nemi	pʷap	‘outside’
Mic	Kiribati	āo	‘upper part of, surface, outside, back’
Mic	Kiribati	(i)ao-	‘on, on top’
Mic	Kosraean	fɛ-	‘above, on’
Mic	Marshallese	ɛwɛ-	‘on; upon; top; surface; over’
Mic	Mokilese	pō-	‘on’
Mic	Chuukese	wɔ̄-ɾ	‘above, on it’
Mic	Puluwatese	wɔ̄-n	‘above, on it’
Mic	Woleaian	wɔ̄-	‘on, topside, upside’
Pn	Samoan	fafo	‘outside, out of doors, a place other than Samoa’
Pn	Rennellese	haho	‘outside’
Pn	Māori	waho	‘outside; open sea; coast, as opposed to inland’
Pn	Hawaiian	waho	‘outside, beyond, out, outer, outward’

Given the tendency for body-part terms to be used by metaphorical extension as relational nouns (cf. p.248), it is unsurprising that the word for ‘head’ and ‘head hair’, POc *qulu[-], also acquired the meaning ‘top’.

PAn		*qulu	‘head’ (ACD)
POc		*qulu[-]	[N] ‘head, (head) hair’; [N LOC] ‘top part’
Adm	Mussau	ulu (bo)	‘headwaters of a river’
Adm	Nauna	kulu(n puli	‘(mountain) peak’
NNG	Yabem	lo-lo(ʔ)	‘topside’
NNG	Bukawa	lu-lu(ʔ)	‘topside’
PT	Molima	ʔunu-ʔunu-	‘head, forehead; (river) source’
MM	Tigak	kuli-	‘top’
MM	Tigak	kul	‘up (there)’
MM	Tiang	kələ	‘topside’
MM	Nalik	kula	‘up (there)’
MM	Tabar	kulu	‘topside’
MM	Lamasong	kun	‘up (there)’
MM	Konomala	ulə	‘topside’
MM	Tolai	ul	‘head, hair, top, apex, crown’
MM	Roviana	ulu	‘top’
SES	Gela	ulu-	‘head, except of a chief; (in compounds) hair; eastern end, upper end’
SES	Bugotu	ulu	‘head, top end’
SES	Lau	ulu(nao)	‘first-born, elder, senior’
SES	Lau	ulu-ulu-	‘(tree) topmost branch’
SES	’Are’are	uru	‘cloud, heaven, sky, top’
NCV	Ambae	ulu-	‘top’
NCV	Tamambo	ulu-	‘top part’
NCV	Raga	ulu-	‘space above’
NCV	Paamese	(n)ulu(ŋout)	‘at the top of the garden’ (< POc *na qulu ni qutan ART top PREP bush)
NCal	Nemi	hule-n	‘top’
Mic	Kosraean	ulu-	‘top’
Fij	Wayan	-ulu	‘head or top part of an animal or thing’
Fij	Bauan	ulu-	‘head, top’
Pn	Tongan	ʔulu	‘head, upper end’
Pn	Samoan	ulu	‘head, hair’
Pn	Samoan	ulu(matua)	‘first-born, eldest child’
Pn	Māori	uru	‘head, (head) hair; chief; top, upper end; (weapon +) point’

In NCV languages we find reflexes of the adverb *qa-qulu ‘up there, up above’: NE Ambae a-ulu ‘up high, on top’, Tamambo a-ulu ‘on top, at the gardens’, Kiai aulu ‘above’.

2.3.4. ‘Side’⇫

It is tempting to look for a POc relational noun which would correspond in its use to the English preposition ‘beside’. However, in many of its English uses ‘beside’ denotes a relative location, and, as I noted in §1, speakers of Oceanic languages do not make use of relative locations. We would expect POc reconstructions corresponding to meanings of English ‘side’ to denote an intrinsic, not a relative, location, and consequently perhaps to denote a part of a particular object.

This expectation is at least partly fulfilled. We can reconstruct POc *[pʷa]pʷaRa[-] ‘side; cheek’, a body-part term whose primary meaning was probably ‘side of the face’. Its uses are analogous to those of POc *mata[-] ‘eye; face; front’ (p.249). However, a good deal of confusion surrounds reflexes of *[pʷa]pʷaRa[-]. Reflexes of POc *baban/*bapan ‘plank; canoe plank or strake’ (vol.1,185) are similar in both form and meaning to those of *[pʷa]pʷaRa[-]. Listed under ‘cf. also’ below *[pʷa]pʷaRa[-] are terms whose glosses include the meaning ‘side’ but whose forms reflect *baban. The Lau and Bauan reflexes have glosses which are associated with both items, suggesting conflation.

PCEMP		*papaR	‘cheek, temple, side,’ (ACD) 14
POc		*[pʷa]pʷaRa[-]	[(N) ] ‘cheek, side of head’; [N LOC] ‘side’
NNG	Kairiru	poreq	‘side of house’ (-q unexplained)
MM	Nalik	par, pāran	‘side’ (dialectal variants)
MM	Tolai	papar, papara-	‘side’
MM	Minigir	papara	‘side’
MM	Ramoaaina	papar	‘side’
MM	Kandas	papori	‘side’
MM	Taiof	pana	‘side’
MM	Mono-Alu	(pa)pala	‘side’
MM	Roviana	papara	‘side of face, cheek’

cf. also:

NNG	Rauto	vava-	‘side’
NNG	Mengen	vava-	‘side’
SJ	Sobei	popa	‘cheek’
SES	Kwaio	baba	‘side, cheek’
SES	Lau	baba	‘side; long side board of canoe’
SES	Arosi	baba	‘cheek, temples; side (of a stream +)’
Fij	Bauan	baba	‘side of s.t., cheek bone; side of a canoe’

POc *pʷala(ŋ) ‘side, part’ is reconstructable, but its exact sense is unclear.

PMP		*balaŋ	‘side, part’ (ACD)
POc		*pʷala(ŋ)	‘side, part’ (ACD)
MM	Tigak	pal	‘part’
SES	Gela	pala	‘side, part’
SES	Lau	bara-	‘side’
NCV	Mota	para	‘sideways, turning aside’
Pn	Tongan	pala	‘side, edge’

POc *bali denoted ‘one of two (opposing) sides or parts’.

PMP		*baliw	‘moiety; answer; oppose; partner, friend, enemy; opposite side or part’ (ACD) 15
POc		*bali[-]	[(N, N LOC) ] ‘one of two (opposing) sides or parts’
SES	Gela	bali	‘bring together (opposite planks of a canoe)’
SES	Kwaio	bali-	‘part, side, portion, half’
NCV	Mota	(ta)vali(u)	‘one of two sides or parts’
NCV	Raga	bal(si)	‘side’
NCV	Lonwolwol	wali	‘one of (a pair); the mate of’
Mic	Ponapean	pali	‘side’
Mic	Woleaian	pariy	‘side’
Pn	Tahitian	pari	‘side’
Pn	Tuamotuan	pari(a)	‘a half’

cf. also:

PT

Motu

badi(nai)

‘beside’ (-d- for expected **-r-)

There are two other reconstructions from which terms for ‘side’ are derived. One, POc *siriŋ ‘side, edge’, is derived from a PMP term whose basic meaning was apparently ‘be close to, be near to’. Only three Oceanic reflexes have been found to date. The other, PEOc *tapa- ‘side, outside’, is limited to Eastern Oceanic, where the earlier sense seems to have been ‘side’ in the sense of ‘outer surface other than front or back’.

PMP		*sidiŋ	‘border on, neighbour; peer, equal’ (ACD)
POc		*siriŋ	[(N, N LOC) ] ‘side, edge’
NNG	Dami	siri-	‘side’
NNG	Takia	siriŋe-	‘side’
Mic	Marshallese	tuṛᵚu-	‘beside’

PEOc		*tapa-	[(N, N LOC) ] ‘side, outside’
NCV	Ambae	tava(lu)-	‘side’
NCV	Tamambo	tava(lu)	‘side part of s.t.’
NCV	Paamese	tav	‘one side’
Pn	Tongan	tafa	‘edge, border’
Pn	Samoan	tafa	‘side’
Pn	Kapingamarangi	taha	‘outside, shore’
Pn	Rennellese	taha	‘outside, beside, near, edge, side’
Pn	Tuamotuan	taha	‘side, margin, edge, border’

2.3.5. ‘Outside’⇫

It is reasonably clear that the ‘inside’/‘outside’ opposition found in European languages did not occur in POc. This is unsurprising, since POc relation terms were nouns denoting parts of an object. The inside of a house is readily conceived as a part of it (POc *lalo-), but the English term ‘outside’ only denotes a part insofar as it refers to the external surfaces of the building. As noted at various points in §2.3, terms which denote (among other things) particular external surfaces are also used metonymically to denote the external surface or ‘outside’ of an object in general. This is true of some reflexes of POc *papo[-] ‘upper surface, top’ (p.252), of POc *muri[-] ‘back part, rear’ (p.261) and of PEOc *tapa- ‘side, outside’ (above).

The terms in the set below could tempt us to reconstruct POc *luku- ‘side, outside’.¹⁶ However, their uneven distribution is suspicious, and it seems far more likely that they reflect PMP *likuD, POc *liku(r) ‘(person’s) back’ (Blust 1981b). The meaning of scattered reflexes of this word has extended to include the backs and rear parts of inanimate objects (like POc *muri[-]), and thence the external surfaces of objects in general. The specification of ‘back’ in the Marshallese and Woleaian reflexes below supports this interpretation.

Yap	Yapese	(wu)ɽuʔ	‘outside of’
NNG	Manam	(e)luku	‘outside’
Mic	Kosraean	lɨkɨ	‘outside’
Mic	Marshallese	liki-	‘outside; ocean side of; behind, in back of’
Mic	Ponapean	liki-	‘outside’
Mic	Mokilese	liki-	‘outside’
Mic	Chuukese	ɾɨkɨ	‘exterior, outside, outside surface, outer edge, immediate environs’
Mic	Woleaian	rɨxɨ	‘outside, back of s.t.’
Fij	Wayan	liku	[N LOC] ‘back side of the island’

2.3.6. ‘Front, time before’⇫

According to Blust (1997), the PMP relational noun for ‘front’ was *qadəp ‘front’, which was also used of the human face. Although reflexes of this term have been replaced by body-part terms (see below) in a majority of Oceanic languages, enough reflexes survive to make it clear that its reflex POc *qaro-, *qarop ‘front; face’ is reconstructable.

PAn		*qadəp	‘front, face’ (Blust 1997)
POc		*qaro-, *qarop	[(N) ] ‘face’; [N LOC] ‘front’
NNG	Mangap	kere-	‘front’
NNG	Manam	aro-	‘space in front’
NNG	Kairiru	aro-	‘in front of (s.t.)’
PT	Dawawa	karo	‘in front’
SES	Sa’a	saro	‘face, turn oneself’
Fij	Rotuman	aro	‘front, side or surface that is usually seen.’
Pn	Tongan	ʔao	‘front’
Pn	East Futunan	ʔalo	‘in front’
Pn	Marquesan	aʔo	‘front’
Pn	Hawaiian	alo	‘front’
Pn	Māori	aro	‘front of body, pubic area of females’

None of the items above reflects final POc *-p, but we can be certain that the form *qarop occurred, as a reflex is preserved in PPn *ʔarofiwaʔe ‘sole of foot’ (e.g. Tongan ʔaofi vaʔe, East Futunan ʔalofi-vaʔe, Samoan alofivae), reflecting POc *qarop qi qaqe, literally ‘front of foot’ (where *qi is the non-specific possessive preposition (Ross 1998b, 2001b)).

Another generic POc term for ‘front’ (but probably not ‘face’) was *muqa[-], which — its reflexes suggest — occurred more often as a zero-valency than a monovalent noun. As a zero-valency local noun it occurred in the prepositional phrase *i muqa ‘in front, formerly’. The reduplicated form *muqa-muqa ‘in front, formerly’ represents a morphological pattern not found with other local nouns.

As the glosses indicate, POc *muqa[-] had the temporal sense of ‘time before’ as well as the local sense of ‘front’. It was thus the antonym of POc *muri[-] ‘back, time after’ (§2.3.7).

POc		*muqa[-]	[(N LOC) ] ‘front’
POc		*muqa	‘front; be in front’
POc		*i muqa, *qa-muqa, *muqa-muqa	[ADV] ‘in front, formerly’
Yap	Yapese	mʔōn	‘front’
Adm	Mussau	mua	‘front’
Adm	Titan	mo(ndrol)	‘bow of canoe’
NNG	Manam	mua	‘go first, precede’
MM	Vitu	muɣa-	‘front’
MM	Nalik	(pa)mua	‘in front’
MM	Notsi	(la)mua	‘front’
MM	Tabar	mu-mua	‘formerly’
MM	Tabar	moa	‘front’
MM	Lihir	(i)muo	‘formerly’
MM	Lihir	muo	‘in front’
MM	Solos	ma-mua	‘in front’
MM	Selau	(to)mua-na	‘old’
MM	Taiof	(i)mua-n	‘formerly’
MM	Teop	(ta)mua-na	‘old’
MM	Banoni	ma-ma	‘in front’
NCV	Mota	mʷoa-i	‘first, foremost, principal; to be first’
NCV	Raga	mua-i	‘first’
NCV	Ambae	mue-	‘front of’
NCV	Lewo	(va)mo	‘front’ (va ‘go’)
Mic	Nauruan	(ā)mʷō	‘front’
Mic	Kiribati	moa	‘front, fore part’
Mic	Woleaian	mmʷa-	‘front, first, tip, before’
Mic	Marshallese	mᵚā-	‘front’
Mic	Marshallese	mʷāha-	‘ahead of, before, in front of’
Mic	Mokilese	mʷō-	‘front’
Mic	Ponapean	mʷowɛ-	‘ahead of, in front of, before’
Mic	Chuukese	mʷ-mʷa-	‘in front of, more than’
Fij	Wayan	mua	‘end-point or tip of a long object; head for or set course for a place’
Fij	Bauan	mua	‘the first; tip, point, prow’
Pn	Tongan	muʔa	‘front’
Pn	Niuafo’ou	muʔa	‘front’
Pn	Rapanui	(ʔi) muʔa	‘front’
Pn	Tahitian	(i) mua	‘front’

In NCV languages we find reflexes of the adverb *qa-muqa ‘in front, formerly’: Mota (a)mʷoa ‘before, first’, Raga (a)mua ‘before, at first, first, in front of’, NE Ambae (a)mue ‘in front, at the front’, Port Sandwich (a)mo (POSTVERBAL ADV) ‘before’.

There is evidence in Western Oceanic languages for a competing form *muga. This occurs far more often in verbal reflexes than does *muqa, so it is possible that *muga was the POc verb, and that *muqa supplanted *muga in PEOc. The possibility of conflating the two terms is illustrated in Vitu and Bali, dialects of the Bali-Vitu language. In one, Vitu, the monovalent term for ‘front’ is muɣa-, reflecting *muqa, whilst in the other, Bali, it is muga-, reflecting *muga. Since reflexes of *muga are otherwise not monovalent, it is reasonable to infer that this is a conflation whereby the noun muɣa- has been replaced by the verb-form muga.

Although *muqa and *muga are formally similar, a historical relationship between them is problematic: we would expect an alternation between POc *k and *g, but not between *q and *g.¹⁷

PWOc		*muga	[(?? N LOC, V) ] ‘front; be in front; formerly’
NNG	Bariai	muga	‘front’
NNG	Bariai	muga(ŋa)	‘forehead’
NNG	Bariai	muga(eai)	‘formerly’ (-eai POSTP)
NNG	Lukep	mugu	‘first of all, formerly’
NNG	Mangap	muᵑgu	‘first of all, formerly, long ago’
NNG	Gitua	muŋga	‘precede, go ahead, future’
NNG	Sio	muga	‘precede; before’
NNG	Tami	muŋ	‘front; in front’
NNG	Bing	mug	‘formerly’
NNG	Gedaged	mug	‘precede’
NNG	Yabem	muŋ	‘precede’
NNG	Adzera	moŋʔ	‘prior’
NNG	Adzera	moŋʔ(an)	‘precede’
PT	Suau	-muga(i)	‘precede’
MM	Bali	muga-	‘front’
MM	Bola	muga	‘frontside’
MM	Nakanai	ma-muga	[RELATIONAL N] ‘front’
MM	Ramoaaina	(nə)mugə	‘in front; formerly’
MM	Kandas	mugu	‘in front’
MM	Bilur	mugo	‘frontside’
MM	Siar	muŋ	‘in front of’

The body-part terms whose reflexes are often used for ‘front’ are POc *mata[-] ‘eye, face, front’ and POc *nako[-] face, front’. The original and basic meanings of *mata[-] and *nako[-] were ‘eye’ and ‘face’ respectively. Nonetheless, reflexes of these terms occur with great frequency in the meaning ‘front’. Scattered reflexes below suggest that *i mata was a POc expression meaning ‘in front’, and other modern uses suggest that it has long been used for the front of an inanimate object, e.g. Nalik (MM) la maran a fal [PREP eye PREP house] and Tolai (MM) ta ra mata-na pal [PREP ART eye-P:3SG house], both ‘in front of the house’.

PAn		*maCa	‘eye’
POc		*mata[-]	[(N)] ‘eye; face’; [N LOC] ‘front’
NNG	Lusi	mata-	‘eye; front’
NNG	Bariai	mata	‘eye; front’
NNG	Mangap	mata-	‘eye; front’
NNG	Takia	mala-	‘eye, front’
NNG	Buang	mala	‘eye, front’
NNG	Adzera	mara-	‘eye, front’
NNG	Kaulong	(e)mara	‘in front’
PT	Kilivila	mata-	‘eye; front’
MM	Nalik	mara-	‘eye; front’
MM	Siar	mata-	‘eye; front’
MM	Tolai	mata	‘eye; front’
MM	Taiof	mata-	‘eye; front’
SES	Gela	(i)mata	‘in front of’
SES	Sa’a	mā	‘eye; front’
SES	Sa’a	(i)mā	‘outside’
NCal	Tîrî	(ɳā)mʷãɽã	‘front’
Mic	Kosraean	mʌtʌ-	‘eye; front’
Fij	Wayan	mata-	‘face, front of head, face of object with both front and back side’
Fij	Wayan	(i)mata	‘in front’
Fij	Bauan	mata-	‘eye; face; front’
Fij	Bauan	(i)mata	‘in front’
Pn	Tongan	mata	‘eye, face’
Pn	Samoan	mata	‘eye, face’
Pn	Māori	mata	‘eye, face’

POc		*nako[-]	[(N, N LOC)] ‘face, front’
Adm	Pak	nogo(gi)	‘front, before, face’
NNG	Gitua	nago	‘face’
NNG	Tami	nao	‘front, face’
NNG	Takia	nao-	‘face’
NNG	Takia	nao(-n na)	‘in front of’ (N-P:3SG POSTPOSITION)
PT	Ubir	na(-na-i)	‘in front of it, him’
PT	Tawala	nao-	‘front/forward position’
PT	Tawala	u nao-na	‘in front’ (PREP N-P:3SG)
MM	Lavongai	(ai)no	‘formerly’
MM	Lavongai	no	‘forehead; frontside’
MM	Tigak	(ai)no	‘formerly’
MM	Tigak	no	‘forehead; frontside’
MM	West Kara	no	‘forehead’
MM	Nalik	no	‘forehead’
SES	Gela	naɣo	‘front, before, face’
SES	Bugotu	naɣo	‘front’
SES	Longgu	naʔo(va-)	‘front’
SES	Lau	nao	‘front’
SES	Kwaio	naʔo-	‘front’
SES	’Are’are	naʔo	‘front’
SES	Sa’a	naʔo	‘front, before, face’
NCV	Mota	nago-i	‘face, front, cutting edge’
NCV	Merlav	nago-i	‘front, before, face’
NCV	Tamambo	naho-	‘face’
NCV	Paamese	nā-	‘face, front’
NCV	Nguna	nako-	‘front, before, face’

2.3.7. ‘Back, space behind, time after’⇫

The generic POc term for the back (of something or someone), the space behind (something or someone), and the time after (an event) was *muri[-]. However, it has a more complex history than other POc relational nouns. Blust (ACD) derives it from PMP *ma-udehi, containing the PMP undergoer-subject verbal prefix *ma- and the root *udehi which he glosses as ‘last; come after or behind; late, later; future; stern of a boat; youngest child.’ Blust’s glosses are not intended to be a claim about the morpholexical class of the item, but it is a reasonable inference from work on the history of PMP and POc *ma- (Evans and Ross 2001) that PMP *udehi was a noun, perhaps meaning ‘that which is behind, that which is last, that which is after or in the future’ and that *ma-udehi was a stative (adjectival) verb derived from it.

There is evidence in the Gapapaiwa (PT), Ramoaaina (MM), Arosi (SES), Bauan (Fij), Samoan (Pn), and Rennellese (Pn) definitions below that POc *muri remained a stative verb, but there is also overwhelming evidence that it was a monovalent relational noun with spatial meanings like ‘back part, rear, behind, space to the rear of, time after’ as well as more concrete uses like ‘stern of a canoe’. It is also glossed as an adverb of place and/or time in a number of languages, but where there is evidence about its morpholexical class, these uses derive from its nominal use with a preposition, suggesting POc *i muri ‘behind, later’ (more literally, ‘in the space behind, at a time after’). POc *muri thus also had a temporal use, referring to time after the time of speaking (p.322). Note that the syntactic behaviour of POc *muri[-] broadly matches that of its antonym *muqa[-]/*muga (§2.3.6), which also had both nominal and verbal uses.

It appears that the PMP root *udehi was also inherited into POc as the base *uri, but only two reflexes have been found. They are both in NNG languages: Gedaged uli ‘follow, pursue; come after, succeed; go to the rear’ and Kaulong e-uli- ‘back’ (e- ART).

PMP		*ma-udehi	‘be last; be after or behind; be late, be later; future’ (ACD) 18
POc		*muri[-]	[(N, N LOC) ] ‘be behind, be after; back part, rear, behind, space to the rear of, time after; (canoe) stern; space outside’
POc		*i muri, *muri-muri	‘at the back, later’
Adm	Wuvulu	muki	‘(canoe) stern’
Adm	Loniu	muʔu (tun)	‘(canoe) stern’
Adm	Drehet	(o)mu(ŋ)	‘back’
NNG	Kove	muhi-	‘s.o.’s back’
NNG	Bariai	mur[-]	‘s.o.’s back’
NNG	Gitua	mur	‘behind, afterwards’
NNG	Gedaged	muli-	‘behind, rear, back part, stern, rear, posterior, outside of s.t.’
NNG	Manam	muri	‘behind’
NNG	Yabem	(ŋa)mu(ŋa)	‘back of s.t.’
NNG	Bukawa	(ŋa)ᵐbu(ᵑga)	‘back of s.t.’
NNG	Kaiwa	mul	‘back of s.t.’
PT	Iduna	muli(ne)	‘back of s.t., behind’
PT	Dobu	muri-	‘behind, afterwards’
PT	Gapapaiwa	muri	‘follow’
PT	Gapapaiwa	muri-	‘back of s.t.; behind, afterwards’
PT	Tawala	muli-	‘back of s.t.; behind, afterwards’
PT	Motu	muri-	‘back of s.t.; space behind’
MM	Bali	muri	‘back of s.t.’
MM	Meramera	(ma)muli	‘back of s.t.’
MM	Meramera	(muli)muli	‘later’
MM	Nakanai	(muli)muli	‘later’
MM	Lavongai	muŋ	‘back of s.t.; s.o.’s back’
MM	Tigak	(ai)muk	‘later’
MM	Tigak	mugi-	‘back of s.t.; s.o.’s back’
MM	Tabar	muri-	‘back of s.t.’
MM	Ramoaaina	muru	‘follow; behind, back; last’
MM	Ramoaaina	(na)mur	‘later, afterwards’
MM	Ramoaaina	mur	‘s.o.’s back’
SES	Gela	muri-	‘behind, afterwards; back; outside of s.t.; afterbirth; posterity’
SES	Lengo	(i)muri(a)	‘after’
SES	Arosi	muri-	‘follow; behind, back; outside of s.t.; afterwards; left hand when facing an object’
Mic	Ponapean	mʷuri	‘behind’
Mic	Woleaian	mʷiẓi-	‘behind, after, backside, rear’
Mic	Mokilese	mʷeri-	‘back of s.t.; s.o.’s back’
Mic	Puluwatese	mʷir-	‘back of s.t.; s.o.’s back’
Fij	Bauan	muri	‘following, after’
Fij	Bauan	(e) muri	‘behind, later’
Fij	Bauan	(ki) muri	‘to the rear’
Pn	Tongan	mui	‘space behind; rear; end, extremity, tip; back, rear; later; young, immature, only partly developed’
Pn	Samoan	muli	‘come last, be last; young, new’
Pn	Rennellese	mugi	‘follow, be or go behind or after; rear end, esp. lower or western end’
Pn	Māori	muri	‘rear, hind part; sequel, time to come; behind, afterwards, backwards; youngest child’
Pn	Hawaiian	muli	‘behind, afterwards; last, following behind; younger, youngest; (canoe) stern’

The reflexes below contain a Northwest Solomonic innovation whereby Proto Northwest Solomonic *mudi[-] is reconstructable (this would reflect POc *mudri) instead of expected **muri[-].

Proto Northwest Solomonic		*mudi-	‘back of s.t.; s.o.’s back’
MM	Nehan	mudi	‘back of s.t.; s.o.’s back’
MM	Petats	muru	‘s.o.’s back’
MM	Halia	muru	‘back of s.t.; s.o.’s back’
MM	Selau	muri-	‘back of s.t’
MM	Selau	mur	‘s.o.’s back’
MM	Banoni	muri	‘behind’
MM	Mono-Alu	(muri)muri	‘later’
MM	Vangunu	(tara)meji-na	‘after’
MM	Varisi	(tara)muzi-na	‘after’
MM	Nduke	mudi-	‘back of s.t.; s.o.’s back’
MM	Roviana	mudi-	‘back of s.t.; s.o.’s back’

The semantic and formal similarity of the reflexes of POc *burit below to those of POc *muri[-] above is evidently due to chance. In the 2003 version of this chapter, I attributed members of the set below to a putative PMP *pa-udehi, paradigmatically related to PMP *ma-udehi (ancestral to POc *muri[-]), but the presence in this set (listed in the ACD) of Bugotu buriti indicates that I was wrong.

PMP		*burit	‘hind part, rear, back’ (ACD)
POc		*burit	‘hind part, rear, back’; [N, N LOC] ‘back part, rear, behind, space to the rear of, time after; (canoe) stern’; [ADV] ‘behind, afterwards’
NNG	Kela	ᵐburi(ya)	‘back of s.t.’
MM	Tinputz	puri	‘behind’
MM	Teop	buri	‘behind’
SES	Lau	buri	‘back, stern’; ‘back; behind, after; stern, rear’
SES	Bugotu	buriti	‘back’
SES	Longgu	buri-	‘behind; after’
SES	Lau	buri	‘back, stern’; ‘back; behind, after; stern, rear’
SES	Lau	buri(wela)	‘after-birth’
SES	Lau	(i) buri	‘afterwards’
SES	Kwaio	buli-na	‘after’
SES	Kwaio	buli	‘after, behind’
SES	’Are’are	puri-na	‘after’
SES	Sa’a	(i) puri	‘back of, behind; stern of a canoe’
SES	Sa’a	puri-na	‘after, back, stern’

One body-part term occurs with fair frequency with the sense of ‘back part of, space behind’. This is POc *takuRu[-] ‘(s.o.’s) back’. The evidence that this was a body-part term in POc is clear. It may also have been used by extension as a POc relational local noun, but it is also possible that local-noun uses in modern languages represent independent parallel developments.

POc		*takuRu[-]	[(N, ? N LOC) ] ‘(s.o.’s) back’
Adm	Titan	lákulo-	‘(s.o.’s) back’ (l- for expected t-)
NNG	Sio	taulo	‘behind’
PT	Gumawana	tolu-	‘(s.o.’s) back’
PT	Dawawa	tauri	‘back of s.t.; s.o.’s back’
PT	Motu	doru-	‘back, behind’
MM	Lavongai	toŋ	‘back of s.t.’
MM	Nalik	toru-	[N LOC] ‘space behind’ (e.g. la toru-gu [PREP N LOC-P:1SG] ‘behind me’)
MM	Minigir	tauru-	‘(s.o.’s) back’
MM	Bilur	taru-	‘(s.o.’s) back’
MM	Siar	taru-	‘(s.o.’s) back’
MM	Taiof	touno-	‘(s.o.’s) back’
MM	Teop	tonu-	‘(s.o.’s) back’
MM	Kia	taɣuru-	‘back of s.t.; s.o.’s back’
MM	Kokota	tagru-	‘back of s.t.; s.o.’s back’
MM	Maringe	tʰagru-	‘back of s.t.; s.o.’s back’
NCV	Mota	tawur, tawuru-	‘back of s.t.; s.o.’s back’
NCV	Mota	(a)tawur	‘behind’
NCV	Ambae	tagu-	[N LOC] ‘space behind’
NCV	Raga	(a)taɣu-	[N LOC] ‘behind’
NCV	Port Sandwich	(a)rax	[N LOC] ‘behind’
NCV	Lonwolwol	tao-	‘lower back (region around hips); behind’
NCV	Lewo	ra-	‘back of s.t.; s.o.’s back’
NCV	Lewo	(va)rau	‘behind’ (va ‘go’)
NCV	Nguna	(na)taku	‘back; the far side, other side’
NCV	Nguna	(e)daku	[ADV] ‘at the back, behind; after’
SV	Sye	(n)toc(-noki)	‘back of skull’
SV	Sye	(n)tocu(-nta-)	‘shoulder blade’
SV	Kwamera	taku-	‘back’
SV	Anejom̃	(i)taɣ	[ADV] ‘behind’ (e.g. itaɣ a niomʷ [ADV PREP N] ‘behind the house’)
NCal	Nyelâyu	dū-	‘(s.o.’s) back; behind’
NCal	Tîrî	ʈɔɔ-	‘(s.o.’s) back’
Mic	Kiribati	akū-	‘back; behind’
Mic	Kosraean	tɔkɔ-	‘back of s.t.; s.o.’s back’
Mic	Marshallese	æliki-	‘(s.o.’s) back’
Mic	Puluwatese	hækɨr	‘(s.o.’s) back’
Mic	Woleaian	taxɨẓɨ-	‘back of s.t.; s.o.’s back’
Fij	Wayan	takū	[N LOC] ‘behind’
Fij	Bauan	daku-	‘back of s.t.; s.o.’s back’

cf. also:

SES	’Are’are	kokoru-	‘(s.o.’s) back’
SES	Sa’a	kokolu-	‘(s.o.’s) back’

2.4. The interrogative local noun ‘where?’⇫

The interrogative local noun ‘where?’ was POc *pai. Micronesian reflexes of *i pai reflect Proto Micronesian *i-fā rather than expected *i-fai.

PMP		*pai	‘where?’ (ACD)
POc		*pai, *i pai	[(N LOC) ] ‘where at?’
NNG	Bebeli	ehae	‘where?’
NNG	Numbami	ai(a)	‘where?’
MM	Bali	ve(ni)	‘where?’
MM	Bola	vai	‘where?’
MM	Meramera	(i)va	‘where?’
MM	Nakanai	-ve	‘where?’
MM	Tigak	ve	‘where?’
MM	East Kara	fa	‘where?’
MM	Nalik	fa	‘where?’
MM	Tabar	ve	‘where?’
MM	Lihir	he	‘where?’
MM	Sursurunga	ai, ai(ə)	‘where?’
MM	Patpatar	he	‘where?’
MM	Minigir	va	‘where?’
MM	Tolai	ve	‘where?’
MM	Ramoaaina	(ə)wai	‘where?’
MM	Teop	(ha)ve	‘where?’
MM	Banoni	vai	‘where?’
MM	Uruava	vei(a)	‘where?’
MM	Lungga	pai	‘where?’
MM	Roviana	(pa)vei	‘where?’
MM	Kia	hae	‘where?’
SES	Gela	(i)vei	‘where?’
SES	Longgu	evei	‘where?’
SES	Lau	(i)fai	‘where?’
SES	Kwaio	(i)fai	‘where?’
SES	Arosi	(naʔi)hei	‘where?’
SES	Bauro	(i)hai	‘where?’
NCV	Raga	(be)he	‘where?’
NCV	Uripiv	(ni)be	‘where?’
NCV	Lonwolwol	be	‘where?’
NCV	Paamese	(e)vē	‘where?’
NCV	Lewo	pe	‘where?’
NCV	Namakir	(-o)be-	‘where?’
Mic	Chuukese	(i)fa	‘where? how? what?’
Mic	Puluwatese	(yi)fa	‘where? what? which?’
Mic	Satawalese	(i)fa	‘where (is it)? which?’
Mic	Carolinian	(i)fa	‘where?’
Mic	Woleaian	(i)fā	‘where? which? what?’
Fij	Bauan	vei	‘where?’
Fij	Wayan	vei	‘where?’

Also found are forms which appear to reflect *pea, *pia and, in Polynesian, PPn *fē. These probably reflect POc *pai-a (cf. Numbami aia, Sursurunga aiə, Uruava veia above), together with vowel sequence reductions which have occurred independently but in parallel. The step from POc *paia to *pea is an obvious one. In a number of languages the height distance between the vowels of *pea has been maximised, giving *pia. And in Polynesian, an innovation which is regular in Tongic and sporadic in some other Polynesian languages apparently produced *fē as an alternant to *fea (< *pea < *paia). The forms are listed below. Where a reconstruction is preceded by a question mark, the forms beneath it may be the result of parallel developments.

Adm	Mussau	bea	‘where?’
NCV	Mota	vea	‘where?’
NCV	Kiai	vea	‘where?’
NCV	Tamambo	(a)bea	‘where?’
Pn	Tongan	fē	‘where?’
Pn	Niuean	fē	‘where?’
Pn	Samoan	fea	‘where?’
Pn	Anutan	pea	‘where?’
Pn	East Futunan	fea	‘where?’
Pn	East Uvean	fea	‘where?’
Pn	Tikopia	fea	‘where? what? when?’
Pn	Ifira-Mele	(i)fea	‘where at?’
Pn	Hawaiian	hea	‘where?’
Pn	Māori	ɸea	‘where?’
Pn	Tahitian	hea	‘where?’
Pn	Kapingamarangi	hē	‘where’
NCV	Tirax	fē	‘where?’
Pn	Nukuria	ihē	‘where?’
MM	Nehan	ia	‘where?’
MM	Solos	īa	‘where?’
MM	Petats	īa	‘where?’
MM	Taiof	ifia	‘where?’
MM	Mono-Alu	hi(na)	‘where?’
MM	Nduke	(o)via	‘where?’
SES	Gela	via	‘wherever, of whatever kind, where, what, which’
NCV	Port Sandwich	(a)ᵐbi	‘where?’
Pn	Māori	hia	‘where?’
Pn	Tahitian	hia	‘where?’

3. Directional verbs⇫

Directional verbs can be conveniently divided into verbs of deictic direction (‘towards speaker’, ‘towards addressee’, ‘away from speaker and addressee’) and verbs of geographic direction, and especially vertical direction (‘go up’, ‘go down’).

3.1. Some Proto Oceanic serial verb constructions⇫

Directional verbs play an important role in certain serial verb constructions in Oceanic languages, and they evidently did so in POc, to judge both from the wide distribution of such constructions today and from grammaticised versions of these constructions (Ross 2003).

Verbs of deictic direction occur in serial verb constructions of deictic direction, where they follow a verb of locomotion (transitive or intransitive) or a verb of geographic direction. The examples below are from Yabem (NNG). In both the first two examples, the deictic directional verb is -yà ‘go away from speaker and addressee’.¹⁹ In the first example it follows the locomotion verb -lob ‘fly’, in the second the geographic directional verb -pi ‘go up’.

• • Yabem (NNG)

    ‘The doves flew off to the mountain.’

    balosi	ge-lob	ge-yà	loʔ
    dove	S:3SG-fly	S:3SG-go:3	mountain

• • Yabem (NNG)

    ‘He climbed up to the men’s house.’

    ke-pi	lom	ge-yà.
    S:3SG-go.up	men’s.house	S:3SG-go:3

Verbs of geographic direction also occur in serial verb constructions of geographic direction, where they follow a locomotion verb (transitive or intransitive). In this example, the locomotion verb is -pwanɛʔ ‘insert’, the geographic directional verb -sep ‘go down’.

• • Yabem (NNG)

    ‘I’ll put the taro into my mouth.’

    ya-pwanɛʔ	mɔ	e-sep	aò-ʔ-sùŋ
    S:1SG-IRR:insert	taro	S:3SG-IRR:descend	mouth-P:1SG-hole

Commonly the two constructions are combined, giving a sequence of locomotion verb, geographic directional verb and deictic directional verb, e.g. -ne ‘sink’, -sep ‘go down’ and -yà ‘go away from speaker and addressee’ in this example.

• • Yabem (NNG)

    ‘The canoe sank into the sea.’

    waŋ	ge-ne	ke-sep	gweʔ	ge-yà
    canoe	S:3SG-sink	S:3SG-descend	sea	S:3SG-go:3

Directional verbs, both deictic and geographic, also occur in sequential serial verb constructions, where the first verb is a directional verb expressing ‘go [up/down] and …’ or ‘come and …’, the second a verb expressing the main event of the predication. This example is from Bali (MM):

• • Yabem (NNG)

    ‘They will go and catch fish.’

    Hizi	mi=ri	zio	ki	vahi-aŋa	ihaŋa.
    they	IRR=HYP:3	go.down	SEQ:3	get-PL	fish

3.2. Grammaticisations of serial verb constructions⇫

Directional verbs are grammaticised in a number of ways in Oceanic languages (Lichtenberk 1991). Three of these grammaticisation paths give rise to morphemes expressing location and direction. Reflexes of directional verbs which have undergone these grammaticisations occur in the cognate sets below, and for that reason are described here.

In the first type of grammaticisation, a directional verb in a serial verb construction loses its subject proclitic/prefix and becomes a directional adverbial enclitic (glossed DIR in cogate sets). In the two Sisiqa (MM) examples below, the directional enclitics =me and =la reflect the POc deictic directional verb forms *ma ‘come’ and *la ‘go:2’ (§3.4) respectively. Each is preceded by a locomotion verb (‘carry’, ‘walk’), reflecting an earlier serial verb construction of deictic direction.

• • Sisiqa (MM)

    ‘I have brought some coconuts.’

    ra	ko-gisu=me	kavia	kuda
    I	S:1SG:REAL-carry=hither	some	coconut

  • Sisiqa (MM)

    ‘S/he is going to Susuka village.’

    ɣōi	ma-zo=la	Susuka
    s/he	S:3SG:IRR-walk=thither	Susuka

In this Manam (NNG) example there is a sequence of locomotion verb (‘take’) and two directional enclitics, -raʔe ‘up, to one’s right when facing sea’ (< geographical directional verb POc *sake ‘go up’, p.273) and -laʔo ‘away’ (< deictic directional verb POc *lako ‘go:3’, p.287) reflecting an earlier three-verb sequence (Lichtenberk 1983:576–582).

• • Manam (NNG)

    ‘He took the axes away upward.’

    Ogi	i-doʔ-i-raʔe-laʔo.
    axe	S:3SG-take-O:3PL-upward-away

In the second type of grammaticisation, described by Pawley (1973) and Durie (1988), a directional verb in a serial verb construction is reanalysed as a preposition or a relator (see below) and comes to form a constituent primarily with the following locative expression. Prepositional reflexes of POc *mai ‘come’ occur in Polynesian languages. In Samoan, for example, we find (Mosel & Hovdhaugen 1992:147):

• • Samoan (Pn)

    ‘The boy jumped down from the horse …’

    Na	oso	i	lalo	le	tama	mai	le	solofanua	…
    PAST	jump	PREP	down	ART	boy	PREP	ART	horse	…

In Meso-Melanesian languages of New Britain and New Ireland and in Longgu (SES), the deictic directional verbs *mai ‘come’ and *ua ‘go:2’ have become respectively ablative and allative relators (Ross 2003). I use the term ‘relator’ for a preposition-like morpheme which differs in its distribution from a preposition in that it precedes either a prepositional phrase or a local noun. This distribution reflects its verbal ancestry: a POc deictic directional verb could be followed by a locative expression consisting of a local noun or a prepositional phrase. Hence in Longgu, vu is the allative relator reflecting *ua:

• • Longgu (SES)

    ‘… and they went into the bush’

    …	m-ara	la	maʔa	vu	masuʔu
    …	and-S:3PL	go	PERFECTIVE	R	bush

  • Longgu (SES)

    ‘we will go into the garden to weed it’

    amalu	ho	la	vu	ta-na	malaba-i	ni	umwani-a
    D:1EP	IRR	go	R	PREP-P:3SG	garden-SG	in.order.to	weed-O:3SG

In the third, least widespread, grammaticisation type, the deictic directional verb in a sequential serial verb construction becomes a pre-verbal clitic indicating the location or direction of the event in relation to the speech act participants. In the best described case, Sinaugoro (PT) (Tauberschmidt 1999:31–32), the clitics are enclitics to the preverbal subject/aspect/mood marking complex. In this example =ma reflects POc *ma ‘come’.

• • Sinaugoro (PT): (Tauberschmidt 1999:31–32)

    ‘Let’s eat here.’

    Si=ma	ɣani-ɣani.
    S:1IPL=near.me	ITR-eat

3.3. Geographic directional verbs and enclitics⇫

Geographic direction verbs occurred both independently and in geographic directional serial verb constructions. From the latter usage, they have often developed into enclitics or adverbs marking geographic direction.

The main semantic domain of geographic direction verbs is that of vertical direction, downward and upward. Vertical direction terms have developed two kinds of secondary meanings in Oceanic languages, and these were probably present in Proto Oceanic. First, ‘downward’ and ‘upward’ often have the secondary horizontal senses ‘to the northwest’ and ‘to the southeast’. Second, because Oceanic speakers often dwell on mountainous islands, in some languages ‘downward’ also means ‘seaward’, and ‘upward’ also means ‘inland’ or, from the sea, ‘landward’.

A pair of antonyms occurred in POc. These were the generic verbs of movement down and up:

*sipo ‘go downward’

*sake ‘go upward’

As I noted earlier (p.229), POc apparently had a subsystem of geographic direction which was based on a river valley and had an up/down axis and a transverse axis with one directionally neutral (‘across the valley’) term. The terms used for ‘down the valley’ and ‘up the valley’ were evidently *sipo and *sake. The transverse term was possibly *pano, which also served as a verb of deictic direction and is reconstructed on p.289. However, there are no known Western Oceanic or Southeast Solomonic reflexes of *pano with this sense, so this may be a later innovation.

Recent work by François (2003, 2004a) suggests strongly that this subsystem was also applied metaphorically to directions at sea. The two cardinal directions at sea were evidently provided by the major winds, POc *apaRat, the northwest storm wind, and *raki, the southeast trade wind, as the reflexes below (repeated from Chapter 5, §4.2) suggest:

PMP		*habaRat	‘west monsoon’ (ACD; Dempwolff 1938)
POc		*apaRat	‘northwest wind; wet season when northwesterlies blow and sea is rough’
Adm	Wuvulu	afā	‘northwest wind’
Adm	Drehet	yaha	‘stormy season, generally from November to March; strong wind and rough sea from the northwest’
NNG	Gitua	yavara	‘north wind’
NNG	Tami	yawal	‘northwest wind’
NNG	Kairiru	yavar	‘northwest wind, makes sea rough’
PT	Muyuw	yavat	‘west, west wind’
PT	Iduna	yavalata	‘rains with wind from the northwest in February and March’
PT	Motu	lahara	‘northwest wind, season of northwest wind’
MM	Bali	vurata	‘northwest wind’

POc		*raki	‘southeast trades’ (probably also ‘dry season when the southeast trades blow’)
Adm	Lou	ra	‘northeast, northeast wind’
Adm	Titan	ⁿray	‘wind from the mainland, mountain breeze, blows at night’
NNG	Kove	hai	‘southeast trade, year’
NNG	Gitua	rak	‘southeast trade’
NNG	Tami	lai	‘southeast trade’
NNG	Maleu	na-lai	‘southeast trade’
NNG	Ali	rai	‘southeast trade’
NNG	Tumleo	riei	‘southeast trade’
MM	Vitu	raɣi	‘southeast trade’
MM	Bulu	laɣi	‘southeast trade’
NCV	Lewo	lagi(pesoi)	‘east wind’
Mic	Marshallese	ṛᵚak	‘south, summer’
Pn	East Uvean	laki	‘southeast or southwest wind’
Pn	Niuean	laki	‘west’
Pn	Samoan	laʔi	‘southwest veering to northwest’

After examining the sea-based directional systems of a sample of Oceanic languages, François concludes that in POc ‘go down’ apparently had the secondary sense ‘go northwest’, whilst ‘go up’ had the secondary sense ‘go southeast’. More tentatively, he suggests that *pano may have been used for movement across the northwest–southeast axis. He suggests that the basis of this metaphor was that sailing into the wind felt to the sailors like going uphill.

In Ross (1995c) I suggested that *sake ‘go up’ and *sipo ‘go down’ were used by POc speakers to denote ‘east’ and ‘west’, i.e. the locations of sunrise and sunset. This inference was based on the fact that the glosses for their reflexes in many Oceanic languages are given as ‘east’ and ‘west’ (this is also true of some of the sources that François 2004a cites). However, François argues in his detailed account of Mwotlap directional systems that this is semantically implausible, as reflexes of *sake and *sipo are used for ‘go southeast’ and ‘go northwest’ respectively, and it is hard to see how these meanings — or ‘go east’ and ‘go west’ — could be derived from ‘go to the place where the sun rises/sets’ (François 2003). In François (2004a) he also presents the systems of a number of languages which display the ‘go up/southeast’ and ‘go down/northwest’ correlations. I find his reconstruction of a terminological subsystem corresponding to the major wind directions convincing, and I think it likely that systems which are oriented to the rising and setting of the sun are probably more recent developments.

François (2004a) is a reconstruction of a POc terminological subsystem, i.e. a system of meanings and the relationships among them. The languages in his sample by no means all use reflexes of *sipo and *sake for ‘go downward’ and ‘go upward’, and he makes no attempt to reconstruct the POc forms, assuming that the relevant POc etyma were *sipo and *sake. I return briefly to the reconstruction of *sipo and *sake as directions at sea in §3.3.3 below.

3.3.1. Downward movement⇫

Three possible verbs of downward movement are reconstructed below. They are:

POc	*sipo	‘go down, downwards’
POc	*sobu	‘go downward, dive down’
POc	*surup	‘(?) enter, penetrate; go down’

The most widely reflected of these is *sipo, the generic verb of downward movement. POc *sobu seems also to have carried the meaning ‘dive down’, as several of its reflexes have to do with action in the sea. I also include POc *surup ‘enter, penetrate, go down’ here on account of reflexes with the gloss ‘go down’, but the latter are found only in Meso-Melanesian and Southeast Solomonic languages, and I question whether it had this sense in POc.

POc		*sipo	‘go down, downwards’
Adm	Mussau	sio	‘go downward’
Adm	Mussau	(la-)sio	‘go down (to)’
Adm	Mussau	(la-)sio(-kasu)	‘come down (from)’ (kasu ‘go from’)
NNG	Kove	(i)ðio	‘go downward’
NNG	Bariai	(ga)dio	[DIR] ‘downward’
NNG	Gitua	zio(vave)	[DIR] ‘downward’
NNG	Tuam	(i)zi(la)	‘sink’
NNG	Yabem	siʔ	[DIR] ‘downward’
NNG	Bing	siy	‘come’
NNG	Takia	-s(-la)	‘go seaward, move downhill; land, arrive (of a boat)’
NNG	Manam	-ria	[DIR] ‘downward; to one’s left when facing sea’
NNG	Kaiep	(a)si	[DIR] ‘downward’
SJ	Sobei	-si	[DIR] ‘downward’
PT	Tawala	-hi	[DIR] ‘towards addressee’
PT	Sinaugoro	(va-)riɣo	‘go down’ (-riɣo occurs as the second element of verbal compounds)
MM	Kia	hi(nae)	‘go down’
MM	Laghu	hi(nae)	‘go down’
SES	Lengo	ðivo	‘go down’
SES	Longgu	sivo	‘go down’
SES	Kwaio	sifo	‘go down’
SES	Sa’a	siho	[DIR] ‘downward’
NCV	Merei	sio	‘move downward/seaward’
NCV	Tamambo	jivo	‘go down’
SV	Sye	-sep, -hep	[DIR] ‘downward’
SV	Sye	yep	‘go down’
NCal	Xârâcùù	βē	[DEM] ‘coming down’
Mic	Kosraean	-yɛ	[DIR] ‘downward’
Mic	Mokilese	-ti	[DIR] ‘downward’
Mic	Puluwatese	-tiw	[DIR] ‘downward, west’
Fij	Wayan	ðivo	[DIR] ‘downward’
Fij	Bauan	ðivo-ðivo	‘wind sweeping down from hills’
Fij	Yasawa	ðivo	[DIR] ‘downward’
Pn	Tongan	hifo	[DIR] ‘downward’
Pn	Samoan	ifo	[DIR] ‘downward’
Pn	Pileni	ifo	[DIR] ‘downward’
Pn	Rennellese	iho	[V, DIR] ‘downward; seaward; northward, westward’

PEMP		*sobu	‘go downward’ (Blust 1978a)
POc		*sobu	‘go downward, dive down’
PT	Gumawana	-sou	‘move down’
PT	Tawala	-hopu	‘go down’
PT	Saliba	dobi	‘go down’
SES	Gela	sovu-sovu	‘splash about in sea’
Fij	Wayan	sovu	‘go down’
Fij	Bauan	sobu	‘go down’; [DIR] ‘downward’
Fij	Boumā	sobu	‘go down’; [DIR] ‘downward’
Fij	Rotuman	jopu	‘dive, swim under water’
Pn	Rarotongan	ʔopu	‘(boat or stone) sink, (sun) set, fade away’

PMP		*surup	‘enter, penetrate’ (ACD)
POc		*surup	‘enter, penetrate; go down (?)’
MM	Barok	su	‘downwards’
MM	Konomala	sup	‘(sun) set’
SES	Bugotu	horu	‘go down’ (-o- for expected *-u-)
SES	Gela	horu	‘go down’ (-o- for expected *-u-)
Fij	Bauan	ðuru	‘enter’
Fij	Rotuman	suru	‘enter’
Pn	Tongan	hū	‘enter’
Pn	Samoan	ulu	‘enter’
Pn	Marquesan	uʔu	‘enter’

The forms listed below also seem to constitute a cognate set, but, as the questions implicit in the reconstruction *[s,j]u[(a,u)] indicate, their history is not fully understood. The New Caledonian reflexes suggest a contrast between a directional adverbial form in *j- and a verb in *s-, but this contrast is not reflected elsewhere in the set.

POc		*[s,j]u[(a,u)]	‘go down vertically, fall’20
NNG	Sio	due	‘downwards’
NNG	Mangap	-su	‘go down’
NNG	Mangap	-su(-la)	‘go down away from speaker’
NNG	Tami	suʔ	‘downwards’
NNG	Lukep	du	‘go down’
NNG	Malasanga	(i)rua	‘(sun) set’
NNG	Roinji	ru	‘(sun) set’
NNG	Mindiri	du(lau)	‘(sun) set’
NNG	Gedaged	-du	‘go down’
NNG	Megiar	-du	‘go down’
NNG	Takia	(i)du(man)	‘downwards’
NNG	Takia	-du	‘go down, fall’
SJ	Sobei	-so	[DIR] ‘downward’
MM	Tigak	(i)sua	‘go down’
MM	Notsi	(bi-)dū	[adverb] ‘from below’21
MM	Notsi	(ta-)dū	[adverb] ‘from the west’
SES	Longgu	su	‘dive, (sun) set’
SES	Lau	sū	‘dive, (sun) set’
SES	Arosi	sū	‘dive, (sun) set’
NCal	Nêlêmwa	du	[DIR] ‘downward’
NCal	Nêlêmwa	tu	‘go downward’
NCal	Nyelâyu	-du	[DIR] ‘downward’
NCal	Nyelâyu	tu	‘go downward’

POc *wau ‘go seawards’ and *bala ‘move downward (?)’ are also tentatively reconstructed, but they are not well supported.

POc		*wau	‘go seawards’
NNG	Bam	wau(la)	[DIR] ‘downward’
NNG	Sissano	eu	[DIR] ‘downward’
PT	Gumawana	-iwo	‘move seaward’
Mic	Puluwatese	-wow	[DIR] ‘seaward’ (contrast -waw ‘towards addressee’)
Mic	Woleaian	waɨ	[DIR] ‘seaward’

POc		*bala	‘move downward (?)’
NNG	Manam	bala	‘move down, move to one’s left when facing sea’
MM	Tolai	ba	[DIR] ‘downward’
PT	Gumawana	-bala	‘move across’
Fij	Nadrogā	bale	[DIR] ‘downward’

3.3.2. Upward movement⇫

The generic verb of upward movement was POc *sake ‘go upward, go southeast’, also used, for example, of boarding a canoe.

POc		*sake	‘go upward’
Adm	Mussau	sae	‘go upward’
Adm	Mussau	sae(-mae)	‘come up (from)’
Adm	Mussau	(la-)sae	‘go up (to)’ (la ‘go away from speaker’)
NNG	Kove	-ðae	‘go upward’
NNG	Gitua	-zage	‘(sun) rise’
NNG	Tami	sai	‘go up to’
NNG	Mangap	-se	‘go upward’
NNG	Mangap	-sa-la	‘go upward away from speaker’
NNG	Yabem	-sa	[DIR clause-final] ‘upward’
NNG	Gedaged	-sa	‘(plant) grow’
NNG	Takia	-s(-da)	‘move up, be high, be full, go up, rise, board (a canoe)’
NNG	Takia	-sa(-la)	‘go inland, move uphill’ (i.e. towards the volcano); ‘depart (by boat)’
NNG	Manam	-raʔe	‘move up, move to one’s right when facing sea’; [DIR] ‘upward’
NNG	Ali	-ha	‘(sun) rise’
NNG	Sissano	ha	[DIR] ‘upward’
SJ	Sobei	-sa[sa]	[DIR] ‘upward’
PT	Tawala	-gae	‘go upward’
PT	Saliba	-sae	‘go upward, eastward’
PT	Sinaugoro	-raɣe	‘upward’ (second element of verbal compounds)
PT	Motu	-dae(roha)	‘(sun) rise’ (roha ‘to come in sight’)
MM	Bali	zaɣe	‘(sun) rise’; [DIR] ‘upward’
MM	Nakanai	sahe	‘climb’
MM	Meramera	saʔe	‘climb’
MM	Tigak	(i)sa	‘go upward’
MM	Halia	sei	‘(sun) rise’
MM	Teop	hae	‘board (canoe)’
MM	Banoni	sai	[DIR] ‘upward’
MM	Mono-Alu	sae	[DIR] ‘upward’
MM	Roviana	saɣe(la)	‘go up’
MM	Hoava	saɣe(la)	‘go up’
MM	Kia	haɣe	‘board (canoe)’
MM	Kokota	hage	‘go up; go landward; go east’
SES	Gela	haɣe	‘enter’
SES	Longgu	taʔe	‘ascend, go up, stand up, get into canoe’
SES	Kwaio	taʔe	‘embark, rise’
SES	Sa’a	taʔe	[DIR] ‘up, inland’
SES	Arosi	taʔe	‘go upward’
NCV	Merei	sa	‘move upward/landward’
NCV	Araki	sa[ha]	‘go up, go inland, go eastward’
NCV	Tamambo	sahe	‘go upward’
SV	Lenakel	(a)hak	‘(sun) already risen’
SV	Southwest Tanna	-hak(ta)	[DIR] ‘upward’
SV	Sye	saɣ	‘go up, go upstream, (tide) rise’
Mic	Kosraean	-ɛk	[DIR] ‘upward’
Mic	Marshallese	tak	‘(sun) rise’; [DIR] ‘upward’
Mic	Mokilese	-ta	[DIR] ‘upward’
Mic	Puluwatese	-tæ	[DIR] ‘upward’
Mic	Woleaian	tax	[DIR] ‘upward, eastward’
Fij	Bauan	ðake	[DIR] ‘upward, eastward’
Fij	Wayan	ðake	‘climb up, mount’
Pn	Tongan	hake	‘go upward, esp. from the sea to the land’; [DIR] ‘upward’
Pn	Samoan	aʔe	[DIR] ‘upward’
Pn	Pileni	-ake	[DIR] ‘upward’
Pn	Marquesan	aʔe	‘upwards, distant in time’

Some or all of the forms below probably also reflect POc *sake ‘go upward’, but they all reflect unexplained anomalies. The New Caledonian forms reflect the same contrast between an adverb in *j- and a verb in *s- as was noted above with regard to POc *[s,j]u[(a,u)] ‘go down vertically, fall’.

NNG	Takia	(-s)da	‘move up, be high, be full, go up, rise, board (a canoe)’
NCal	Nêlêmwa	da	[DIR] ‘upward’
NCal	Nêlêmwa	(o-)da	‘go upward’ (o ‘go’)
NCal	Nyelâyu	-da	[DIR] ‘upward’
NCal	Nyelâyu	ta	‘go upward’
NCal	Nemi	-da	[DIR] ‘upward’
NCal	Nemi	ta	‘go upward’
NCal	Cèmuhî	da	[DIR] ‘upward’
NCal	Tîrî	ɖa(-jɯ)	‘go up’
Fij	Wayan	ða(va)	‘ascend, go up a slope’
Fij	Wayan	ða(dra)	[DIR] ‘upward’

3.3.3. *sipo and *sake as directions at sea⇫

Above I noted François’ (2004a) reconstruction of POc speakers’ use of terms for ‘go downward’ and ‘go upward’ for ‘go northwest’ and ‘go southeast’, i.e. directions corresponding with those of the major winds. François assumes that the relevant POc terms were *sipo and *sake, reconstructed in §§3.3.1–3.3.2. As these were the generic terms for ‘go downward’ and ‘go upward’, his assumption is probably correct, but it is not particularly well supported by the available data, as the sets below show. In fact, all supporting data for both terms in these meanings come from Eastern Oceanic languages. The Yabem and Motu reflexes of *sipo have ‘wrong’ directions in their glosses. This may mean that this use of *sake and *sipo was an Eastern Oceanic innovation, but it may also mean that insufficient Western Oceanic (and Admiralties) systems have been carefully recorded. In a number of Western Oceanic languages, ‘east’ and ‘west’ are translated as ‘place where the sun rises’ and ‘place where the sun sets’, but these phrasal expressions may be modern renderings of English ‘east’ and ‘west’.

POc		*sipo	‘go downward; go northwest’
NNG	Yabem	-sep	‘go down, go east’
PT	Motu	diho	‘south, south wind; down; go down, descend’
NCV	Ambae	hivo	‘move downward/seaward/northwestward’
NCV	Araki	si[vo]	‘go down, go seaward, go westward’
SV	Anejom̃	-se[h]	[DIR] ‘down, north, west’
Mic	Woleaian	tiw	[DIR] ‘downward, westward’
Pn	Rennellese	iho	[V, DIR] ‘downward; seaward; northward, westward’

POc		*sake	‘go upward, go southeast’ (Dempwolff 1938)
NCV	Mwotlap	hag	[DIR] ‘(on land) eastward, (at sea) south-eastward’
NCV	Ambae	hage	‘move upward/landward/southeastward’
NCV	Araki	sa[ha]	‘go up, go inland, go eastward’
SV	Anejom̃	-tʃai	[DIR] ‘upward, southward, eastward’
Mic	Kiribati	rake	‘(sun) rise’; [DIR] ‘upward, eastward’
Mic	Woleaian	tax	[DIR] ‘upward, eastward’
Fij	Bauan	ðake	[DIR] ‘upward, eastward’
Pn	Rennellese	ake	[V, DIR] ‘upward; inland; southward, eastward’

François also alludes to members of the Polynesian sets below. However, these are clearly locative nouns, and may just as well reflect an orientation to sunset and sunrise, as Biggs (1994:25) implies.

PPn		*si-sifo	‘(N) west’ (Biggs and Clark 1993)
Pn	Tongan	hihifo	‘west’
Pn	Niuean	hifo	‘go west’
Pn	Samoan	sisifo	‘west’
Pn	Tokelauan	sisifo	‘west’

PPn		*sa-sake	‘(N) east’ (Biggs and Clark 1993)
Pn	Tongan	ha-hake	‘east’
Pn	Samoan	sa-saʔe	[ADV] ‘in the east’
Pn	Samoan	(ŋā-ŋ)aʔe	[ADV] ‘eastward’

The important point about François’ reconstruction is that the equation of ‘go downward’ and ‘go upward’ with ‘go northwest’ and ‘go southeast’ occurs widely, suggesting that the equation itself should be reconstructed for POc, even if the forms themselves are hard to reconstruct. Except for Wayan, the terms listed below are drawn from François (2004a).

		‘downward, northwest’	‘upward, southeast’
PT:	Saliba	sae	dobi
SES:	Longgu	alaʔa	toli
NCV:	Mwotlap	hag	hōw
NCV:	NE Ambae	hage	hivo
SV:	Anejom	-jai	-se(h)
NCal:	Nemi	-da	-dic
Mic:	Woleaian	-tiw	-tax
Fij:	Wayan	vua i rā	vua i ata (vua ‘direction’)
Fij:	Bauan	sobu ‘go down, west’	ðake ‘go up, east’

3.3.4. Geographic direction adverbs derived from verbs⇫

In a scattering of Oceanic languages, the verbs POc *sipo ‘go downward’ (p.271) and POc *sake ‘go upward’ (p.273) are also reflected as (or as the roots of) locative and/or allative adverbs. These are distinct from directional adverbials in that they often form part of the locative demonstrative paradigm, with meanings like ‘down below’ and ‘up here’. These reflexes are sufficiently well distributed to arouse the suspicion, at least, that this was also one of their POc functions.

The meanings of the items listed below overlap substantially with reflexes of the POc local nouns *tanoq ‘down below’ (p.241) and POc *atas ‘top; space above’ (p.243). There are also a few reflexes below of *sipo and *sake which function as nouns, but it seems certain that these are the results of locally restricted developments.

POc		*sipo	‘go downward’; [ADV] ‘downwards, down below’
NNG	Kove	sio	[ADV] ‘down below’
NNG	Lusi	sio	[ADV] ‘below, down there’
NNG	Bariai	(ga)dio	[ADV] ‘downward’
SES	Kwaio	(ʔai)sifo	[ADV] ‘downwards, northwesterly’
NCV	Merei	(ai)sio	[ADV] ‘down here’
NCV	Araki	sivo(su)	[ADV] ‘down there’ (-su DEM)
SV	Sye	(ye)hep	[ADV] ‘down here’
Pn	Samoan	si-sifo	[ADV] ‘in the west’
Pn	Samoan	(ŋā-ŋa)ifo	[ADV] ‘westward’

POc		*sake	‘go upward’; [ADV] ‘upwards, up top’
Adm	Mussau	sae-sae(na)	‘upwards’
NNG	Lusi	sai	[ADV] ‘on top, above’
NNG	Bariai	(ga)dae	[ADV] ‘above’
NNG	Tuam	(i)za	[ADV] ‘upwards’
NNG	Gitua	sage	[ADV] ‘up above’
PT	Motu	dae-	[N] ‘above’
MM	Siar	sai(gali)	[ADV] ‘up there, over there’
MM	Label	sa	[ADV] ‘up there’
MM	Label	(u)sa	[ADV] ‘upwards’
MM	Minigir	(ke-na)sa	[ADV] ‘up there’
NCV	Merei	(ai)sa	[ADV] ‘up here’
NCV	Araki	saha(su)	[ADV] ‘up there’
NCV	Paamese	(ne)sa	[N LOC] ‘up, above, on top’
SV	Sye	(ya)haɣ	[ADV] ‘up here’
NCal	Iaai	(e)ðə̄	[ADV] ‘upward, inland’
Mic	Kiribati	rake	‘up, above’
Fij	Wayan	ðake	[DIR] ‘upwards’
Fij	Bauan	ðake	[N LOC] ‘up, above’
Pn	Tongan	ha-hake	[N] ‘east’; [ADJ] ‘eastern’
Pn	Samoan	sa-saʔe	[N LOC] ‘east’
Pn	Samoan	(ŋā-ŋ)aʔe	[ADV] ‘eastward’

3.4. Deictic directional verbs and enclitics⇫

3.4.1. A note on deixis in Oceanic⇫

Proto Oceanic deixis was person-oriented. That is, there were forms with the meanings ‘near speaker’, ‘near addressee’ and ‘distant from both speaker and addressee’. This pattern is well represented in Oceanic demonstrative forms right across the Pacific. Individual languages may lose a member of the paradigm, finishing up with a proximal/distal system. Some languages have reinterpreted the three-way system in terms of orientation to the speaker alone (‘near speaker’ vs ‘an intermediate distance from speaker’ vs ‘far from speaker’), but such languages are by no means as widespread as the person-oriented system. A good many languages have added members to the system, distinguishing, for example, between referents that can and cannot be seen by the speech act participants, or adding a member for referents the speaker points at, but none of these additions can be reconstructed as a POc category.

A reconstruction of POc demonstrative forms is beyond the scope of this chapter, but a sample of person-oriented systems is given below. The forms given are those used adnominally, except where shown (they may also have other language-specific uses).

		near speaker	near addressee	distal
Adm:	Mussau	toko	o[ia]	teke
NNG:	Lukep (Pono)	i	in	ni
NNG:	Yabem	tɔnɛʔ	tɔnaŋ	tone
NNG:	Kairiru	an	at	nai
PT:	Gapapaiwa	we-ni	na-ni	noko-ni
PT:	Gumawana	ame	moe	amo
PT:	Sinaugoro	mai	mani	mo[a]
MM:	Bali	-ani	-ina	-ini
MM:	Tigak	gura	gara	tara
MM:	Roviana SG	hie	isa	hoi
SES:	Bugotu	ani, eni	ŋeni, ari	ŋgeri
NCV:	Araki	ne, kesi	ho-ni, v̈aha-ni	v̈a[ha]-su
NCV:	Lewo	nini (also nene ‘near speaker and hearer’)	namʷā	nena
SV:	Anejom PRO SG	niñ[ki], nī	nā[nai]	nai[kou]
NCal:	Tinrin	=ha	=mʷã	=ṛa
NCal:	Iaai	āŋ	ē	e-lē
Mic:	Kosraean	ʌ	an	o
Mic:	Ulithian	-ē	-lā	-lāy
Fij:	Boumaa	yai, ī	yā	mayā
Fij:	Wailevu	ɣā	ɣāri	ɣaðei
Pn:	Tongan	ni	na	ia (also ē ‘pointing’)
Pn:	Pileni	ne[i]	na	la
Pn:	Marquesan	nei	nā	ʔā, aʔa

The person-oriented system of deixis was manifested not only in demonstratives, but also in the system of deictic directional verbs. POc had a system with at least three members: ‘come to[wards] speaker’, ‘go/come to[wards] addressee’ and ‘go to a point away from both speaker and addressee’, glossed here as ‘come’, ‘go:2’ and ‘go:3’ respectively. There were possibly two ‘go:3’ verbs, however. One licensed a location expression, i.e. its basic meaning was ‘go to’. The other simply meant ‘go away (from speaker)’ and did not license a location expression. Thus in Kele (Adm) there is a contrast between la ‘go to’ and aw ‘go away’ (Ross 2002f). This suggests that POc had a four-member system, and I will assume this here. Against this is the fact that I have found no language in which a four-member system is preserved.

Three-member deictic directional verb systems are found right across Oceania,²² and a sample is given below. Some are reflected as directional enclitics, rather than as verbs. It is sometimes difficult to tell from a source whether the third member should be assigned to ‘go:3’ or ‘go away’. However, in languages where deictic directional verbs have become directional enclitics, the ‘go:3’ form has lost its capacity to license a location expression and inevitably means ‘go away’. Note that the two systems from the Admiralties make the contrast between ‘go:3’ and ‘go away’ which supports the reconstruction of a four-member system.

		come	go:2	go:3	go away
Adm:	Loniu	mɛ	—	la	yaw
Adm:	Kele	sa	—	la	aw (also doh ‘come from’)
NNG:	Yabem	-mèŋ	-wàʔ	-yà	—
PT:	Gumawana	-ma	-wo	-na	—
MM:	Hoava	-mae	-atu	-la	—
NCV:	NE Ambae	vanai	vanatu	—	vano
Mic:	Kosraean DIR	-ma	-ɔt	—	-lᴂ
Mic:	Mokilese DIR	-to, -tɔ	-wɛ	—	-la
Fij:	Bauan DIR	mai	yani	—	tani
Fij:	Wayan DIR	mai	ati	—	dei
Pn:	Tongan DIR	mai	atu	—	aŋe
Pn:	Samoan DIR	mai	atu	—	ʔese
Pn:	Pileni DIR	mai	atu	—	aŋe

3.4.2. Reconstructing Proto Oceanic deictic directional forms⇫

The main deictic directional forms of POc, which account for the majority of the forms listed above, were as follows:

		verb	directional adverb
come	*mai, *ma	‘come’	‘towards speaker’ (p.281)
go:2	*ua	‘go towards addressee’	‘towards addressee’ (p.283)
	*watu	‘go towards addressee’	‘towards addressee’ (p.286)
go:3	*lako, *la	‘go (to)’	‘away from speaker’ (p.287)
go away	*pano, *pa	‘go away’	‘away from speaker’ (p.290)

Note that I reconstruct these POc forms as both verbs and directional adverbs. In §3.2 I briefly discussed the grammaticisation of deictic directional verbs as directional adverbs. This process raises the question, Were there already directional adverbs in POc, or are modern Oceanic directional adverbs the outcomes of independent parallel developments? The answer appears to be: both. Directional adverbs are so widespread in Oceanic languages that one may infer that they were already present in POc, otherwise we would not find such a plethora of reflexes of verbs as enclitics in modern Oceanic languages. On the other hand, there are cases where it is clear that the development of the directional adverb is more recent because it reflects not just the verb root but also accretions to it that have occurred in the history of the particular language.²³ The Sobei (SJ) directional enclitics -ema ‘towards speaker’ and -ewo ‘away from speaker’ transparently reflect the verbs -ma ‘come’ and -wo ‘go’ with a fossilised third person singular subject marker e-. The Sye (SV) directional adverb mpelom ‘towards speaker’ is transparently derived from the verb velom ‘come’. This must be a late development, since ve-lo-m seems itself to reflect the concatenation of three roots, *pano/*pa ‘go away’, *lako/*la ‘go’, and *ma ‘come; towards speaker’.

Three of the verbs listed above, *mai/*ma, *lako/*la and *pano/*pa, have long and short forms. It is difficult to reconstruct the distribution of the long and short forms with any certainty, as they do not correspond with the division between verb and enclitic. However, there seem to be two contexts in which the short forms may have emerged, both of them in serial verb constructions. One was as the final verb of a deictic directional serial verb construction. The other was as the first verb of a sequential serial verb construction (the gloss V AUX is used to label these below). POc *watu and *ua look superficially like a long/short pair, but they are regionally distributed, unlike the other three pairs (p.286).

Two other verbs seem to have been used as deictic directional verbs, but probably not as directional enclitics, in POc. They are less well attested than the verbs reconstructed above.

come	*pʷati	‘come’ (p.283)
go:3/go away	*[y]aku	‘go (to)’ or ‘go away’ (p.293)

It is unclear whether *[y]aku meant ‘go (to)’ or ‘go away’.

I also reconstruct the prepositional verb *tani ‘(go) away from’ below (p.293). It was not deictic, but it overlaps semantically with the verbs reconstructed in this section.

3.4.3. ‘Come towards speaker’⇫

Blust (ACD) reconstructs PAn *aRi, which in its root form was used imperatively as ‘come on’, ‘let’s go’ (Paiwan [Formosan] ari ‘let’s go!’) and apparently survives in the Takia interjection aria with the same meanings.²⁴ The indicative form was PAn *maRi ‘come’ (from *um-aRi, where *um is the actor focus infix; vol.1,29), and this form is well reflected in Formosan and WMP languages. However, it seems likely that the variant *mai existed from the earliest times, as Blust notes Favorlang (Formosan) mai and Yami (WMP) mai, and that *mai ousted reflexes of *maRi throughout the Central/Eastern Malayo-Polynesian grouping (e.g. Manggarai, Sikka, Rotinese mai) to which POc belonged.

There is just one Oceanic reflex, Mangap (NNG) -mar ‘come towards speaker’, which appears to reflect *maRi rather than *mai. However, the fact that Mangap has a distinction between -mar ‘come towards speaker’ and -ma ‘come towards addressee’ suggests that a pre-Mangap reflex of *ma expanded its use from ‘towards speaker’ to ‘towards us, speaker and addressee’, and that a further morpheme, perhaps a demonstrative, was then suffixed to it to disambiguate ‘towards speaker’, giving -ma-r.

Also reconstructable is the POc form *ma, reflected in WOc and CEOc languages. There is no reconstructable functional distinction between *-mai and *-ma, as reflexes of both occur as verbs and as directional enclitics, and both should probably be glossed in POc as (V) ‘come’ and (DIR) ‘towards speaker’. In many languages, a reflex of *mai or *ma occurs as the second element of one or more compounds, and these are taken to be reflexes of an earlier final verb of a serial construction or reflexes of a directional enclitic (it is impossible to tell which). A number of these are listed below.

PAn		*maRi, *mai	‘come’ (ACD)
PCEMP		*mai	‘come’
POc		*mai, *ma	‘come’; [DIR] ‘towards speaker’
Adm	Mussau	mae	‘come’
Adm	Mussau	(sio)mae	‘come down (from)’
Adm	Loniu	-mɛ	‘come’; [DIR] ‘towards speaker’
Adm	Aua	-mai	[DIR] ‘towards speaker’
NNG	Mangap	-ma	‘come towards addressee’
NNG	Mangap	-ma(r)	‘come towards speaker’
NNG	Mangap	(-le)-ma	‘come inside’ (-le ‘enter’)
NNG	Yabem	-mè(ŋ)	‘come’ (-ŋ is a suffix of unknown origin)
NNG	Sio	mɔ	‘come’
NNG	Tuam	(ka)miai	‘come’
NNG	Gitua	(la)m	‘come’
NNG	Manam	mai	‘move towards speaker from neither up nor down’; [DIR] ‘towards speaker’
NNG	Kairiru	-myai	‘come’
SJ	Sobei	-(e)ma	[DIR] ‘towards speaker’
SJ	Sobei	-ma	‘come’
PT	Tawala	mai	[DIR] ‘towards speaker’
PT	Gumawana	-ma	‘come’
PT	Saliba	ma	[DIR] ‘hither’
PT	Sudest	ma	[DIR] ‘towards speaker’
PT	Sinaugoro	(iaɣo)ma	‘come’ (iaɣo ‘go’ < POc *lako)
PT	Sinaugoro	-ma	[preverbal clitic] ‘near speaker’25
PT	Sinaugoro	-ma(riɣo)	‘come down’ (-riɣo ‘downward’ < POc *sipo occurs in verbal compounds)
PT	Motu	-mai	‘come’
MM	Vitu	mai	‘come’
MM	Bali	mai	‘come’
MM	Bulu	mai	[DIR] ‘towards speaker’
MM	Harua	mai	‘come from’
MM	Tigak	(i)ma	‘come’
MM	Notsi	(kala)me	‘come’
MM	Selau	(la)ma	‘come’
MM	Taiof	(o)m	‘come’
MM	Taiof	-ma	‘first/second person object enclitic’26
MM	Banoni	ma	[DIR] ‘towards speaker’
MM	Banoni	(tai)ma	‘come’ (tai ‘go’)
MM	Mono	(lao)ma	‘come’
MM	Babatana	me	[DIR] ‘towards speaker’
MM	Hoava	mae	‘come’
MM	Kokota	mai	‘come’
SES	Bugotu	mai	‘come’; [DIR] ‘towards speaker’
SES	Gela	mai	‘come’; [DIR] ‘towards speaker’
SES	Longgu	mai	[DIR] ‘towards speaker’
SES	Kwaio	mai	[DIR] ‘towards speaker’
SES	Kwaio	(leka)mai	‘come’ (leka ‘go’)
NCV	Mwotlap	mɛ	[DIR] ‘towards speaker’
NCV	Ambae	-mai, -mei, -ai	[DIR] ‘towards speaker’27
NCV	Merei	ma	[DIR] ‘towards addressee’
NCV	Araki	ma	[DIR] ‘towards speaker’
NCV	Tamambo	mai	‘come’
NCV	Sakao	(la)m	‘come’
SV	Sye	(ve-lo-)m	‘come’ (ve ‘go’)
SV	Sye	(-mpe-lo-)m	[DIR] ‘towards speaker’
SV	Anejom̃	-(pa)m	[DIR] ‘towards speaker’ (pan ‘away from speaker’)
SV	Anejom̃	(ha)m, (apa)m	‘come’ (han, apan ‘go’)
NCal	Nêlêmwa	me	[DIR] ‘towards speaker’
NCal	Nêlêmwa	ō-me	‘come’ (o ‘go’)
NCal	Nyelâyu	-me	[DIR] ‘towards speaker’
NCal	Xârâcùù	mɛ̃	[DEM] ‘near speaker’
NCal	Iaai	(jē)m	[DIR] ‘towards speaker’
Mic	Kosraean	-ma	[DIR] ‘towards speaker’
Mic	Kiribati	mai	‘come’
Fij	Nadrogā	mā, mei	[DIR] ‘towards speaker’
Fij	Wayan	mai	[DIR] ‘towards speaker’
Fij	Bauan	mai	[DIR] ‘towards speaker’
Fij	Boumā	mai	[DIR] ‘towards speaker’
Pn	Tongan	mai	‘come’; [DIR] ‘towards speaker’
Pn	Samoan	mai	[DIR] ‘towards speaker’
Pn	Marquesan	mai	[DIR] ‘towards speaker’

In a number of languages POc *mai/*ma is reflected as a preposition or a relator. However, the reflexes listed below are probably the result of parallel innovations in different groups of languages, and it is unnecessary to reconstruct a preposition or relator usage for POc *mai/*ma (cf. §3.2).

POc		*mai, *ma	‘come’; [DIR] ‘towards speaker’
PT	Tawala	mei	[PREP] ‘like, resembling’
MM	Meramera	maʔ-	[R-] (ablative)
MM	Lamasong	ma-	[R-] (ablative)
MM	Madak	me-	[R-] (ablative)
MM	Barok	mu-	[R-] (ablative)
MM	Konomala	mə	[R] (ablative)
MM	Patpatar	ma-	[R-] (ablative)
MM	Patpatar	ma	(PREP w PLC; ablative)
MM	Label	mi-	[R-] (ablative)
MM	Tolai	ma-, ma-ma-	[R-] (ablative)
SES	Longgu	mi	(R with placename or local noun; ablative)
Fij	Bauan	mai	[PREP] (ablative)

One other POc morpheme with the same meanings as *mai may be reconstructable. This is *pʷati. Note, however, that most reflexes are in the South New Ireland grouping of MM, and that the POc status of this reconstruction is dependent on the Arosi reflex alone.

POc		*pʷati	‘come’; [DIR] ‘towards speaker’
NNG	Wogeo	(e)wot	‘come’
MM	Minigir	(vana)uti	‘come’
MM	Tolai	pot	‘come’
MM	Tolai (Nodup)	(le)poti	‘come’
MM	Label	(la-m)ut	‘come’
MM	Label	hot	‘towards speaker’
MM	Bilur	vot	‘come’
MM	Kandas	(uan)pat	‘come back’
SES	Arosi	boi	‘come’

3.4.4. ‘Go towards addressee’⇫

Two alternant forms, POc *ua and *watu, mean ‘go towards addressee’, (DIR) ‘towards addressee’.

It is just possible that POc *ua reflects PAn *kuSa ‘go’. However, no reflexes of PAn *kuSa have been found in non-Oceanic languages outside Taiwan,²⁸ and it is more likely that the Taiwan and Oceanic sets reflect different etyma, the more so as no member of the Oceanic set reflects PAn *k-.

Some of the reflexes of *ua can be confused with those of POc *pa and *ba. That there is a contrast between *pa and *ua is attested by the pairs Adzera fa ‘go’ (p.291) and waʔ- ‘go out’ (p.286) and Kiriwina va (PREP, p.292) and wa (VF, below).

POc		*ua	‘go towards addressee’; [DIR] ‘towards addressee’
Adm	Seimat	-wa	[DIR] ‘away from speaker’
SJ	Sobei	-(e)wo	[DIR] ‘away from speaker’
SJ	Sobei	-wo	‘go’
PT	Gumawana	-wo	[DIR] ‘towards addressee’
PT	Saliba	-wa	[DIR] ‘thither’
PT	Kilivila	-wa	‘go (to addressee)’
PT	Sudest	-wo	[DIR] ‘away from speaker’
PT	Sudest	wa	‘go’
PT	Lala	-ovo	[DIR] ‘away’
MM	Bali	ua	‘go’
MM	Tolai	vue	[DIR] ‘away’
MM	Halia	wa	[DIR] ‘towards (a specified destination)’ (Ross 1982: 44–45)
Fij	Nadrogā	wā	[DIR] ‘thither’
Fij	Wayan	ā	[DIR] ‘thither’

In a number of languages POc *ua is reflected as a preposition or a relator (cf. §3.2).

POc		*ua	‘go towards addressee’; [DIR] ‘towards addressee’
Yap	Yapese	u	[PREP] (locative, ablative)
PT	Tawala	u	(PREP w N LOC; locative, allative)
PT	Kilivila	o	(PREP w N LOC; locative ‘in, into’)
PT	Muyuw	u, wa	[PREP] (locative, allative)
MM	Bulu	o	(PREP w PLC; locative)
MM	Nakanai	o-	(VF formative; locative)
MM	Meramera	u-	[R-] (allative)
MM	Lamasong	u-	[R-] (allative)
MM	Madak	u-	[R-] (allative)
MM	Barok	u-	[R-] (allative)
MM	Sursurunga	u(r)	[R] (allative)
MM	Tangga	ua, u	[R] (allative; ua w DEM, u elsewhere)
MM	Konomala	uə	[R] (allative)
MM	Patpatar	u-	[R-] (allative)
MM	Patpatar	u	(PREP w PLC; allative)
MM	Label	u-	[R-] (allative)
MM	Kandas	u-	[R-] (allative)
MM	Kandas	u	(PREP w N LOC; locative, allative)
MM	Ramoaaina	u-	[R-] (allative)
MM	Ramoaaina	u	[PREP] (locative, allative)
MM	Minigir	u-	[R-] (allative)
MM	Tolai	u-	[R-] (allative)
SES	Longgu	vu	[R] (allative, towards)
Fij	Bauan	vuā, vei	[PREP] (locative, allative, dative, cause)

Proto South Vanuatu *un-, which forms locative nouns from common nouns (Lynch 2001c:132), may also belong here.

Longgu vu appears to be cognate with forms in the Malaita/Makira subgroup (SES) that reflect Proto Malaita/Makira *vua or *vuni. Lichtenberk (1985b) attributes all these forms to POc *pani (V) ‘give’, (PREPV) ‘beneficiary case-marker’. The Malaita/Makira forms certainly have benefactive meanings, but Longgu vu is clearly allative, and the best account of both its form and meaning is given by attributing it to *ua — although this means assuming that initial *v- has arisen by epenthesis. But what are we to do with the Malaita/Makira forms? The best explanation seems to be that there was a conflation of pre-Proto Malaita/Makira *vua ‘allative relator’ (cognate with Longgu vu and reflecting POc *ua) and *vani ‘beneficiary prepositional verb’ (reflecting POc *pani), resulting in *vua with a benefactive function and *vuni with a form cobbled together from both items but in benefactive function.

This interpretation is supported by the Longgu dative verbal preposition wini-, which takes an object pronoun suffix when it governs a first or second person referent, e.g. wini-o DATIVE-O:2SG ‘to you’, but assumes the allomorph wa- with a possessor pronoun suffix when it governs a third person, e.g. wa-na DATIVE-P:3SG ‘to it/him/her’.²⁹ As Hill (1992:245) points out, wa- overlaps semantically with vu (< POc *ua). It appears that wini reflects POc *pani and wa- POc *ua, but the two form a single paradigm in Longgu. This inference would also explain the non-etymological initial *v- of vu: it is inherited from pre-Proto Malaita/Makira *vua, where it resulted from ‘infection’ by *vani.

The Bauan Fijian forms are tentatively included in the set above. The form vuā again has epenthetic v-, but its meaning and the parallel with the prepositional use of mai suggest that it reflects *ua. More specifically, vuā apparently reflects pre-Fijian *vua-i-a (go.towards-TR-O:3SG) and means ‘to/for/from/with him/her’. The form vei is more problematic. It may reflect either *vua-i (go.towards-ART) or *pa-i (go.away-ART). Either way, *i is the personal article. Semantically, it is also possibe that *pani ‘give’ has played a role in the history of these forms, as they have a dative function.

Other reflexes of *ua apparently occur as demonstratives in a number of languages.

POc		*ua	‘go towards addressee’; [DIR] ‘towards addressee’; [DEM] ‘away from speaker’ (anaphoric)
NNG	Lusi	(ɣe-ra)wa	[LOC] ‘away from speaker’
NNG	Lusi	(ne-dua)wa	[PRO] ‘that one remote from speaker’ (cf. ne-dua (PRO) ‘away from speaker’)
NNG	Bariai	oa	(ADN; away from speaker, anaphoric)
NNG	Manam	(ena)wa	(ADN; 3)
PT	Saliba	-wa	(ADN ENCL; anaphoric)
PT	Kilivila	-we	[ADN AFFIX] ‘away from speaker’
PT	Lala	uʔa	(3) [PRO]
Mic	Kosraean	æ	(ADN, POST; anaphoric)
Mic	Mokilese	-wa	(ADN ENCL; anaphoric SG)
Mic	Puluwatese	(ye)we	(ADN, POST; anaphoric)
Mic	Woleaian	we	(ADN, POST; anaphoric SG)
Mic	Ulithian	-wē	(ADN ENCL; anaphoric SG)

Forms reflecting POc *watu are listed below. It is tempting to reconstruct this as *uatu in view of its probable relationship to *ua, discussed below. However, the reflexes listed below point to POc *watu, even if this perhaps reflects pre-POc *uatu.

Irregular loss of initial *w- is reflected in New Caledonian and Central Pacific (Fij and Pn) languages, which reflect *atu for expected **watu. POc *w is lost regularly in all positions in Hoava and Roviana, word-initially in Gela and Bugotu, and sporadically in Mota and NE Ambae and other NCV languages.

POc		*watu	‘go towards addressee’; [DIR] ‘towards addressee’
Adm	Aua	-wau	[DIR] ‘away from speaker’
NNG	Yabem	-wàʔ	‘go (to addressee)’
NNG	Adzera	waʔ-	‘go out’
PT	Motu	vasi	‘go:2/3’
MM	Hoava	atu	‘go (to addressee)’
MM	Roviana	atu-atu	[INTERJECTION] (implying movement away, of speaker or addressee)
SES	Gela	(g)atu	[DIR] ‘away from speaker’
SES	Bugotu	atu	[DIR] ‘away from speaker’
SES	Kwaio	ka[ʔ]u	[DIR] ‘thither’
SES	Lau	kou	[DIR] ‘away from speaker’
SES	Longgu	hou	[DIR] ‘thither’
SES	Arosi	wou	[DIR] ‘thither’
SES	Sa’a	wau	‘there’
NCV	Mota	at	‘outwards or away from speaker’s point of reference’
NCV	Ambae	-atu	‘towards addressee, towards past/future deictic centre’ (lexicalised in some compounds)
NCal	Nemi	-ec	[DIR] ‘away from speaker’
Mic	Kiribati	wati	[DIR] ‘away, hence’
Mic	Kosraean	-ɔt	[DIR] ‘towards addressee’
Mic	Marshallese	wac	[DIR] ‘towards addressee’
Mic	Mokilese	-wɛ	[DIR] ‘towards addressee’
Mic	Puluwatese	-waw	[DIR] ‘towards addressee’
Fij	Wayan	ati	[DIR] ‘away from speaker towards addressee or elsewhere’
Fij	Kadavu	atu	[DIR] ‘outwards or away from speaker’s point of reference’
Pn	Tongan	atu	[DIR] ‘away from speaker towards addressee or elsewhere; onward in time’
Pn	Niuean	atu	[DIR] ‘away from speaker towards addressee’
Pn	Samoan	atu	[DIR] ‘away from speaker towards addressee’

Reflexes of *ua and *watu have a distribution which roughly matches major subgroups: *ua is found throughout Western Oceanic, *watu elsewhere. But the distribution is imperfect. Possible reflexes of *ua occur in Southeast Solomonic and Fijian (and apparently as demonstratives in Micronesian). Reflexes of *watu occur in a few Western Oceanic languages. And reflexes of both forms appear in the Admiralties. Despite these imperfections, however, the distribution of the two forms is quite different from that of the other deictic directional verbs, where geography plays no significant role. It seems legitimate to suggest that both forms occurred in POc, and that as POc broke up and diversified, one form or the other tended to win out on an areal basis.

Why did POc have the two forms *ua and *watu? Evidently, as hinted by Blust (ACD, under the entry for PAn *-Cu ‘near addressee’), *watu represents an innovation whereby the POc demonstrative morpheme *-tu ‘near addressee’ was added to *ua, stretching it to the canonic CVCV shape of POc morphemes. An obvious alternative suggestion is that *ua represents the short form of *watu in the same way as *ma and *la represent the short forms of *mai (p.281) and *lako (see below). This is unlikely, however, as the external evidence indicates that the inherited POc form was *ua, and the fairly neat geographic distribution of *ua and *watu is very different from the scattered, interlaced distributions of *mai and *ma and of *lako and *la.

3.4.5. ‘Go away to’⇫

Just as POc *mai ‘come’ had a short form *ma, so POc *lako ‘go’ had the short form *la. Again, reflexes of both occur as verbs and as directional enclitics, and each probably had both functions in POc, viz. *mai, *ma glossed as (V) ‘come’ and (DIR) ‘towards speaker’ and *lako, *la glossed as (V) ‘go (to)’ and (DIR) ‘away from speaker’. In some languages, a reflex of *lako or *la occurs as the second element of one or more compounds, and these are again taken to be reflexes of an earlier final verb or directional enclitic.

PMP		*lako	‘go’
POc		*lako, *la	[V] ‘go (to)’; [DIR] ‘away from speaker’
Adm	Mussau	la	[DIR] ‘away from speaker’
Adm	Mussau	lao	‘go to’ (la in compounds)
Adm	Loniu	-la	[DIR] ‘away from speaker’
Adm	Loniu	-lɛ	‘go to’
Adm	Kele	la	‘go to’
NNG	Gitua	lago	‘go’
NNG	Mangap	-la	‘go’
NNG	Mangap	-sa-la	[VF] ‘ascend away from speaker’
NNG	Sio	lɔ	‘go’
NNG	Takia	la	‘move away from speaker’; ‘go round the island’
NNG	Manam	laʔo	[DIR] ‘away from speaker’
NNG	Manam	-la[ʔo]	‘move away’
PT	Kilivila	-la	‘go (to some place away from here)’
PT	Tawala	-nae	‘go away’
PT	Gumawana	-na	[DIR] ‘away from speaker and addressee’
PT	Saliba	lao	‘go across’
PT	Sinaugoro	iaɣo	‘go’
PT	Sinaugoro	-a	[preverbal clitic] ‘away from speaker and addressee’30
PT	Motu	la	‘go away’
MM	Tigak	inaŋ	‘go away’
MM	Notsi	la	‘go’
MM	Banoni	nau	[DIR] ‘away from speaker’
MM	Torau	lao	‘go’
MM	Babatana	ka	[DIR] ‘away from speaker’
MM	Hoava	la	‘go away from speaker and addressee’
MM	Kia	lao	‘go’
MM	Kokota	lao	‘go’
SES	Bauro	raɣo	‘go’
NCV	Mota	lago	‘step, stretch the legs’
NCV	Raga	lago	‘walk, travel’
NCV	Sakao	la(m)	‘come’ (from POc *lako + *mai)
SV	Sye	(ve)laɣ	‘go ahead’ (ve ‘go’)
NCal	Cèmuhî	-lɛ	[DIR] ‘away from speaker’
Mic	Kiribati	nako	‘go’
Mic	Kosraean	-læ	[DIR] ‘away from speaker’
Mic	Marshallese	lᵚɒk	[DIR] ‘away from speaker’
Mic	Mokilese	-la	[DIR] ‘away from speaker’
Mic	Puluwatese	-lɔ	[DIR] ‘away, south’
Mic	Woleaian	rax	[DIR] ‘away from speaker’
Mic	Ulithian	loxo	[DIR] ‘away from speaker’
Fij	Bauan	lako	‘go’
Fij	Bauan	la(i)	‘go and …’
Fij	Wayan	la(i)	‘go and …’

It is reasonable to expect that the processes that have created prepositions and relators from the deictic direction verbs *mai/*ma ‘[come] towards speaker’ and *ua ‘[go] towards addressee’ may also have created them from *lako/*la ‘go (to)’, (DIR) ‘away from speaker’. Although there are a good many prepositional reflexes, many of them are problematic because they have more than one possible source. In the set below, it is probable that some items reflect POc *lalo-, *lo-, *la- (N LOC) ‘inside’ rather than POc *lako, *la ‘go (to)’, (DIR) ‘away from speaker’.

1.	POc	*lako, *la	‘go (to)’; (DIR) ‘away from speaker’ (p.287)
2.	POc	*lalo-, *lo-, *la-	(N LOC) ‘inside’ (p.246)
	Adm: Loniu	lo	(PREP) ‘in’
	NNG: Arawe	lu-O:, li-O:	(PREP w N PERS, PRO PERS) locative, allative
	NNG: Mamusi	la	(PREP) general
	PT: Motu	lalo	‘the inside, the mind’
	MM: Tigak	lo	(PREP w N COM, N LOC) locative, temporal
	MM: Tiang	lə	(PREP) locative, temporal
	MM: Kara	la	(PREP) locative
	MM: Nalik	la	(PREP w N LOC) locative, temporal
	MM: Notsi	la(n)	(PREP) locative
	MM: Tangga	lo	(PREP w N PERS, PRO PERS) locative
	MM: Konomala	lə	(PREP w PLC, N LOC) locative, temporal
	MM: Label	la	(PREP w N LOC) locative, temporal
	MM: Bilur	la	(PREP w N LOC) locative
	NCV: Mota	lo	(PREP) ‘in, inside’
	NCV: Tasiko	lo	(PREP) ‘in, inside’
	NCV: Mwotlap	lV-	(PREP prefixed to N LOC) locative, allative
	NCV: NE Ambae	lo	(PREP w N LOC) locative
	NCV: Maewo	le	(PREP) ‘in, inside’
	SV: Lenakel	le	(PREP) locative, allative
	Mic: Marshallese	(i)lɔ	(PREP) locative
	Mic: Puluwat	le-	(PREP) ‘in, because of’

Formally, items reflecting *lo presumably reflect *lo-, one of the short forms of *lalo-. It is tempting to attribute all forms reflecting *la to the short form of *lako, but there is evidence against this. In Lihir (MM), la is a short form of lilie- ‘inside’ (reflecting *lalo-), as we find phrases like the one below where la must be a (relational) noun:

• • Lihir (MM)

    ‘in the house’

    i	la	liom
    PREP	inside	house

Semantically, all the reflexes listed above are locative, which sits better with a derivation from *lalo- ‘inside’ than one from *lako ‘go (to)’, from which one would expect an allative. But it is possible that some reflexes represent a conflation of the two etyma.

3.4.6. ‘Go away’⇫

POc *pano, reconstructed below, perhaps had two uses. Firstly, it was a deictic directional verb meaning ‘go away (from speaker), depart’. Evidence for this meaning is also widespread in non-Oceanic languages (Blust, ACD). Reflexes of both *lako/*la and *pano occur as directional enclitics with the meaning ‘away from speaker’. However, there is evidence that as verbs they had different meanings. Most verbal reflexes of *lako ‘go (to)’ have a valency which implies or requires a destination (expressed, for example, as a prepositional phrase), whilst those of *pano are intransitive.

As noted in §3.3, some reflexes of POc *pano indicate that it was also a geographic directional verb meaning ‘move in a transverse direction’, contrasting with ‘go up, go inland’ and ‘go down, go seawards’. However, it is not entirely clear whether this usage occurred in POc. On one hand, there is a non-Oceanic reflex with this meaning, namely Aralle-Tabulahan (South Sulawesi) pano (DIR) ‘along the level’ (McKenzie 1997). On the other hand, within Oceanic the meaning ‘move in a transverse direction’ is reflected only in North–Central Vanuatu and New Caledonian languages. There are two interpretations of these data: either there were independent parallel innovations in South Sulawesi and Remote Oceanic, or this usage was inherited into POc but happens to have been lost in Western Oceanic and Southeast Solomonic.

PMP		*panaw	‘go away, depart, leave on a journey’ (ACD)
POc		*pano	‘go away’; [DIR] ‘away from speaker’; ‘move in a transverse direction’
MM	Vitu	vano	‘go (away)’
MM	Harua	mano	‘go away’ (see text below)
SES	Bugotu	vano	‘go, come’; [DIR] ‘thither’; ‘beyond, more’ (used in comparisons)
SES	Gela	vano	‘away, further off; to go’
SES	Arosi	hano	‘make a journey, set out; go’
NCV	Mota	van(o)	‘go, come’
NCV	Mwotlap	van	[DIR] ‘thither’
NCV	Ambae	vano	‘move in transverse direction’
NCV	Merei	va, van(a)	‘move in transverse direction’
NCV	Tamambo	vano	‘go away from speaker’
NCV	Lonwolwol	van	‘go, pass (and so also of time); continue (to do s.t.)’; [DIR] ‘away’
NCV	Southeast Ambrym	haen	[N] ‘going, departure’
SV	Lenakel	-pən	[DIR] ‘distant’
SV	Lenakel	vən, (a)vən	‘go, walk’
SV	Anejom̃	-pan	[DIR] ‘away from speaker’
SV	Anejom̃	han, (a)pan	‘go’
NCal	Nêlêmwa	ve	[DIR] ‘in a transverse direction’
NCal	Nêlêmwa	o	‘go’
NCal	Nyelâyu	-van	[DIR] ‘in a transverse direction’
NCal	Nyelâyu	van	‘go’
NCal	Nemi	en	[DIR] ‘in a transverse direction’
NCal	Nemi	hen	‘go’
NCal	Tîrî	(ã)va	‘there, the other side of stream’
NCal	Xârâcùù	fɛ	[DEM] ‘away from speaker’
NCal	Iaai	hããŋ	[DIR] ‘away from speaker crosswise’31
Pn	Niuean	fano	‘go’
Pn	Samoan	fano	‘(of time) be gone, past; perish’
Pn	Nanumea	fano	‘go’
Pn	Rennellese	hano	‘go; depending on, according to; on and on; little by little; one by one’
Pn	Māori	ɸano	‘go, proceed; lead, of a road; verge towards; be on the point of; act, behave’

The Harua form mano appears to reflect the application of the PMP Actor focus morpheme *‹um› to the root *panaw (vol.1,29), suggesting that an alternation between *pano and *mano may have survived in POc.

POc *pano evidently had a short form *pa, giving a pair analogous to *mai/*ma and *lako/*la (pp.281, 287).

POc		*pa	‘go away; move in a transverse direction’; [V AUX] ‘go and …’
NNG	Lukep	pa	‘go’
NNG	Adzera	fa	‘go’
PT	Sinaugoro	va(riɣo)	‘go down’
PT	Motu	ha	[V AUX] ‘go and …’
SES	Gela	va	[V AUX] ‘be going to …’
NCV	Mota	va	‘go, come’; [V AUX] ‘go on …-ing’
NCV	Araki	v̈a	‘go; go in a direction other than north or south’
NCV	Lonwolwol	va	‘go’
NCV	Southeast Ambrym	ha	‘go, leave, depart’
NCV	Paamese	vā	‘go’
NCV	Nguna	vā	‘go’ (short form of vano, Clark 1996)
SV	Sye	-mpe	[DIR] ‘away from speaker’
SV	Sye	-ve	‘go’

The meanings of reflexes of *pa agree with those of *pano, and the function of the Motu, Sinaugoro,³² Gela, and Mota reflexes of *pa (in three different subgroups) as a preverbal auxiliary is similar to that of the Sinaugoro reflexes of *ma and *la above. Clark (1996) notes that Nguna vā ‘go’ is also described as a short form of vano. Paton (1973) describes Lonwolwol va as a short form of van ‘go, pass’, and therefore as a reflex of POc *pano, but Blust (ACD) argues that his inference is unjustified since original medial nasals are otherwise retained in Lonwolwol. If, however, va reflects POc *pa, the objection disappears.

This leaves a loose end. Blust (ACD) takes the cognate set above to reflect PCEMP *ba ‘go, go away, walk’. Clark (1996), on the other hand, infers that the set above and PCEMP *ba are etymologically separate. It is true that the forms attributed to POc *pa above could reflect PCEMP *ba: there is no phonological objection to this. But there are just a few Oceanic forms which reflect a POc locomotion verb *ba ‘go’ (Tolai [MM] ba ‘tread, go’, Talise [SES] ba ‘go’), and it seems likely that this *ba reflects PCEMP *ba, whilst POc *pa is the short form of *pano.

There are a number of apparent prepositional reflexes of *pano/*pa ‘go away’, but most are very problematic, as there are two other possible sources of the items listed.

These are POc *pani ‘give’, (PREPV) ‘benefactive’ and PWOc *pʷa (PREP) ‘instrumental, comitative’.³³ There is also evidence of conflation.

The clearest piece of evidence that *pano/*pa played a role in the history of some of the items in the set below is that Hoava pa behaves as a relator, i.e. it occurs before a preposition.

• • Hoava (MM)

    ‘they who come close to the tree’

    ria	pu	tata	mae	pa	tani	sa	gato
    D:3PL	REL	close	come	R	PREP:3SG	ART:SG	tree

Relators reflect erstwhile deictic directional verbs (Ross 2003), and so *pano/*pa is the most likely candidate for the ancestor of Hoava pa. Other probable straightforward reflexes of *pano/*pa are the Kiriwina, Roviana and Nguna forms, and perhaps the Label locative preposition ha, as it contrasts with instrumental pa (from PWOc *pʷa).

The NNG items below, all from the Vitiaz Strait area, probably reflect conflation of the POc benefactive prepositional verb *pani and a PWOc instrumental preposition *pʷa (Ross 1988:106–108, 112–115).

Bound items below are shown with the suffix paradigm that they take, one of object (O:), disjunctive (D:) or possessor (P:). The gloss of each item is formulated as carefully as the data allow, but should not be treated too seriously, as there are likely to be gaps in the glosses.

1.	POc	*pano, *pa	‘go away; move in a transverse direction’ (p.289)
2.	POc	*pani	‘give’, (PREPV) benefactive (Pawley 1973, Lichtenberk 1985b)
3.	PWOc	*pʷa	(PREP) instrumental, comitative
	NNG: Kove	pa, pa-O:	(PREP) locative, temporal, allative, ablative
	NNG: Bariai	pa-O:	(PREP) locative, allative, ablative, benefactive
	NNG: Malai	pa-D:	(PREP) allative, instrumental
	NNG: Gitua	pa-O:	(PREP) temporal, allative, benefactive, ablative, instrument
	NNG: Malalamai	pa-O:	(PREP) allative, benefactive
	NNG: Lukep	pa-O:	(PREP) allative, benefactive
	NNG: Malasanga	pa-O:	(PREP) benefactive, comitative
	NNG: Roinji	pa-P:/O:	(PREP) allative, benefactive
	NNG: Sio	pa-O:	(PREP) allative, benefactive
	NNG: Tami	pa, pa-D:	(PREP) temporal, benefactive
	NNG: Mangap	pa, pa-O:	(PREP) locative, benefactive, ablative, instrumental
	NNG: Rauto	pa	(PREP w N PERS) locative, allative
		pe	(PREP w N COM, PRO PERS) locative, allative, instrumental
	PT: Kiriwina	va	(PREP) ‘in the direction of’
	MM: Tiang	pa-P:	(PREP) locative, instrumental, comitative
	MM: Nalik	pana	(PREP) locative, instrumental, comitative
	MM: Label	ha	(PREP) locative
	MM: Roviana	pa	(PREP) locative, allative
	MM: Hoava	pa	(R, PREP w PLC, non-human N) locative, allative
	NCV: Nguna	pa(ki)	(PREP w PLC) allative

Functionally and semantically the members of the small cognate set below resemble reflexes of POc *lako, but phonologically they do not reflect it. The medial consonant is reconstructed on the basis of Yabem low tone, which reflects the loss of a Proto Huon Gulf voiced obstruent, probably either *v or *ɣ, lenis reflexes of POc *p or *k. Of these, both are lost intervocalically in Takia, but only *k is lost in the Admiralties languages.

POc		*[y]aku	‘go (to)’; [DIR] ‘away from speaker’
Adm	Loniu	yaw	[DIR] ‘away’
Adm	Kele	aw	‘go away’
Adm	Titan	aw	‘go away, leave’
NNG	Yabem	-yà	‘go (to her/him/them)’
NNG	Takia	-au	‘go (from the speaker)’

3.4.7. ‘Away from a specified point’⇫

POc *tani was a prepositional verb, reconstructed by Pawley (1973). It was not deictic, i.e. not oriented with regard to speaker or hearer, and so strictly does not belong here. It was transitive, and the object of the verb was the point of orientation from which movement takes place. I include it because its reflexes have become deictic directional adverbs in a few languages.

POc		*tani	[PREPV] ‘(go) away from’
PT	Motu	tani	[PREPV] ‘away from’
PT	Mekeo	-ani	‘away from’ (in compounds: fossilised DIR)
NCV	Merlav	dani-	[PREPV] ‘away from’
NCV	Ambae	dene	[PREPV] ‘away from’
NCV	Sesake	deni-	[PREPV] ‘away from’
Fij	Bauan	tani	[DIR] ‘away, elsewhere’
Fij	Boumā	tani	[DIR] ‘away’

A note on sources

In addition to the sources of lexical items listed in Appendix 1, a number of grammars and other grammatical sources were consulted during the research on which this chapter is based. Other than my fieldnotes, these are: NE Ambae (Hyslop 2001), Anejom (Lynch 2000b), Araki (François 2002), Arosi (Lynch & Horoi 2002), Awad Bing (Bennett & Bennett 1998), Bali-Vitu (Ross 2002a), Banoni (Lynch & Ross 2002), Bariai (Gallagher 1998), Bauan Fijian (Churchward 1973, Schütz 1985), Boumaa Fijian (Dixon 1988), Bugotu (Ivens 1933, author’s fieldnotes), Cèmuhî (Lynch 2002a), Drehu (Moyse-Faurie 1993), Erromangan (Sye) (Crowley 1998), Gapapaiwa (McGuckin 2002), Gela (Crowley 2002a), Gumawana (Olson 1992), Halia (Allen 1987), Hoava (Davis 1997), Iaai (Ozanne-Rivierre 2004), Ifira-Mele (Clark 2002), Kairiru (Wivell 1981b, Ross 2002e), Kele (Ross 2002f), Kiriwina (Senft 1986), Kiribati (Groves, Groves & Jacobs 1985), Kokota (Palmer 1999), Kosraean (Lee 1975), Kwaio (Keesing 1985), Kwamera (Lindstrom 1986), Label (Peekel 1930), Lenakel (Lynch 1978c), Lewo (Early 1994a), Longgu (Hill 1992, 1997), Loniu (Hamel 1994), Lukep (Pono) (D’Jernes & D’Jernes n.d.), Lusi (Counts 1969), Manam (Lichtenberk 1983), Mangap-Mbula (Bugenhagen 1995), Marquesan (Lynch 2002b), Mekeo (Jones 1998), Merei (Chung 1998), Minaveha (Lovell 1994), Mokilese (Harrison 1976), Motu (Lister-Turner & Clark 1954b), Mussau (Ross 2002b), Mwotlap (Crowley 2002b), Nadrogaa Fijian (Geraghty 2002), Nakanai (Johnston 1980), Nalik (Volker 1998), Nêlêmwa (Bril 1994), Nguna (Schütz 1969), Niuafo’ou (Early 2002), Notsi (Erickson & Erickson 1992), Nyelâyu (Ozanne-Rivierre 1998), Paamese (Crowley 1982), Pileni (Næss, forthcoming), Puluwatese (Lynch 2002c), Ramoaaina (Davies & Fritzell 1992), Roviana (Corston-Oliver 2002), Saliba (Margetts, forthcoming), Samoan (Mosel & Hovdaugen 1992), Siar (Ross 2002c), Sinaugoro (Tauberschmidt 1999), Sio (Clark & Clark 1987), Sobei (Sterner & Ross 2002), Sudest (Anderson & Ross 2002), Tamambo (Jauncey 1997), Tawala (Ezard 1997), Tigak (Beaumont 1979), Tinrin (Osumi 1995), Tobati (Donohue 2002), Tolai (Mosel 1982, 1984, Rinderknecht 1987), Tongan (Churchward 1953), Ulithian (Lynch 2002d), Wailevu Fijian (Ritsuko Kikusawa pers. comm.), Wayan (Pawley & Sayaba, 2003), Woleaian (Sohn 1975), Xârâcùù (Moyse-Faurie 1995), Yabem (Dempwolff 1939, Zahn 1940, Ross 2002d), Zabana (Fitzsimons 1989).

Notes⇫