Moving along the grammaticalisation path: locative and allative marking of non-finite clauses and secondary predications in australian languages

4.4 Warumungu (Pama-Nyungan)

In Warumungu, as in Wakaya and the Arandic languages discussed in the following sections, the Ergative, Locative and Instrumental have the same form. In Warumungu this is due to complete vowel assimilation to the preceding vowel. This suffix (LOC/ERG) is used for location in intransitive sentences (27), while the ALL suffix on the object is used for location in transitive sentences (the AN construction), regardless of whether the object only (28), or the subject and the object (29) are at the location (examples from Simpson 2002).

(27) Karnanti nyi-nta warlukun-jju.

mother sit-PRESENT fire-LOC/ERG

‘Mother sits by the fire’
(28) Ngulya-kana ajjul kupunta kuyu maraji.

hole-ALL 3plS cook-PRESENT meat euro

‘They (more than two) cook euro in holes in the ground.’
(29) Karttingki ngurraji-kina kayin karta-nta.

man-Actor camp-ALL boomerang trim-PRESENT

‘The man is trimming a boomerang at home. ’
LS is not found in Warumungu, but note remarks below about the possible historical origin of the ‘same subject’ form in an LS construction.
AS is found with object control subordination in Warumungu.
(30) Nya-nyi arnangkku kanyi-nji-kina.

see-PAST 1sgS.2sgO step-NFIN-ALL

‘I saw you stepping (on it). ’

This contrasts with a complementiser suffix, -kil/-kul, used just for subject control (glossed here ‘while’). This is not the same as the current Warumungu LOC/ERG case suffix (various allomorphs including –ngkV, -njjV)¹² but historically the –l could descend from proto-Pama-Nyungan *–la LOC with a preceding ligative element; see 4.4 below for a possible cognate form ligative +rl in Wakaya.
(31) Kirriji wangka-n warlukun-kul parrpi-nji-kil.

woman talk-PRESENT fire-while light-NFIN-while

‘The woman is talking while lighting the fire’

4.5 Wakaya (Warluwarric, Pama-Nyungan)

Wakaya, south-east of the Eastern Mindi languages of the Barkly Tablelands, has the AN construction. In contrast to Warumungu above, in Wakaya the AN construction (ALL on object-controlled secondary locative predication) is reportedly only found where only the object is in the location, not where both subject and object are. In (32) and (33), the subject agent and the object are in the same location and physically interacting but only the object ends up in the specific location ‘ground’ and ‘water’ specifically.¹³

(32) Kurrerniy pulu, ngintuwerniy miingkert

die-PAST that, bury-PAST ground-ALL

‘He died and they buried him. ’ [Eastern Wakaya]
(33) Inkapurl wera ngurrungunta yukarrirt.

kangaroo-ERG dog drown-PAST water-ALL

‘The kangaroo drowned the dog in the water’
The use of ALL to mark object controlled and spatially separated objects contrasts with the use of the LOC for events seen as wholes. Whether the criterion for selection of LOC is lack of physical separation between subject or object or predication of the location on the whole event rather than the object needs further investigation, but examples like (32) and (33) suggest some version of the former (where ultimate location of the object separate from the subject invokes ALL).
Regarding non-finite subordination, Breen writes: "Simultaneous action and immediate consequence are expressed by a nominalised form of the verb marked with the operative [ergative]/locative case suffix if the subjects of the two actions are the same, or the allative case if they are different." That is, Wakaya also has the LS pattern where LOC -rl is attached to the verb following a ligative –nge- and is used for simultaneous action "if the subjects of the two actions are the same". (34) illustrates LOC/ERG as same-subject marker.
(34) Wirriwirriyarn penkangerl marniy yungkumet

along-1sgS go- NFIN-LIG-LOC/ERG kill-PAST bluetongue.lizard

‘I killed a lizard while I was going along’
If the two subjects are not the same, then there are several possibilities. Breen states that the ALL is used if the two subjects are different (i.e. the AS pattern).

ALL as different subject marker

(35) Jirrewenhthariyarn, nguninhthangert.

spear-PURP-1sgS sleep- NFIN-LIG-ALL

‘I'll spear him while he's asleep’ (E Wakaya)
(36) Ingentiyarn kerewul therremanhthangert.

see-PRES-1sgS child-PL play- NFIN-LIG-ALL

‘I'm watching the kids playing’
(37) Wayipalarrarn ingenyji kilyalirt nenkerrangert.

white.man-1sgS see-IMPF meat-ALL cut-NFIN-LIG-ALL

‘I've been watching the white fellow cutting up a bullock’
Note the use of the ALL on the subordinate object as well as the verb in (37), as in other languages like Gurindji. All the examples Breen gives are of object control, not of control by non-objects.¹⁴

4.6 Kaytetye (Arandic, Pama-Nyungan)

It is unclear if Kaytetye has a full-blown AN construction. Turpin (2000:48) gives the following example where the LOC could be used in other languages rather than the allative –(w)arle, but on the other hand it does involve a ‘putting’ action which may be conducive to use of allative.

(38) Weye (re) eletnhenke apmere kwereyeng-arle

meat (he) puts.down camp his-ALL

‘He puts the meat down in his camp’
The allomorphs -le, -nge are used for both ERG and LOC (due to neutralisation of an original difference in vowels in Arandic). Object control does not cause LOC to be changed to ALL in (39), but this is a case perhaps of the whole event being in the scope of the location, which takes LOC in some languages with AN constructions.
(39) Elye-nge-pe atanthe-pe pweyayne-ee

Shade-in-FOCUS they-FOCUS cook.used.to-CONTINUE

‘They would cook (the meat) in the shade for a long time (Turpin 2000:83)
-le following a ligative ­nge- (like Wakaya) is used for same subject ­ Locative subordination (glossed as ‘same actor’ by Turpin 2000:116-117)
(40) Ayenge ane-nge-le angkerrane

I.ABS sit-LIG-LOC talking

‘I am sitting talking’
(41) Atye aherre are-nge-le aytnenyerre

I.ERG kangaroo saw LIG-LOC speared

‘When I saw the kangaroo I speared him’
ALL ­warle following a ligative ­nge is used ‘where the doers of the two

sentences are different’. (Turpin 2000:117). Turpin glosses them as ‘same and different actor’, but the form is identical to ligative+LOC/ERG (LS) and ligative+ALL (AS).

(42) Artweye atye arenherre ape-nge-warle

man I saw go-LIG-ALL

‘I saw a man going’
(43) Artweye atye arenherre ape-nge-le

man I saw go-LIG-LOC

‘I saw a man as I was going’
In other Arandic languages such as Alyawarre (Yallop 1981) a form which is otherwise used to denote dative case the case can be used as in (45)-(46) instead of LOC as in (44). The contrast between LOC and DAT in this language may be making a similar distinction as that between LOC and ALL in other languages including perhaps the northern Arandic language Kaytetye: LOC indicates the whole event is in the scope of the location but ALL restricts it to the main clause object alone. However (47) casts doubt on this interpretation, since DAT is used where both subject and object are in the same location.
Both S and O at location:
(44) Amulya akngirra plain-ila atuna.

lizard many.NOM plain-LOC kill-PCON

‘(We) killed a lot of lizards on the plain’
DATIVE as locational OBJECT only
(45) Antimirna arrpmila arika, atnkirrika

sugarbag.NOM again see.PAST coolibah-DAT

‘(I) saw a sugarbag again, in a coolibah tree’
(46) Aylinapa line-irnima kwatjika-ntiya iwulalpika

we (du nom 1)+but line+SOME+NOM water+DAT+STILL leave+PAST

‘But we left the lines by the water’
DATIVE with both SUBJECT and OBJECT in location
(47) Artula arrpiyla ayliyla apmirika.

man+ERG carve+PRES boomerang+NOM camp+DAT

‘The man is carving a boomerang in camp’
The reasons for choosing DAT over LOC are not entirely clear in Alyawarre, and we may be dealing with a degree of pure optionality here, as is the case in the choice between ALL and LOC in a number of languages where AN occurs. This connection is all the more salient because of the frequent overlap in function between DAT and ALL case-markers in many languages: DAT used for destinations and ALL used for purpose etc. The difference between the northern and southern Arandic languages in this regard requires explanation; possibly Kaytetye has been affected by structural diffusion from its non-Arandic northern neighbours, but this requires more research.