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

2. Types of locational case marking examined, in Central Australia

2.1 Terminology used

In this paper we are dealing with a number of phenomena in which the semantics of case do not line up with the form of case-marking, as with ‘Allative’ case being used in a ‘Locative’ meaning discussed above, so we need to be clear about which we are referring to. We shall refer to a set of ‘locational’ meanings which include locative, which refers to the static location of an entity, and allative, which refers to movement to a goal entity; where necessary we refer to these as ‘locative semantic function’ ; allative semantic function’ etc. On the other hand, there are local case affixes which we shall refer to by capitalised abbrevations eg LOC for Locative case; ALL for Allative case. So in the situation referred to as AN, locative semantic function can be expressed by LOC or ALL, depending on control variables.

Another terminological issue is that of non-finite subordination. In these languages there are typically non-finite verb forms which are variously referred to as ‘nominalised’, ‘gerunds’, ‘infinitives’ etc. by different authors. There may be good reasons for the choices in particular cases but these are not relevant to our purposes in this paper, so all such forms are referred to here as ‘non-finite’ (NFIN). A related issue is that of ‘coverbs’ which form the main lexical content part of two-part complex verbs in a number of languages under consideration here. These are variously referred to in different languages as ‘preverbs’, ‘particles’, ‘main verbs’ etc. Again they have different morpho-syntactic properties in different languages but these are not particularly in focus in this paper, and so we shall refer to them all as ‘coverbs’. In some languages such coverbs occur in subordinate clauses in a similar role to non-finite forms of verbs.

2.2 Local cases in Australian languages

Australian languages have systems varying from 3 to about 8 cases marked on nominals, generally as suffixes. These usually include ergative/instrumental and dative/benefactive especially in Pama-Nyungan languages. Most Australian languages have nominal local case suffixes, usually LOC (Locative: at, in etc); ALL (Allative: to, towards) and ABL (Ablative: from) with some additional ones in some languages. This is particularly true of languages of the Pama-Nyungan family, and in the case of this family at least two of these suffixes can be relatively confidently reconstructed to proto-Pama-Nyungan: LOC *-ngka,* -la (allomorphs determined by whether the host is bimoraic or polymoraic, respectively); ABL*-ngu(ru) (Alpher 2004); ALL is too variable across the family to be confidently reconstructed at this stage.

The prime functions of these local case suffixes are semantic, expressing ideas that in many languages would be expressed by adpositions, but are also grammaticalised in various ways in some languages. In this paper we discuss the three types of grammaticalised constructions listed and briefly exemplified below as AN, LS, and AS.

2.3 Construction AN: Allative expressing object control on NPs

ALL is used instead of LOC in a secondary predication where the subject of that predication has the same reference as an object or oblique in the main predication.

For instance in Gurindji one can say (1) with formal ALL marking on the object instead of LOC.²
(1) ngu-rna karrap nya-nya jurlak karnti-kurra

CAT-1sgS see see-PAST bird tree-ALL

‘I saw a bird on a tree’³

2.4 Construction LS: Locative marked subordination

LOC case-marking on the verb and other elements is used to mark types of non-finite subordinate clauses. There is some variation in the elements which are case-marked in subordination in different languages, both in the LS construction and the AS construction below. In most, non-finite forms of verbs receive the marking, but in addition other elements in the subordinate clause are also case-marked; in others, coverbs in addition or instead of the verb may be case-marked and in some, adverbs in main clauses are case-marked in a similar way.

In Gurindji LOC can mark a clause of simultaneous action on the gerund of a verb as in (2), with the subordinate object also marked LOC Locative, or on a coverb⁴ with the verb omitted as in (3). In both instances the subject of the gerund (NFIN) or coverb and the subject of the main clause are identical.
(2) Ma-rna-ni jarrak-ap kurrupartu-la yinka-rnu-la

say-IMPF-PAST talk-ACT boomerang-LOC scrape-NFIN-LOC

‘He was talking while planing a boomerang’
(3) Pulwarr-kuk jiya-wu talwirr-a-ma

dry-INCHOAT burn-FUT hang-LOC-TOPIC

‘It will dry hanging up

2.5 Construction AS: Allative marked subordination expressing object control

AS combines features of AN and LS, usually where LS is also present in the language.

Gurindji also exemplifies this type: if the verb of the main clause is transitive and the subject of the subordinate clause is the same as the object of the main clause, then the ALL may be used.
(4) Ngu=rna karu pura nya-nya yunpa-rnu-yirri.

CAT=1sgS child hear see-PAST sing-NFIN-ALL

‘I heard a child singing’
(5) Ngu=rna nya-nya lilaj-jirri warrija.

CAT=1sgS see-PAST swim-ALL crocodile

‘I saw a crocodile swimming’
Thus the fact that this construction involves a subordinate clause links it to the LS construction, and the fact that the subject of the subordinate clause is controlled by the object of the main clause links it to the AN construction.

3. Other treatments of grammatical uses of local cases

3.1 Outside Australia

Case-markers or adpositions denoting allative are is considered ‘notorious’ for both itstheir polysemy as a case-marker or adposition and its their heterosemy (crossing over to other word classes), and which has been the subject of attention in recent work, including within grammaticalisation frameworks (Rice and Kabata 2007:451). Rice and Kabata’s survey shows that it is not only the extension of the local case-form ALL to other functions (e.g. purposive: 2007:459) which is at stake here, but moves in the opposite direction from more abstract grammatical functions to the local ALL. While the function of marking non-finite subordination is not central to their article, they note the significance of (Rice and Kabata 2007:453):

the extensions or case syncretisms affecting allatives in particular, especially in languages where there is overlap between spatial goal-marking function and the marking of purpose or clausal subordination, as happens with English to, German zu, or French à – all classic allatives and all infinitive markers as well. (Rice and Kabata 2007:453)
Haspelmath (1989) also suggests the following source-to-target extension hierarchy:
allative > purpose > infinitive > complementizer
However, while the extension of ALL via a purposive use to non-finite clause marking can be found in Australia, the subject of the current paper suggests a quite different type of pathway linking allative to non-finite clause marking which is not purposive in current manifestations, or in origin.
The use of local case-marking to indicate clausal subordination is well attested outside Australia. Finnic languages are replete with such patterns (eg Tamm 2008 and p.c.) as are Daghestanian languages of the Caucasus (Forker 2008 and p.c.). Anderson (2002), as well as discussing case-marked clausal subordination in Burushaski, provides a brief survey of similar phenomena in other languages. The case-marked clausal subordination strategy is, he says, relatively common in morphologically rich (SOV) languages of Eurasia which make use of nominal cases. He exemplifies the construction from Munda languages, and mentions that it also occurs in Tibeto-Burman, Turkic, native languages of Siberia (macro-areal feature) such as Ket (Grishina 1979), Basque, and Yimas (New Guinea). He implies that it may have been even more widespread in the past but that it is relatively easily lost through contact with other languages which allow importation or calquing of a subordinator element. ⁵
3.2 ‘Fictive motion’ in Finnic languages
The AN phenomenon has parallels with another phenomenon in Finnic languages know as ‘directional locatives’ (eg Fong 2003) and ‘fictive motion’ (Humoo 2006, 2007, to appear). In some cases the allative (or the corresponding ‘interior’ case illative) is used for the static location of objects, not the directional path, as with the AN construction in Australia. For instance, consider the (i) interpretation of (6) in Finnish (Fong 2003:223-4): in this case the illative ( ILL, the ‘interior’ version of the allative) is used.
(6) Metsästäjä ampu-i karhu-n metsä-än

hunter.NOM shoot-past.3s bear-ACC forest-ILL

(i) ‘A/the hunter shot a/the bear in (lit. ‘into’) a/the forest.’

(ii) ‘A/the hunter shot a/the bear into a/the forest (i.e. the bear was

ejected/propelled into the forest).’
Actual motion of the object into the location can be involved in the meaning of such examples (as in interpretation ii of 6) but commonly there is no real motion. The common element seems to be that the object remains in a ultimate position in the location in the ultimate phase of the event. The allative and illative are commonly used for the location of objects with verbs of ‘leaving’ in Finnish, where the subject is separated from or moves away from the object entity (Huumo 2006: 48; 2007:85).
This is similar to the AN constructions in some Australian languages where both the subject and object can be in the location in a general sense but the key point is that the ultimate location of the object is the Allative marked NP, as in the Wakaya sentence (33) below, ‘the dog drowned the kangaroo in(to) the water’.
Fong attributes this unusual usage in Finnish to the diphasic nature of the event structure involved, and the fact that Finnish directional cases can represent one phase – in this case the ultimate location – rather than inherently Path. She relates the phenomenon in Finnish not to grammatical object control but to the semantics of the verbs involved, and the occurrence of the allative and illative for location in intransitives seems to bear this out e.g. (7).⁶
Finnish

(7) Karhu kuol-i metsä-än

bear.NOM die-past.3s forest-ILL

‘A/the bear died in (lit. ‘into’) a/the forest.’

Also in the Finnic languages is the construction where the location of the object is marked ablative (or elative, the ‘interior’ equivalent). This is found with a wide range of transitive verbs including perception and possession. The subject is understood as outside the location and some property of the object fictively ‘moves’ to the subject’s dominionain of possession or consciousness (Huumo to appear:12) .
Finnish

(8) Löys-i-n seinä – stä halkeama – n

find-PST.3sg wall – ELA crack- ACC

‘I found a crack in [lit. ‘from’][ the wall’

Fong (2003) gives other examples where the explanation that a property of the object moves into the domain of the subject does not appear appropriate. In (9) the resultant broken pieces of the vase can be understood to remain on the shelf.
Finnish

(9) Kissa rikko-i maljako-n hylly-ltä

cat.NOM break-PAST.3s vase-ACC shelf-ABL

‘A/the cat broke a/the vase off a/the shelf.’

In all the cases of uses of ‘directional locatives’/’fictive motion’ the subject and object of the verb are perceived as being in separate places, with the subject outside the place, at least at the end of the process. We return to the relevance of the Finnic data to the Australian cases at the end of the paper.