Previous Next

Aboriginal Placenames

16. ‘Where the spear sticks up’

The variety of locatives in placenames in the Victoria River District, Northern Territory

Patrick McConvell1


In the Victoria River District (VRD) of the Northern Territory, the citation and neutral form of many placenames is marked with the locative case, both in the Pama-Nyungan (PN) languages (Eastern Ngumpin) and in the non-Pama-Nyungan (NPN) languages (of the Jarragan, Mindi and Yangmanic2 families). In the central area of the district both the PN languages (Ngarinyman, Birlinarra, Gurindji) and the NPN languages (Jaminjung, Ngaliwurru, Nungali) there is often an additional element added to the locative, which in the modern languages has a variety of functions including ‘exactly’ (for discussion of semantics of such suffixes, McConvell 1983; Schultze-Berndt 2002). The actual form of the elements is unrelated in the different families but the patterning is highly similar.

The purpose of this paper is to investigate how these patterns – LOCATIVE and LOCATIVE+EXACTLY – came to be distributed across language families linguo-genetically and typologically distant from each other. If there was structural diffusion what was the source and how did it occur?

Firstly it is pointed out that the locative patterning is not universally present in the district. In the Mudburra territory in the south-east it is very rare and another placename suffix, not found elsewhere in the district, is common, which has cognates in a similar function in a wide area of Pama-Nyungan. This provides a hint that the locative pattern may have been borrowed by other Ngumpin languages because of contact (adstratal or substratal) with the NPN languages.

Before approaching this question however, it is necessary to examine just how similar the use of locative is across the languages. Use of locative as part of placenames is widespread among NPN languages not only in the VRD but commonly in the Kimberley, Top End and Arnhem Land. In these languages, generally, the locative form is used as the citation/neutral form AND as the locative form. In the PN languages with locative placenames, however, the locative form is not used in the locative function – another locative suffix is added. This does not negate the hypothesis of structural diffusion from the NPN languages but shows that the pattern was not simply copied in all its detail, but was adapted to the PN grammar.

The paper then moves to looking at the LOCATIVE+EXACTLY pattern in more detail. This has not been reported from other regions to my knowledge. Again while at first sight the PN and NPN neighbouring languages which have this formation appear to mirror each other, there are differences. Most placenames consist of either a nominal or a coverb3 with this pair of suffixes. In Jaminjung/Ngaliwurru, there are other placenames which do not have a LOCATIVE suffix, but simply an EXACTLY suffix, which in these languages can signal secondary predication; this is not possible in the Ngumpin languages. In a few names in the PN languages there are phrasal placenames with the configuration:


While no phrasal placenames have been so far collected in pure Jaminjung/Ngaliwurru, there is one curious example of a code-switched placename in which the first part is a Ngarinyman noun with LOCATIVE and the second is a Jaminjung coverb with EXACTLY – following the different pattern of Jaminjung grammar above.

These patterns could also suggest a different hypothesis, that the current placename derivations reflect not only older locative patterns in PN but also patterns of non-finite subordination in the respective languages. The presence of -LOC+rni (= the EXACTLY morpheme in Ngumpin) as a form of non-finite subordination in Warlpiri, a Ngumpin-Yapa language closely related to Ngumpin, might suggest an inherited rather than borrowed source for the LOC forms.

Locative-marked placenames

Inherent locatives

In many areas of Australia a sizable proportion of the placenames bear locational case marking in citation form. This is true of both Pama-Nyungan and non-Pama-Nyungan languages, although it is more prevalent in the latter.

For instance Harvey (1999: 174) cites the following example from the Wardaman language in the north-east of the VRD.

  1. 1. Wardaman Jarrug-ja egg-LOC

The normal interpretation of this inflected noun would be (a) below.

  1. a. ‘at the eggs’
  2. b. ‘where the eggs are’

However, as Harvey points out, the (b) interpretation is more likely. This type of locative refers to a Dreaming being or an object left by such a being. Harvey considers this interpretation as “deriving from a headless relative clause with an existential meaning” and this analysis is endorsed by Baker (2002: 109). However even if correct, and notwithstanding the extended discussion of why unmarked placenames function as locatives (Harvey 1999), this does not explain clearly why the locative can have a non-locative function in some of these languages, such as Wardaman.4 It is not the case, for instance, that other locatives apart from placenames can be interpreted as nominative/absolutive noun phrases in these languages. Harvey’s discussion revolves more around the parallels between absence of locational case-marking on placenames in these Australian languages and in colloquial English such as ‘I’m going down the beach’.

Harvey (1999) lists a number of languages where “place names generally fail to take locational marking in circumstances where other nominals would bear it”. The NPN languages are as follows: Alawa, Gooniyandi, Mangarayi, Marra, Ngalakgan, Ngandi, Nunggubuyu, Wardaman and Warndarang. They are shown on Figure 16.1.

Additional NPN languages with this pattern include probably all NPN languages in the Kimberley: Bardi (Nyulnyulan), Worora (Wororan), Kija and Mirwung (Jarragan). Harvey’s list is shown together with these, and a few PN languages having this feature, on Figure 16.1. Some of these Kimberley languages also have inherent locatives (a locative suffix incorporated into the placename as its citation form).

Figure 16.1: Languages with inherent locative placenames


A Bardi example is the following (Aklif 1999: 81, 150, 179)

2. Bardi Jilarnboo-goon porcupine_fish-LOC ‘Evans Rocks’

In the eastern Mirndi languages, there is no evidence in published sources of inherent locative suffixes, nor of absence of locatives on placenames used with locative function in Jingulu (Chadwick 1997; Pensalfini 2003). In Wambaya, a Mirndi language further east, the situation appears to be the same, although one placename in Nordlinger’s wordlist (1998), Ganjarrani (a Gudanji placename), may be analysable as including -ni the locative suffix (ganjarra is not glossed).

Jaminjungan, the western branch of Mirndi, has quite a high proportion of inherent locatives, particularly the LOCATIVE+EXACTLY variant which will be discussed below.

Not all NPN languages north of Mirndi have the inherent locative type of system. ‘Place’ suffixes in Kunwinykuan (Harvey 1999) are distinct from the locative type.

A cursory glance at published descriptions of PN languages in Central Australia (Arandic), Western Australia and the south-east does not turn up any examples of inherent locatives in placenames, although discovery of examples through more directed research should not be ruled out. Apart from the examples in central Ngumpin languages in the VRD, to be discussed, inherent locatives do appear in two other regions – Cape York Peninsula (Yir Yoront, cited by Harvey from Alpher (2002)) and South Australia around Adelaide.

The selective sampling of inherent locatives carried out here is summarised on Figure 16.1.

In these languages placenames are marked locative in their citation forms and in their unmarked forms, e.g. when used as intransitive subjects.


In some – but not all – of these languages the inherent locative forms are also used for the locative function (‘at’, ‘in’ a place) even if the locative morphology is archaic and cannot be omitted. The presence of this distinction in the syntactic behaviour of inherent locatives in different languages is the basis of a typology which we shall examine in more detail in relation to VRD languages.

A second major difference between languages is that in some languages (mostly NPN), placenames which are neither locative currently, nor can be reconstructed as locative, function as locatives in the syntax. I will refer to this situation as ‘zero-locative’. Whether this is confined to placenames or occurs with other spatial nouns varies between languages (Harvey 1999 cites cardinal direction terms as behaving similarly to placenames in some languages) and is not investigated comparatively across Australia here. There is some overlap between the case of ‘archaic inherent locatives’ and the ‘zero-locative’ situation which would require detailed examination for each language. Koch (1995: 47-48) also discusses locatives in Australia as developing into ‘morphological zeroes’ in contexts like placenames.

Unmarked placenames functioning as locatives also occur in some dialects of Aboriginal English/Kriol.

  1. 3. I been sit down Darwin.

This usage appears to be a substratum effect: where locative is used in the traditional languages of the area, a preposition generally longa/la is used in the pidgin/creole variety.

  1. 4. I been sit down la Darwin.

Sometimes in NPN languages the locative is not the current form but an archaic form, as in the PN language Yir Yoront. For instance Baker (2002: 125; see also Heath 1981: 92) points out that in Roper River languages the form with the oblique prefix nya- is used for placenames, which was probably the older way of marking locative. In the modern languages there is additionally a locative suffix, and this occurs with some placenames. This appears to point to a historical layering of different forms of placename formation, although Baker does not go into the ‘stratigraphic’ implications of this.

Placenames as indicators of previous occupation

Baker (2002: 123) shows that in Ngalakgan terrritory, transparently Ngalakgan names intermingled with Alawa and Marra names, which are not however synchronically transparent. He interprets this as pointing to a prehistoric move into this area of the Ngalakgan group, although this could not be recent, based on linguistic evidence.

Harvey (1999) also points to a number of examples in the ‘Top End’ of the Northern Territory where movement of groups into the territory of others yields a layering of different types of placename structures as well as different language sources for the component elements of the names.

Hercus et al. (1997: section 4.1.1) note the similarity of the formation of Kaurna and Ngarrindjeri names, both featuring inherent locatives (the latter more consistently than the former). Some Kaurna names appear to be phonologically adapted forms of the Ngarrindjeri forms with the appropriate Kaurna locatives added. Although they do not make this proposal, this may be due to Kaurna taking over Ngarrindjeri names. In Central Australia, areas which are currently the territory of speakers of Western Desert dialects have placenames which are clearly adaptations of earlier Arandic names, reflecting eastward spread of the Western Desert language in the last millennium (cf. McConvell 1996).

Placenames in the VRD

In what follows the placenames of the VRD will be analysed to determine the extent to which the inherent LOCATIVE and LOCATIVE+EXACTLY patterns, which are found in both PN and NPN languages, are the result of the adoption of the pattern by one group from the other. Secondly the distribution of these patterns and other examples of ‘foreign’ patterns and roots in placenames will be analysed to determine if some languages have taken over the country of other language groups in recent prehistory.

Language families and subgroups in the VRD

The VRD at first European contact in the 1880s and in the succeeding century exhibits a great deal of superficial similarity in culture and social practices. The different groups engaged in intensive contact, intermarriage, and ritual and subscribed to a roughly parallel set of laws and customs. However the languages, despite quite substantial borrowing, were fundamentally different, and some aspects of the kinship systems remained different despite superficial similarity. Not only does the major divide between Pama-Nyungan and non-Pama-Nyungan languages pass through the middle of the region, but there are three distinct NPN families represented which are substantially different from each other in vocabulary and grammar.

Figure 16.2 shows the language families and groups in the Victoria River District.

Figure 16.2: Language families and groups in the VRD


The Eastern Ngumpin languages

The Eastern Ngumpin languages are a branch of the Ngumpin-Yapa subgroup of languages, established using the comparative method as exhibiting a set of common innovations (McConvell and Laughren 2004). Geographically Eastern Ngumpin languages occupy a ‘bulge’ into the northern riverine region from the territory of the other branches of Ngumpin-Yapa, Yapa and Western Ngumpin, which occupy the semi-deserts of the Tanami and Great Sandy Deserts to the south.

Among the Eastern Ngumpin languages, the central members Gurindji, Birlinarra and Ngarinyman are closely related, with Mudburra distinctively different.5

The Jaminjungan (Western Mirndi) languages

The Jaminjungan languages lie immediately to the north of Eastern Ngumpin around the lower Victoria River and they are the source of most of the borrowings and structural influence on Eastern Ngumpin. It has been shown that Jaminjungan languages are a branch of a wider Mirndi family which includes Jingulu, Wambaya and other languages in the Barkly Tablelands (Chadwick 1984, 1997; Harvey 2008).6 The two branches are discontinuous, being separated by Wardaman and Eastern Ngumpin.

Jaminjung and Ngaliwurru are said to be dialects of a single language (Schultze-Berndt 2000). Nungali, to the east, which no longer has any speakers, appears to be more conservative, retaining nominal class-prefixation which varies for case.

The Jarragan languages

These languages are centred in the East Kimberleys and are distinctively different from either Jaminjungan or Wardaman; they may be more closely related to Bunaban in the west. The northern languages Miriwung and Kajirrabeng do come into the Northern Territory in the north-west. Two extinct languages in the territory now covered by western Ngarinyman – Nyiwanawung (cf. Rose 1990, 2000) and Jiyilawung (Tindale 1974; Mathews 1901)7 – may have been Jarragan, but no details of the languages are known.


Wardaman is the most south-westerly of a group of related languages to the west and north and south of Katherine, including Daguman, Yangman and Wagiman. They may be part of, or closely related to, a broader Gunwinyguan family.

The role of placenames

Traditional function

Indigenous placenames throughout the VRD are thought by local Aboriginal people to be associated with the deeds of the Dreamings, the creative figures which roamed the land at a distant era. The stories of the Dreamings are still told and many are celebrated in song and ritual, but much of this lore and knowledge is being lost. Many of the placenames have meanings which relate to events in these Dreaming stories, although some have no transparent meanings, perhaps because they are too old and relate to words and languages which are no longer remembered. Those which have meaning may be a single item like a noun, but more often have some kind of morphology like a locative suffix or a ‘having’ suffix, or some kind of suffix whose meaning is obscure.

The roots of placenames are also used (with different suffixes) as personal names of people. This indicates a special relationship between the named individual and the named place. Often the place is part of a songline within the traditional country of the individual’s clan estate.

Historical impact of Indigenous placenames

As compared to other areas of Australia, rather few Indigenous placenames have been adopted in the Victoria River District by the European settlers; rather names of early settlers or evocations of the early settler history generally providing the source of contemporary names. Nevertheless there are several dozen Indigenous placenames which have been used for some time and which are found on maps of the region (not spelled using any systematic orthography).

Contemporary significance of placenames

Since the 1970s there have been some attempts by white authorities to assign Aboriginal names to places in the VRD, particularly Aboriginal settlements. For various reasons (detailed in McConvell 2002a) the wrong names have usually been given to places, and the standard of spelling has not significantly improved from the early days, although better information has been available.

A substantial amount of land in the VRD has come into the hands of Aboriginal people since the 1970s through the Northern Territory Land Rights Act, although this has not been always matched by the Aboriginalisation of the placenames or the use of systematic spelling. In the last decade, several cattle station properties have been acquired by the Northern Territory (NT) government as National Parks, primarily the very large Gregory National Park, which includes several small Aboriginal communities inside it or near to it (see Figure 16.3). Recently the NT government has embarked on a ‘joint management’ strategy to allow Aboriginal participation in running the park along with the Parks and Wildlife Service. It remains to be seen what level of recognition of Aboriginal traditional knowledge and naming of places will take place.

Figure 16.3: Gregory National Park


Linguistic prehistory of the VRD

It has been proposed that the Eastern Ngumpin languages moved north into the Victoria River District in the late Holocene, probably around 2000 years ago, but the chronology requires stronger evidence than currently exists (McConvell 1996; McConvell and Smith 2004). It has also been hypothesised that this spread was based at least in part on language shift by resident populations, who were most likely speakers of the three non-Pama-Nyungan language families described in the previous sections (McConvell 1997, 2001).

Linguistic evidence for spread based on language shift includes:

  1. a. massive borrowing of riverine flora and fauna terms;
  2. b. strong structural influence of the northern complex verbs on the Eastern Ngumpin verbs while the basic deep differences in grammatical type were maintained (McConvell and Schultze-Berndt 2001);
  3. c. adoption of key kinship terms from the north as the kinship system adapted.

Economy and technology of the Eastern Ngumpin speakers is a blend of the southern desert culture from which their language arose (by hypothesis) and that of the northern NPN speakers. Seed-grinding technology and some of its accompanying terminology spread from south and west into the VRD, but other aspects of the terminology, and other technologies such as the fire-drill, are more in tune with the autochthonous VRD patterns (McConvell and Smith 2004). The evidence in this field as presently assessed does not point unambiguously in the direction of language shift by NPN speakers to the Eastern Ngumpin languages nor to heavy migration by PN speakers and displacement of the previous population. Biogenetic studies can be called upon to shed light on such questions, but evidence on these questions is not available in sufficient detail for this area to be of assistance.

Adoption of placenames and placename structures from the northern languages would be another indication of Eastern Ngumpin language spread with substrate influence, but as already noted this is not a straightforward matter and will be examined in more detail in this paper.

Glossonyms and ethnonyms

For most of VRD, glossonyms (language names) and ethnonyms (names of groups of people) are identical. One exception is the term Karranga which signifies the language, whereas Karrangpurru is generally used to indicate the group (with the -purru suffix indicating ‘group’), although Karrangpurru can also be used to mean the language (as Ngaliwurru with the same suffix can indicate both language and people).

There are five recognisable ethnonymic stuctures which are used for language groups in the VRD (displayed on Figure 16.4):

  1. a. a root of obscure meaning with a suffix -man. McConvell (2002b, in press) argues that these roots are mostly characteristic interjections in the respective languages. -man is a semi-productive suffix in the Yangmanic languages from which this structure appears to be derived;
  2. b. a root of obscure meaning with the suffix -pung~-wumg, derived from Jarragan languages, and applied to them, with the exception of Jaminjung (?< Jaminji-wung);
  3. c. the root ngali (‘good’, ‘all right’) with a noun class prefix in the case of Nungali(wurru) and the ‘people’ suffix -purru~-wurru in the case of Ngaliwurru;
  4. d. a root with the form of an archaic environmental term (pirli-mat ‘hill-MAN?’; jiyil ‘?spring’) and the suffix -jurrung;
  5. e. a version of the Yangmanic (Wardaman) word korrong ‘south’ either as a bare stem or with the suffix -ji.8

Additionally there is the ethnonym Birlinarra, which appears to be constructed from pirli ‘stone, hill’ and an otherwise unattested suffix -narra.

The placenames based on ‘south’ in Wardaman are immediately to the south of the current territory of Wardaman with the exception of Kuurrinyji (Gurindji), which is separated from Wardaman by Birlinarra and Karrangpurru. Possibly at the time when the ethnonym Kuurinyji arose, languages ancestral to Wardaman and Kuurrinyji were contiguous. Various scenarios might be proposed but one is that Birlinarra and Ngarinyman which are closely related to Kuurrinyji, split off from the ancestral language located around the upper Victoria River, and occupied the sandstone hills to the north, creating the separation and leading to the adoption of a placename based on the term for ‘rock, hill’ (McConvell 2002). Ngarinyman has been expanding west even up until the early 20th century, replacing languages now extinct which were probably Jarragan.9

Structure of placenames

Placenames are divided into the following categories for discussion. Here ‘meaning’ indicates the intension of whole or parts of the toponym as a lexical item (e.g. ‘having a fig-tree’) and ‘reference’ indicates the onomastic extension of the term (the place it refers to).10

  1. 1. Opaque – neither the meaning of the term as a whole nor parts of it are known to current speakers.
  2. 2. Semi-transparent – either the whole or parts of the toponym have no meaning for current speakers in the current language, but they may be interpreted by comparative historical linguistics, or occasionally by speakers attributing an ‘old meaning’ to the placename or part of it.
  3. 3. Transparent – the whole toponym has a meaning for current speakers, whether it consists or one or more morpheme.

Figure 16.4: Ethnonyms in the VRD


Increasing opacity over time is a property of all long-lived naming systems and is not confined to Australian Aboriginal languages by any means. Many English placenames are semi-transparent in that the form and meaning of elements has changed since they were first coined and some of the component stems and affixes have not been used for some hundreds of years in the everyday language. Much the same is true of Aboriginal placenames, but we do not have the written documentation that can help us to discover the meanings of the names in most cases.


Fully opaque placenames are relatively rare in the VRD, but of course as knowledge recedes the composition and meanings of placenames become inaccessible. Tindale (1974) correctly points out that opaque names are old, but that is a relative matter and it may take only a generation or two for information to be lost. Examples of such names in Ngarinyman are Pukaka and Layit, associated with Flying-fox dreaming in the mid-Wickham River.

There may be names which could contain more than one morpheme, but since the morphemes involved are not regularly found, this could only be a suspicion rather than a proven fact. For instance the name Lipananyku, the large waterhole at Wave Hill settlement, may have once been segmentable (-ku may be a suffix, for instance), but that is only speculation.


These can be divided into three sub-types:

  1. a. those with an archaic stem
  2. b. those with an archaic affix
  3. c. those with an archaic stem and archaic suffix.

A simple example of the (a) type is the Gurindji placename Warrijkuny discussed in McConvell (2003). It is clearly segmentable as follows

5. warrij -ku -ny crocodile DATIVE NOMINALISER ‘belonging to Crocodile’

There is a transparent connection since the story of Crocodile is associated with this place and is still known. The only archaism here is that the word for ‘crocodile’ in the placename has the form warrij, not warrija, the current form for ‘crocodile’. The latter is clearly a relatively recent loanword from Mudburra and the older form warrij is found in Ngarinyman. This indicates that the placename must have been coined before the borrowing of warrija.

More common is the situation where the meaning of the stem is no longer known although the suffix is of a common type found with placenames – such an example is that of Ngaliminya in the next section, where -nya is very likely to be the common suffix, but people generally no longer seem to know what Ngalimi is (presumably a noun), and so far investigations of vocabularies of neighbouring languages have been unable to assist.

There are quite a number of suffixes in the languages of the VRD which seem to attach to stems to form placenames but are not interpreted as having a particular meaning.

Two found in Gurindji/Birlinarra/Ngarinyman are the following, exemplified here from the southern Ngarinyman area in the south of Gregory National Park (in the case of 6a) and from central Daguragu station.

6. Eastern Ngumpin placename suffixes of obscure meaning and origin

  1. a. -malulu e.g. Janka-malulu, in which the first element is Janka ‘woman’ and refers to a Women dreaming area
  2. b. -maliny e.g. Warna-maliny, in which the initial stem is warna ‘snake’ (not the most common word for this meaning today); the area is connected to the Jurntakal giant poisonous snake Dreaming.

Both these suffixes occur over a wide area and seem to refer to creeks, but what specifically, either in terms of geography or mythology, remains unclear. Possibly they are related to each other. The basic suffix could be -malu. The suffix -ny could be related to the Mudburra placename suffix -nya discussed below (Mudburra has a regular sound change of a-augmentation to consonant-final stems). The -lu is an ‘exactly’ suffix in western dialects like Wanyjirra similar in meaning to -rni. There are some other examples of -lu being used as a placename suffix without a preceding locative in western dialects Malngin and Wanyjirra, where the locative placename construction is absent e.g. Ngalja-lu (frog EX) ‘Frog Dreaming’.


Stems with no affixation which have a clear present meaning are relatively rare as toponyms but do occur, e.g. Karrikurn ‘yellow ochre’. In this case there is also yellow ochre at the site.

Many placenames consist of a stem (usually a noun, sometimes a coverb) with a known meaning equivalent to its current use, and a currently used suffix, or more than one suffix. The suffix may have its normal everyday meaning such as Gurindji/Birlinarra/Ngarinyman -jawung~-yawung ‘having’ as in:

7. Jampa-yawung ‘having a Leichhardt tree’

A place with a name like this may have a Leichhardt tree, or may have had one at some point, but the name is no longer simply a description – it has undergone the process of onymisation – conversion to a term which is mainly a referential label. Even such labels which tend to be descriptive may refer to the Dreaming, or mythological time, rather than the present or recent past.

Affixes may not have their normal function but may have a special function with placenames, as will be discussed with LOCATIVE and LOCATIVE+EXACTLY. Some such suffixes may be optional – both LOCATIVE and EXACTLY are known to be omitted on occasion but this is rare in the VRD.

Placenames with locative in the VRD

Distribution of locative pattern

Some kind of inherent locative (including the LOCATIVE+EXACTLY type) is found on a proportion of placenames throughout most of the VRD, Nowhere is this pattern categorical, but varies between around 20-40 per cent of the placenames in the central zone (Gurindji/Birlinarra/Ngarinyman, Jaminjungan and Wardaman). As we shall see, it becomes rare in the south-east (Mudburra) and western dialects related to Gurindji (Malngin, Wanyjirra and Nyininy). In the Jarragan languages in the west, however, locative is found with a great many placenames, and especially in Kija this seems more like a grammatical rule, that placenames shall take locative endings even in citation forms. Miriwung tends in this direction but there appear to be more cases of placenames not in the locative. Of a list of 221 Miriwung placenames compiled by Kofod (n.d.), 97 (44 per cent) are inherently locative in form (with the LOC suffix -m or allomorphs). Of these many are transparently formed from current Miriwung stems, like (8).

8. Miriwung

  1. a. Bananggam ‘hill at dam near Coolibah Pocket’
  2. b. bananggang ‘nightjar’ (Kofod 1978; -ng is the general nominal non-feminine suffix).

There is another feature already mentioned above, the zero locative, which is independent of the ‘inherent locative’ trait. In some languages placenames may function as locations without overt (additional) locative marking whether or not they have inherent locative marking. This feature seems confined to some of the NPN languages, never occurring in PN languages.11

In Mudburra country, in the south-east of the VRD, inherent locative marking of placenames is vanishingly rare. On the other hand there is another placename suffix which is quite common. Figure 16.5 shows the western limits of the distribution of this suffix which corresponds quite closely to the western boundary of Mudburra traditional territory. The examples of placenames with -nya are as follows:

9. Mudburra -nya

Wirntiku+nya ‘Curlew Dreaming Place’

Karu+nya ‘Child Dreaming Place’

Kirrawa+nya ‘Goanna Dreaming Place’

Janga+nya ‘Sickness Dreaming Place’

A gap in the distribution: Mudburra -nya

Figure 16.5: -nya and LOC distribution


There is some evidence supporting the idea that -nya may be an old placename suffix within the Pama-Nyungan family. In Nyangumarta (and other Pilbara languages) -nya is commonly found on names; e.g.

10. Nyangumarta placenames with -nya

  1. a. Wanangkuranya – rocky outcrop near Marble Bar
  2. b. Mikurrunya – hill at Strelley turnoff

< miku-rri ‘get jealous’ (the associated Dreaming story is about jealousy)

In the Western Desert dialects (particularly southern) -nya is an absolutive suffix used on all names (primarily personal names). A suffix *-ny(a) is reconstructable to proto-Pama-Nyungan which is the source of accusative and dative suffixes on pronouns and names in some languages, and is probably the source of the same suffix as an absolutive on names. This seems a more likely source for the Mudburra -nya than other look-alikes in the area, such as the West Mirndi feminine -nya.

Implications of the distribution

In terms of our initial question of where inherent locative originally came from, this distribution of -nya versus LOCATIVE on placenames is evidence which tends to weigh on the side of LOCATIVE on placenames being diffused from the northern NPN languages into the central eastern Ngumpin languages (Gurindji/Birlinarra/Ngarinyman) either by contact or by substratal influence (retention by NPN speakers as they shifted to PN languages). In this scenario, the Mudburra pattern of absence of inherent locative would be a result of failure of this diffusion to reach Mudburra. Similarly the -nya placename suffix could be interpreted as an older pattern which was not replaced by the new inherent locative marking.

Eastern Ngumpin (PN) plain locatives

Turning now to the main topic of this paper, the inherent locative marking of placenames, we shall look first at plain locative marking then at the rarer LOCATIVE+EXACTLY formation.


These languages are closely related and occupy a central column in the VRD, from around the headwaters of the Victoria River and Hooker Creek north to around Victoria River Downs and the western tributaries of the Victoria River. Ngarinyman also extends further west to the East and West Baines Rivers, although this may be relatively recent (as discussed further below). They all have plain LOCATIVE placenames as well as the LOCATIVE+EXACTLY type. Both of these types fade out in the west of the associated dialect chain (Malngin, Wanyjirra and Nyininy). This distribution might also be taken as support for the idea that the LOCATIVE patterns diffused from NPN languages in the north.

McConvell (2004: 47-48) proposed that the LOCATIVE type was relatively productive in these languages and was therefore used to form placenames from the name of the Dreaming known to have been at a site, where the original name had been forgotten or tabooed due to death or ritual secrecy. (See also Merlan 1998 for attribution of information to newly discovered sites near this region). The example given in McConvell (2004) was:

11. Jantura -la

bustard LOC

‘Turkey Dreaming place’

This behaviour was shown to be consistent with the fact that Ngarinyman plain LOCATIVE names are found in the north of traditional Ngarinyman country. Putatively they were assigned relatively recently because older names were forgotten or suppressed, and since Ngarinyman was the lingua franca in the area in recent times, this is the language used. On the other hand, names with the LOCATIVE+EXACTLY structure are confined to traditional Ngarinyman country prior to the expansion of Ngarinyman as a lingua franca. This more complex formation is not used for ad hoc bestowal of names in the way the plain LOCATIVE is.

Figure 16.6 shows the distribution of such Locative names both south and north of the Ngarinyman traditional boundary.

Figure 16.6: Locative and locative-exactly placenames on the Ngarinyman-Ngaliwurru boundary


The following are some representative examples of LOCATIVE placenames in Ngarinyman country. The positions are shown on Figure 16.7.

12. Ngarinyman placenames with LOCATIVE

  1. a. Lapa- ngka ‘Little Corella-LOC’
  2. b. Wirnirni-la ‘Emu chick-LOC’
  3. c. Kumuyurra-la ‘Butcher bird-LOC’

The characteristic of these placenames and the locative examples on Figure 16.6 above are that they are constituted from the name of a Dreaming animal (mostly birds in this sample but this may be fortuitous). Other placenames with LOCATIVE+EXACTLY have more varied roots, indicating other elements of the scene in Dreaming stories, coverbs representing actions etc., types which are not found in the plain LOCATIVE type of placename.

LOCATIVE placenames do not show any particular geographical distribution as compared with LOCATIVE+EXACTLY forms except that there is an area in the west which has LOCATIVE but not LOCATIVE+EXACTLY forms.

Figure 16.7: Ngarinyman/Birlinarra/Gurindji locative placenames (small sample)


There are a large number of Jaminjungan (Jaminjung, Ngaliwurru and Nungali) placenames with the locative suffix (in Ngaliwurru -ni following a vowel and -ki following a consonant, shown in Figure 16.8), as well as those with the LOCATIVE+EXACTLY pattern. Most of them are of the same type as those illustrated for the Eastern Ngumpin languages – the name of a dreaming with the locative suffix. However at least one, Kaljaki (below), is formed from a coverb representing the action of the dreaming kaljak ‘insert into hole’ +ki locative (with degemination of the k). This item is an exception to the generalisation proposed for the Eastern Ngumpin locative placenames that they are not formed from coverbs and that the LOCATIVE+EXACTLY pattern is used in such cases. However this seems rare.

Jaminjungan (NPN) Locative placenames

Figure 16.8: Jaminjung/Ngaliwurru locative placenames


13. Jaminjungan Locative Placenames

  1. a. Karra-ni Spider-LOC Head of Slatey Creek
  2. b. Muying-ki Black plum-LOC billabong near Legune Station
  3. c. Jamurru-wurru-ni Green plum-PL-LOC Roller Creek, east of Timber Creek
  4. d. Wukartij-purru-ni Rock Wallaby-PL-LOC Auvergne Crossing
  5. e. Jakarla-ni ?-LOC rock face northwest of Victoria River Crossing, also name for the location of the Pub
  6. f. Jurlu-ni ?-LOC across from Timber Creek a bit upstream from Yurlyuli
  7. g. Kaljak-i ‘insert into hole’-LOC Sundown Hill
  8. h. Marlirn-ki ?-LOC on Depot Creek.

And many others, e.g.: Yirrbangki, Karlbangki, Wirliyingki, Wamanki, Wangkangki, Waninyjanki, Minjilini, Yuntuluni, etc.

While the great majority of these kinds of placenames are within or very close to what is today considered Jaminjungan territory, a few are found to the south in what is currently seen as Ngarinyman country; e.g. Marlirnki (above), which is some 20 kilometres south of what is now thought of as the southern boundary of Ngaliwurru. This is likely to be a remnant name indicating that Jaminjungan languages once occupied an area farther south than they currently do.

Placenames with ‘exactly’ suffixes

Just as common as the placenames with plain locative, in both the Eastern Ngumpin and Jaminjungan languages, are those with a combination of a locative with a following suffix glossed here ‘exactly’. These types are exemplified respectively in (14), Birlinarra being Eastern Ngumpin and Ngaliwurru Jaminjungan.


  1. a. Birlinarra “Old Man Dreaming Place”Marluka-la-rni
  2. b. Ngaliwurru “Place where Python Dreaming tied up her topknot (wangurtu)”Wangurtu-ni-wung topknot-LOC-EX

The EXACTLY suffix in VRD languages

The Gurindji suffix -rni (-parni following consonants and -warni following liquids) has been described (McConvell 1983: 18ff) as functioning to modify expectations, and its function seems to be identical in all the Eastern Ngumpin languages. While some instances may be accurately translated into English as ‘only’, many are not, and there is a range of translations depending on context (i-vi in Table 16.1 below).

Table 16.1: Meaning of Eastern Ngumpin -rni

i. ‘only’

ii. ‘exactly’

iii. ‘even’

iv. intensive

v. ‘still’

vi. ‘all the time’

McConvell (1983) regards all these translations and functions as deriving from a single meaning. Unlike English ‘only’ in which the expectation is modified to a lesser target by the assertion, with -rni the expectation can be modified to something greater (‘even’) or something more specific (‘exactly, precisely’), also, not just to a lesser target. It is used also temporally to modify expectation of some event already having been completed (‘still’) or to modify expectation that an event has a limited time span more generally towards it being continuous (‘all the time’).

In the Non Pama-Nyungan languages Wardaman -peng ~-weng and Jaminjungan -pung~-wung have somewhat similar ranges of meaning to Eastern Ngumpin, but in Jaminjungan the suffix additionally functions to mark secondary predicates (Schultze-Berndt 2002).

It is not immediately obvious how this suffix relates to placenames. Harking back to Harvey’s suggestion that X-LOC in a placename signifies ‘where the X is’ one might think X-LOC-EXACTLY renders ‘where the X still is’ using a temporal sense of the expectation modifier, given that local people generally consider that some essence or power of the Dreaming inheres in sites visited by the Dreaming. This explanation is not totally convincing and we will return to the question of the origin of this type of word formation at the end of this essay.

LOCATIVE-EXACTLY placenames in Eastern Ngumpin

There are numerous examples of this pattern and only a small sample is given on Figure 16.9, drawn from around the eastern and western boundaries of where this type of placename is found. These include example (14a) above and those in (15) below. They include both nouns naming the animal, some other object associated with the dreaming and coverbs related to the Dreaming action.

Figure 16.9: Gurindji/Birlinarra/Ngarinyman LOC-EX placenames


15. Placenames with Locative-Exactly in Eastern Ngumpin

a. Jiwayirru-la-rni ‘bower_bird-LOC-EX’

b. Karna-ngka-rni ‘spear-LOC-EX’

c. Pula-ngka-rni ‘call_out[archaic coverb]-LOC-EX’

d. Warlmayi-la-rni ‘spearthrower-LOC-EX’

e. Kurrpu-karra-la-rni ‘dive-ACTION-LOC-EX’

f. Karu-ngka-rni ‘child-LOC-EX’

For (f) compare the Mudburra form Karu-nya discussed above in (9).

In the western dialects of Gurindji like Wanyjirra, the -rni suffix is replaced by -lu, which has the same range of functions as a discourse suffix as in Eastern Gurindji, Birlinarra, Ngarinyman and Mudburra -rni, and can be glossed EXACTLY, abbreviated to EX. In the western parts of Gurindji country around the headwaters of the Victoria River on present-day Riveren station, near the boundary of Wanyjirra dialect in the west, the placenames have the suffix sequence LOC+-lu instead of LOC+-rni, for instance:


  1. a. Jalwa-ngka-lu Heron- LOC-EX

Revolver Creek Junction, upper Victoria River

b. Jarnpa-ka-lu Grub/Featherfoot LOC-EX

Head of creek running south-east into upper Victoria west of Buchanan Springs

This seems to be a transitional region as locatives in placenames disappear altogether further west, although some names with -lu and without locative are found (see note 12).

LOCATIVE-EXACTLY placenames in Jaminjungan

Here are a few examples.


a. Karra-ni-wung Spider-LOC-EX

Hill overlooking the walking track near Timber Creek

b. Jipij-ki-wung bird sp-LOC-EX

near Marralum, salient rock left of road to M.

Red Ochre place, Dreaming little bird Jibij

c. Lirrimi-ni-wung Centipede-LOC-EX

Hill behind Garraniwung, Timber Creek area

d. Wungung-ki-yung Owl-LOC-EX

near Munbug, has a bottle tree that is fallen over

Dreaming for Owl (Wungung) who is looking back for Goanna (Goanna Dreaming at Laltha) at this place.

There appear to be no clear examples of LOCATIVE+EXACTLY placenames based on a coverb in Jaminjung. This would go along with the rarity of LOC on coverbs in Jaminjungan placenames; a historical explanation is attempted towards the end of the paper.

Jaminjungan placenames with exactly and without LOCATIVE

A different type of placename formation in Jaminjungan is the adding of the EXACTLY morpheme without LOCATIVE. This is not possible in Eastern Ngumpin.12 At least some of these are formations with coverbs and may relate to the rarity of the Locative with coverbs already observed. In these forms, -pung becomes -puk by Nasal Coda Dissimilation where the final ng is preceded by a nasal-stop cluster (see McConvell 1988). It does not appear that this is a current phonological rule in the Jaminjungan languages but it must have operated at the time of placename formation or afterwards.


a. Mun-puk face-down (coverb)-EX

mud flats near junction north of East Baines River (on Auvergne station)

b. Wularriny-puk ?-EX

Waterhole on East Baines, downstream from Bulla

c. Yalunjuk < Yalunju+wung

d. Mampuyuk < Mampi+wung?

Nungali – gender and case in prefixes

Nungali alone of the Jaminjungan languages has retained a four-class system for nominals expressed by prefixes (Bolt, Hoddinott and Kofod 1970: 69).13

19. Nungali gender/case prefixes


I masculine ti- nyi- ki-

II feminine nya- nyani- kanyi-

III neuter nu-/ni- nyi- ki-/ku-

IV plant ma- - ki-

Case inflection is expressed by use of different prefixes, and for some case functions also suffixes. For the neuter class of nominal the locative is expressed by just the prefixal desinence:

20. Nungali

  1. a. ni-yawa ‘foot’
  2. b. nyi-yawa foot LOC

According to Bolt et al. (1970) Nungali nominals which do not take prefixes normally express case by means of suffixes only. The locative is said to be -ki (1970: 72), although it is likely that it is also -ni following vowels as in the other Jaminjungan languages.

There are a few placenames that look Nungali both with ni- and niya- (absolutive) and nyi- (locative) scattered around. On the east at the border with Wardaman there are Nungali placenames in the same area as clearly Wardaman names; e.g. (21a-b) on the same hill.


  1. a. nyi -ma-ni-wung


  1. b. parnangka -ya

nightjar LOC (Wardaman)

(22) is an example in an area which is on or outside the western edge of Nungali country and what looks like a Nungali form with a locative prefix and suffix. The regular word for this eucalypt has a regular masculine Nungali prefix ti-; a prefixed form (timarlan) has also diffused throughout the VRD as the tree name, alongside the plain root (marlan).14

22. Nungali

Nyi -marlan - purru-ni

LOCI river.gum PL LOC

(Timber Creek area)

There is also a Wardaman placename Timarlam-beng with the same stem and a suffix related to the Jaminjungan -pung/-wung. The regular word for the tree in Wardaman is timalan ‘Eucalyptus camalduensis’ – no doubt a loanword from Nungali. The change of final -n to -m in the placename is inexplicable.

Niyampula, hills associated with Possum Dreaming in present Ngarinyman territory, is possibly Nungali or a related language. The folk etymology given by Ngarinyman people is that it is derived from Ngarinyman niyan ‘flesh’, which seems unlikely although this word also could be an old loan with a class prefix. There could be an old root *yampu(la) or *jampu(la) – if the -la is a Ngarinyman locative suffix this would be an interesting example of mixing of morphology from two languages. Example (23) appears to have a suffix -rti (which is possibly an allomorph of Jaminjungan LOC).

23. Wuyuwurti

rock face upstream from Nawulbinybiny, can be seen from Victoria Highway; also name for Old Victoria River Crossing

Wuyuwuj (Black Whip snake) dreaming, Wuyuwuj was menstruating there and then caught by a Lunggura (Blue Tongue) man in a cave; sacred place for women only

Phrasal placenames

There are some phrasal placenames in Gurindji and Ngarinyman country which consist of a noun marked locative followed by a coverb also marked locative. This is the same pattern as used in subordinate clauses in these languages. In the following name of a place on Wattie Creek in Jiyiljurrung Gurindji country, the clause is intransitive and the noun is the subject.

24. Gurindji

‘Where the devil died’

mararla-la tuk-kula

devil-LOC die-LOC

There are a few Ngarinyman non-finite phrasal names with nominal marked LOC and coverb marked LOC+rni, but I have not seen any Jaminjungan equivalents.

25. Ngarinyman “Where the spear sticks up”

Karna-ngka jirrp -kula-rni

spear-LOC stick.up LOC-EX

Stand of trees south of Wickham River

There is also one (possibly unique) phrasal name which combines both languages, Ngarinyman and Ngaliwurru.

26. Ngarinyman/Ngaliwurru mixed “Where the spear sticks up”

Karna-ngka that -puk

spear-LOC stick.up -EX

On East Baines River

Here both the lexical item and the suffix are Ngarinyman in the first word, and both are Ngaliwurru in the second half (the use of a Jaminjungan sound th in the coverb further reinforces the language difference), so there is a ‘code switch’ between the two words in the name. Interestingly the locative is found with the Ngarinyman noun, and the EXACTLY morpheme -puk (-pung15) with the Ngaliwurru coverb. This fits in with the placename formation in Jaminjungan noted earlier, that coverbs take this suffix in placenames rather than locative as is possible for Eastern Ngumpin languages.

In many languages to the north of the Victoria River District placenames can be formed from finite clauses without any additional affixation. This is also possible in Ngarinyman, although rarely, as in (27).

27. Ngarinyman whole clause placename


hill a long way (maybe 50 kilometres) west of Yarralin Community: the name refers to the two women who have lice on them that are biting them – if you go touch the rock you’ll get lice too.

This is as recorded by Caroline Jones, who is unsure of the parsing of the morphology (pers. comm.). Possibly it is as follows: 16

wudu -rlu =wula-nyunu baya-la

louse-ERG 3DUA-RFL bite-PRES ‘the two lice are biting themselves’

In Malngin, a western variety similar to Gurindji it is possible to have a phrasal placename consisting of the subject noun (including ergative suffix where appropriate) and a coverb – no inflecting verb – as in (28)

28. Malngin phrasal placenames

a. “Where the nail-tailed wallaby stands up’

Kururungkuny warawarap

n-t.wallaby stand

b. ‘Where the Bluetongue made a gap (by throwing a boomerang)”

Luma -ngku tingkalp


Recall that the use of locatives in placenames disappears in the Malngin area.

In south-western areas (Wanyjirra) there are also some phrasal names with finite verbs (not coverbs) without locative. In this example the verb form is not the current verb form in that language, but is identical to the present tense form in Birlinarra; this is probably an archaism.

29. Wanyjirra

‘Where the hat stays’

Kumununga karra

ritual.hat stay.PRES

Evidence of previous occupation

Jaminjungan names in Ngarinyman country

The presence of Jaminjungan names in Ngarinyman country has already been noted in relation to Marlirnki above. The evidence for names with Nungali features in Ngarinyman country is not strong, and it may be that this involves not Nungali as such but an earlier stage of Jaminjungan languages more generally when gender prefixes were still in wider use. However, aside from this there are some indicators that Nungali speakers originally had a more westerly distribution, including at least the Timber Creek area. What evidence there is of Jaminjungan placenames in Ngarinyman country indicates possible Jaminjungan occupation of an area as far south as the mid-Wickham River at a distant time, probably at least hundreds of years ago.

Names with Miriwung/Kajirrabeng -m LOC

Figure 16.10: Placenames with Jarragan -m Locative


To the west of the Jamijungan languages are the Jarragan languages of the East Kimberley (see Figure 16.10). Today the Miriwoong and Kajirrabeng have recognised territories inside the Northern Territory borders, and there is evidence that at least one other group existed (Nyiwanawu) in the nineteenth century along the West Baines River, in part of Amanbidji (Kildurk) station and neighbouring western Victoria River Downs, which probably spoke a Jarragan language but which succumbed to a combination of settler violence and internecine warfare (Rose 1990). Early sources (Mathews 1901, relying on information from N. H. Stretch, in turn obtaining information from a speaker of this language) name another language in this region also – Geelowng – probably Jiyilawung. The glossonymic suffix does not definitively prove that the language named is Jarragan, because people can adopt exonymic glossonyms from other language groups as their own, as probably happened with Jaminjung, which seems to have had the -wung suffix at some time in the past. However it is an initial clue that Jarragan languages were spoken further east than their current domain and the evidence below about locatives in placenames tends to support that view.

All the Jarragan languages appear to have a strong tendency to mark placenames with a locative suffix.17 They do not add any further locative when a meaning ‘at X place’ is required – that is, they are of the zero-locative type. In the northern Jarragan languages which are the ones relevant to us here, the locative suffix following a vowel is -m with allomorphs -me, -mi, -be following consonants. As expected there are quite a number of placenames in -m in the north-western corner of the Northern Territory where the Miriwoong and Gajirrabeng territories are. However there are also such -m placenames to the east in what is now Jaminjungan and Ngarinyman country. The westernmost examples are in the area probably previously inhabited by the Nyiwanawu and subsequently used by Ngarinyman mainly. Even further east however there are a few such names which may indicate that Jarragan languages were spoken over a wider area of Jaminjungan and Ngarinyman country at some more distant time in the past.

30. Placenames with Jarragan locative -m

a. Jaljini ~Jalyinim Mount Razorback

b. Jinmiyam Kneebone; Gajerrawoong or Jaminjung country? (Legune area)

c. Kurrpijim Bulla Community

d. Papikurnam On West Baines, Kildurk

e. Jurum On West Baines, Kildurk

f. Julngayim~ Tulngayim18Argument Gap (Kildurk area)

g. Thukparrim On Pinkerton Range

h. Lumukem near Nyunyuarim?

Bluetongue place, met up with Walujabi – Blackheaded Python (Kildurk area)

Lumuke- is the Jarragan word for ‘bluetongue’ (Kija lumuku-l (fem)). This confirms the presence of this language family in the area. This site is well outside current Jarragan territory and probably beyond the 19th century Nyiwanawu area.

(31a) is even further east in the southern central area of Ngaliwurru country, with the -am Locative suffix of Jarragan, suffixed to the name of a small brown bird (sp.?). There are other placenames formed from this bird species name, the plain stem (b) in Karrangpurru country and the Ngarinyman LOCATIVE+EXACTLY formation (c). Their locations are shown on Figure 16.10.


a. Tuwak-am

Bird sp. LOC Bob’s Yard

b. Tuwak East of Jasper Gorge

c. Tuwak-kula-rni

bird sp. LOC EX near Jutparra

Even further south-east is the site Takam near Jutparra. The meaning of the root here is unknown but the -m has a Jarragan appearance. Final -m is very rare in Ngarinyman, especially in nominals.

Hypotheses about placename locatives(+exactly) in the VRD

Structural diffusion from Non-Pama-Nyungan (especially Jaminjungan) to Pama-Nyungan (eastern Ngumpin)

The formation of placenames from locative forms of nouns for Dreaming beings and objects is fairly common in the area of NPN languages north of Eastern Ngumpin, Jaminjung, Wardaman and Jarragan. It is not found in other Ngumpin-Yapa languages or even in Eastern Ngumpin languages to the immediate east and west (Mudburra, Malngin) but only in the Central Eastern Ngumpin languages Gurindji, Birlinarra and Ngarinyman. The most attractive hypothesis therefore is that this pattern was borrowed from NPN, and probably Jaminjungan, into these languages, but the actual morphemes involved were not borrowed.

Locative placenames elsewhere in Pama-Nyungan

One might raise an objection that the locative means of forming placenames is also found in some Pama-Nyungan languages elsewhere, and that therefore it could be an inherited pattern in a few Eastern Ngumpin languages but lost elsewhere in PN languages of the region.

So with the Yir-Yoront tract name Puyvl which has an archaic locative suffix -vl, the citation form is also used as a locative as in (28) (Alpher 2002: 135)

32. Ngethn oylt artm athan ngathn oylt nhinvnh Puyvl

we there mother my we there stayed [place]

‘We, my mother and I, were staying there’

When such names occur with other locational cases, allative and ablative, the locative is retained prior to the other case marking; e.g.

33. I pal thalvnh Puyvl+uyuw

there hither returned [place] ALL

‘They were coming back this way towards Puyul’

This latter behaviour is reminiscent of the use of the Locative suffix prior to other case-marking in southern Western Desert languages such as Pitjantjatjara

34. Alice Spring-ala-kutu


Alpher suggests that the accretion of the locative to placenames has been a process of ‘historical drift’, citing Lauriston Sharp recording names without locatives in 1933 which Alpher recorded as having obligatory locatives in 1966.

A similar historical drift might have happened in South Australia: Teichelmann and Schürmann record some placenames without locative but the bulk of ‘Kaurna’ placenames are regularly marked locative. However the ‘drift’ in this case may well have been influenced by a more well-established pattern of locative marking in the Ngarrindjeri languages. As far as I know, such a contact explanation is not available for Yir-Yoront, but may be for locative marking in the PN languages of the VRD. We return to this issue at the end of the paper after looking at VRD placenames in more detail.

Inheritance of locative subordination within Ngumpin-Yapa

Another potentially relevant factor is that LOCATIVE is also used a means of marking non-finite subordination in Ngumpin-Yapa. In Gurindji for instance one can add locative to a coverb19 to indicate a simultaneous action by the main clause subject, as in (35). If the main clause verb is transitive, the coverb is usually ergative-marked instead.

35. makin-ta ngu=rna jarrakap marn-ani

lie.down-LOC CAT=I talk say-PST.IMPF

‘I was talking while lying down’

If the clauses do not share an NP referent, the subject may also be present with locative marking.

36. Jangala-la makin-ta ngu=rna jarrakap marn-ani

J.-LOC lie.down-LOC CAT=I talk say-PST.IMPF

‘I was talking while Jangala was lying down’

Coverb and object NP can also take locative marking in a subordinate non-finite clause where the object of the main clause is coreferential with the subject of the subordinate clause as in (37a). An alternative is to use allative marking as in (37b). This kind of pattern is areal but appears to originate in some Pama-Nyungan grouping (McConvell and Simpson 2009).

37a. Jangala ngu=rna paraj pu-nya ngarin-ta jaart-karra-la

J. CAT=I find pierce-PST meat-LOC eat-ACT-LOC

b. Jangala ngu-rna paraj pu-nya ngarin-jirri jaart-karra-yirri

J. CAT=I find pierce-PST meat-ALL eat-ACT-ALL

‘I found Jangala eating meat’

The interpretation of placenames could therefore be different based on this background of locative subordination. Makinta could mean ‘when/where [someone] was lying down’ and Jangalala could mean ‘when/where Jangala did something’. This is quite amenable to the kind of interpretation placenames with locative actually have in relation to Dreaming events. This is a distinctly different from the interpretation in terms of headless existential clauses offered by Harvey.

There are several problems with this solution however. The locative strategy for non-finite subordination is common in Ngumpin-Yapa languages (e.g. Jaru, see Tsunoda 1981: 172 ff), but the locative placenames phenomenon is quite restricted to the central Eastern Ngumpin languages, as we have seen. In order to probe this problem further we need to look at LOCATIVE+EXACTLY constructions.

LOC+EX subordination in Ngumpin-Yapa

LOCATIVE+EXACTLY placenames formation is even more restricted and more puzzling. It occurs only with the central Eastern Ngumpin languages and with the Jaminjungan languages, not in the other NPN languages to the north, nor in any other PN languages that I know of.

However one quite salient fact is that the combination of locative case marking with a morpheme -rni does occur, in the southern Ngumpin-Yapa language Warlpiri, marking a certain type of non-finite subordination.

This is one of a number of types of subordinating suffixes which are used to mark coreferentiality of subjects, or objects, or absence of such coreferentiality. The LOC+rni form is used among other functions as the default form appropriate when there is no relevant coreferentiality in a main clause. It is therefore most appropriate for expressing a general meaning of ‘where X happened/where X did something’.

Let us suppose for the moment then that such a clause type were present in early Ngumpin-Yapa, and while it was lost as a productive syntactic mechanism in the Ngumpin branch, it was retained in placenames. This hypothesis seems to be counter to the earlier idea being pursued that the locative pattern could have been borrowed from Jaminjungan into eastern Ngumpin, but as we shall see below, there may be a way of reconciling these two approaches.

It raises other issues too. In Warlpiri -rni apparently does not have the meaning ‘EXACTLY’ (etc. – note the range of senses discussed earlier) but only functions as a formal syntactic marker of a type of non-finite clause linkage with the LOCATIVE, defined by Hale (1995: 42) as follows in (38) with the example (39).

38. Warlpiri -ngkarni/-rlarni

‘relative: main clause arguments not coreferential with subordinate subject (the latter inflected for dative case)’20

39. ngalapi-nyanu ka manyu-karrimi kirda-nyanu-ku karli jarnti-rninja-rlarni

child-his AUX play-PRES father-its-DAT boom trim-INF-LOC+rni

‘The child is playing while its father is trimming a boomerang’

Simpson (1991: 391) clarifies that LOC+rni clauses in Warlpiri ‘describe an action which happens at the same time as event denoted by the argument-taking predicate in the nominalised clause’. This contrasts with another type of clause linkage, sequential, which is marked by -rla, which is the same form as for the Warlpiri locative case suffix, exemplified in (40).21

40. Warlpiri

wati-ngki kuyu purra-nja-rla nga-rnu

man-ERG meat cook-INF-LOC eat-PST

‘The man cooked and ate the meat’

It is possible, then, that the -rni element was added to the locative at some earlier stage of Ngumpin-Yapa to provide the simultaneous, continuous interpretation, rather than sequential, since this appears to a key sense of this element. The -rni /-parni element is also found as an ‘emphatic’ marker including this kind of meaning in Walmajarri (e.g. in texts in Hudson 1978: 88-92). In Jaru this suffix is replaced by -lu, as noted for western dialects of Gurindji, but the range of meanings is similar including ‘keep …-ing, still now’ (Tsunoda 1995: 210). This distribution of -rni EXACTLY in the west and east of Ngumpin-Yapa but not in the central languages is found with other elements also. A plausible interpretation of this distribution is that the morpheme in question is old in the sub-group and has been replaced by innovation in the central languages.

One might claim that this -rni suffix did begin as an expectation modifier and continued (and perhaps expanded) that function in Ngumpin while it also took on the function of marking a type of non-finite subordination also, at which time it also transferred to placenames. The grammaticalised syntactic function solidified as a cross-clause marker in Yapa (Warlpiri) but was lost in favour of the kinds of plain case marking illustrated above for Gurindji, in the other languages. In Jaru, plain locative marks non-finite subordination, and as in Gurindji, both the object and verb are marked locative as in (41). In (42) the verb is replaced by a ‘preverb’ (similar to Gurindji coverb, but with less freedom in position, see Tsunoda 1995: 173). According to Tsunoda, this construction is used when the main object is coreferential with the subordinate subject, and there is an alternative with the Allative replacing the Locative, as in Gurindji but seemingly more rare.

41. Jaru

ngumbirr-u mawun nyang-an ngaba-ngka nganu-ngka

woman-ERG man see-PRES water-LOC drink-LOC

‘A woman sees a man drinking water’

42. Jaru

ngumbirr-u yambagina nyang-an manyan-da

woman-ERG child see-PRES sleep-LOC

‘A woman sees a child sleeping’

The emergence of LOC+EX placenames as compromise solution

Let us assume then that at a certain stage of Ngumpin there was both a locative and LOCATIVE+EXACTLY non-finite subordination mechanism and that both of these were applied to yield placenames from events in the Dreaming stories. So phrases like ‘water-LOC drink-LOC’ could be interpreted as ‘where [some being] drank water’. More to the point single nouns or coverbs could have either LOCATIVE or LOCATIVE+EXACTLY added to yield placenames like X-LOC (EX) ‘where X did/was doing something, or had something done to it/her/him’ or in the case of a coverb ‘where some being did X’.

As far as we can tell, however, these patterns are not similar to or very compatible with Jaminjungan grammars at the relevant period. Neither Jaminjungan nor the other NPN families in the region would have had a gerund/coverb nominalisation pattern at all, and certainly not with locative case-marking or LOCATIVE+EXACTLY. However two factors could have disposed Jaminjungan to adopt this pattern:

  1. a. the common use of locative forms of placenames as citation forms, and along with it in some NPN languages the ‘zero-locative’ rule of not adding further locatives to such forms;
  2. b. the use in Jaminjungan and possibly other NPN languages of the region of expectation-modifying (restrictive) markers as grammatical markers of secondary predication and subordination.

Schultze-Berndt (2002) discusses the Jaminjungan -pung, which is similar in range of meaning to -rni in central Eastern Ngumpin but is moving in the direction of being a secondary predication marker. Further evidence that this played a role is in the discussion above of the use of -pung alone on coverbs to form placenames.

A fascinating example of how placename formation processes intertwined is the ‘code-switched’ example (26) Karna-ngka that-puk ‘where the spear stands up’, where the noun has just locative marking and the coverb just EX marking. On the other hand the almost parallel fully Ngarinyman example (25) Karna-ngka jirrp-kula-rni ‘where the spear sticks up’ has only LOCATIVE on the noun but LOCATIVE+EXACTLY on the coverb. In a subordinate construction in Ngarinyman omission of the locative case-marker on the coverb would be ungrammatical.

Thus it would seem that although the fundamental LOCATIVE+EXACTLY construction came from Ngumpin-Yapa (PN), the parts of it were sufficiently congruent with preexisting elements in Jaminjungan to allow a compromise in which LOCATIVE+EXACTLY became a standard placename pattern in Jaminjungan too. Traces of its origin remain for instance in the apparently relatively small number of LOCATIVE and LOCATIVE+EXACTLY placenames based on coverbs in Jaminjungan. Another factor influencing the situation may have been the relationship to other placename formation strategies involving subordination markers, e.g. the subordinating/relativising morpheme -ngarri in Jarragan and Worrorran. The notion of adding a subordinating morpheme to form placenames could have been calqued from Jarragan into Ngumpin initially, then the specific LOCATIVE+EXACTLY construction calqued from Ngumpin into Jaminjungan later. Finally, another factor smoothing the path to this convergence may have been the semantics of the -pung and -rni elements, both of which carry an element of ‘still’ – continuity in time against the odds. When people see the stand of trees that represent the spears thrown at the sites called (25) and (26) in the Dreaming long ago, surviving still is surely in their mind.

Typology of inherent locatives in placenames

Alongside these questions of the origin of particular placename formations, and interacting with them, is the issue of the syntactic behaviour of placenames. The notion of zero-locatives was introduced earlier and within the VRD alone we can see various settings of this feature. In the Jarragan languages it is quite prominent. In the Eastern Ngumpin languages it is virtually absent and where placenames with inherent locatives occur in a locative function, a second locative (or other locational marker) is added as in (43a) below. In Wardaman however, zero-locative is usually found as in (44), and this also extends to placenames which do not contain a Locative.22

43. Gurindji

a. Kujarra ngu-wula mamangkurl

two CAT-3du.s

wani-nya Karu+ngka+rni -la.

fall-PAST child–LOC-EX-LOC

‘Two (Dreaming children) ‘went down’ at Children-Dreaming-Place (Wave Hill)’

b. Ngu-wula ya-ni Karu+ngka+rni -yirri

CAT-3du.s go-PAST child-LOC-EX-ALL

‘They both went to Wave Hill’

44. Wardaman

a. Yurrb we-ndi julu-ya

stand 3sg-PAST hill-LOC

Barnangga+ya / Bijbarnang

bird sp.-LOC [place]

He stood on the hill Mt. Gregory / Bijbarnang

b. Wurr-yanggi Barnangga-warr / Bijbarnang-garr

3pl.go-PAST bird sp.-ALL [place]-ALL

‘They went to Mt. Gregory / Bijbarnang’

Table 16.2 provides an overview of the typology of placename behaviour in the VRD languages.

Table 16.2: Distribution of features of placenames in VRD languages

IL= Inherent Locative; EX = EXACTLY suffix; PN = placename; LOC = locative case ending; Loc = locative function; ? = no clear data obtained








IL with EX suffix







EX suffix, no IL





[coverbs only?]



Coverbs form IL’s +

Nominals only -







Other cases forms added to IL form +

root -







LOC case added to IL form in Loc function +

IL form used in Loc function -







IL form citation








Any PN used with LOC as loc








Phrasal with IL marking +

Phrasal without IL marking -







Phrasal with LOC+EX marking on coverb, LOC on noun







Phrasal with EX on coverb, LOC on noun




+ [?only code-switched example]



This array confirms the distinctiveness of the Eastern Ngumpin languages and Jaminjungan languages together in having the LOCATIVE+EXACTLY type of placename formation, despite the linguogenetic unrelatedness of the languages. The explanation for this remains difficult but a detailed hypothesis about the interaction of the languages leading to this result has been outlined in the last section, where the role of non-finite subordination/secondary predication in the history of placename formation is discussed.

A wider grouping of languages – including all the NPN languages of the region but not all the Ngumpin (PN) languages - has the feature of a significant number of placenames being what has been called here ‘Inherent Locatives’. Nevertheless the behaviour of such placenames is distinctively different in the different languages. The NPN languages largely follow the ‘zero-locative’ pattern in which inherent locatives (and other placenames) do not generally add locative case-marking in locative function, but the Ngumpin (PN) languages do. This indicates that while the PN languages may have been influenced by NPN neighbours to mark placenames locative, they did not adopt the ‘zero-locative’ grammar.

It has been shown that the LOCATIVE+EXACTLY pattern of placename formation may be some kind of hybrid made up of elements from both the Pama-Nyungan Ngumpin languages and the Jaminjungan languages during their extended period of contact and the spread of the Ngumpin languages north. There are intriguing hints that the LOCATIVE+EXACTLY pattern is related to another pattern of marking especially coverbs in placenames with the EXACTLY suffix only, not just in Jaminjungan but perhaps in the other NPN languages of the region with even some parallels in some PN dialects. Further data and analysis, including on the rather rare instances of clausal and phrasal placenames, are needed to shed light on this.


Aklif, Gedda (compiler) 1999, Ardiyooloon Bardi Ngaanka: One Arm Point Bardi Dictionary, KLRC, Halls Creek.

Alpher, Barry 2002, ‘The Archaism and Linguistic Connections of some Yir-Yoront Tract-Names’, in The Land is a Map: Placenames of Indigenous Origin in Australia, L. Hercus, F. Hodges and J. Simpson (eds), Pandanus Books, Canberra: 131-140.

Baker, Brett 2002, ‘“I’m going to where-her-brisket-is”: Placenames in the Roper’, in The Land is a Map: Placenames of Indigenous Origin in Australia, L. Hercus, F. Hodges and J. Simpson (eds), Pandanus Books, Canberra: 103-130.

Bolt, Janet, W. Hoddinott and F. Kofod (n.d.), Nungali, MS.

Chadwick, Neil 1997, ‘The Barkly and Jaminjungan Languages: A Non-Contiguous Genetic Grouping in North Australia’, in Boundary Rider: Essays in Honour of Geoffrey O’Grady, D. Tryon and M. Walsh (eds), Pacific Linguistics C 136, Canberra: 95-106.

Evans, Nicholas 2003, ‘Introduction: Comparative Non-Pama-Nyungan and Australian Historical Linguistics’, in The Non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region, Nicholas Evans (ed.), Pacific Linguistics 552, Canberra: 2-25.

Harvey, Mark 1999, ‘Place names and Land-Language Associations in the Western Top End’, Australian Journal of Linguistics, vol. 16, no. 2: 161-195.

— 2002, ‘Land Tenure and Naming Systems in Aboriginal Australia’, The Australian Journal of Anthropology, vol. 13, no. 1: 23-44.

— 2008, Proto-Mirndi: A Discontinuous Language Family in Northern Australia, Pacific Linguistics 593, Canberra.

Hercus, Luise, Flavia Hodges and Jane Simpson (eds), 2002, The Land is a Map: Placenames of Indigenous Origin in Australia, Pandanus Books, Canberra.

Hercus, L., H. Koch, D. Nash and J. Simpson 1997, ‘Placenames’, Paper to Second Workshop on Australian Languages, University of Melbourne.

Koch, Harold 1995, ‘The Creation of Morphological Zeroes’, in Yearbook of Morphology 1994, Geert Booij and Jaap van Marle (eds), Kluwer Academic Publishers, Dordrecht/Boston/London: 31-71.

Kofod, Frances 1978, The Miriwung Language (East Kimberley): A Phonological and Morphological Study, MA Thesis, University of New England.

— (n.d.), Miriwung Placenames’, MS.

McConvell, Patrick 1983, ‘“Only” and Related Concepts in Gurindji’, unpublished MS.

— 1985, ‘The origin of subsections in northern Australia’, Oceania vol. 56: 1-33.

— 1996, ‘Backtracking to Babel: The Chronology of Pama-Nyungan Expansion in Australia’, Archaeology in Oceania, vol. 31: 125-144.

— 2002a, ‘Changing Places: European and Aboriginal Styles’, in The Land is a Map: Placenames of Indigenous Origin in Australia, L. Hercus, F. Hodges and J. Simpson (eds), Pandanus Books, Canberra: 50-63.

— 2002b, ‘Linguistic Stratigraphy and Native Title: The Case of Ethnonyms’, in Linguistics and Native Title, J. Henderson and D. Nash (eds), Aboriginal Studies Press, Canberra: 261-292.

— 2004, ‘A Short Ride on a Time Machine: Linguistics, Culture History and Native Title’, in Crossing Boundaries: Cultural, Legal, Historical and Practice Issues in Native Title, S. Toussaint (ed.), Melbourne University Press, Melbourne: 34-49.

— (in press) ‘Shibbolethnonyms, Ex-Exonyms and Eco-Ethnonyms in Aboriginal Australia: The Pragmatics of Onymization and Archaism, ONOMA 41

— and Mary Laughren 2004, ‘Ngumpin-Yapa Languages’. in Australian Languages: Classification and the Comparative Method, Claire Bowern and Harold Koch (eds), Benjamins, Amsterdam: 151-178.

— and Jane Simpson 2009, ‘Moving along the Grammaticalisation Road: Locative and Allative Marking of non-finite Clauses and Secondary Predications in Australian Languages’, Paper presented at the PIONIER Workshop on Locative Case, Nijmegen, 25-26 August 2008,

Mathews, R. H. 1901, ‘Ethnological Notes on the Aboriginal Tribes of the Northern Territory’, Queensland Geographical Journal (Proceedings of the Royal Geographical Society of Australasia), vol. 16: 69-90.

Merlan, Francesca 1982, Mangarayi, Lingua Descriptive Series, no. 4, North-Holland, Amsterdam.

— 1994, A Grammar of Wardaman, A Language of the Northern Territory of Australia, Mouton de Gruyter, Berlin.

— 1998, Caging the Rainbow: Place, Politics and Aborigines in a North Australian Town, University of Hawaii Press, Honolulu.

Nash, David, 1986, Topics in Warlpiri grammar, Outstanding Dissertations in Linguistics, Garland, New York/London.

Nordlinger, Rachel 1998, A Grammar of Wambaya, Northern Territory, Pacific Linguistics C-140, Canberra.

Pensalfini, Robert 2003, A Grammar of Jingulu: An Aboriginal Language of the Northern Territory, Pacific Linguistics 536, Canberra

Rose, Deborah Bird 1990, Hidden Histories, Aboriginal Studies Press, Canberra.

— 1990/2000, Dingo Makes Us Human: Life and Land in an Aboriginal Australian Culture, Cambridge University Press.

Schultze-Berndt, Eva 2000, Simple and Complex Verbs in Jaminjung: A Study of Event Categorization in an Australian Language, Nijmegen: MPI series in Psycholinguistics 14.

— 2002, ‘Grammaticalized Restrictives on Adverbials and Secondary Predicates: Evidence from Australian Languages’, Australian Journal of Linguistics, vol. 22, no. 2: 231-264.

Tindale, Norman B. 1974, The Aboriginal Tribes of Australia: Their Terrain, Environments, Controls, Controls, Limits and Proper Names, The Australian National University Press, Canberra.

Walsh, Michael, 2002, ‘Transparency versus opacity in Aboriginal placenames’, in The land is a map: Placenames of Indigenous origin in Australia, eds. L. Hercus, F. Hodges and J. Simpson (eds), Pandanus Books, Canberra: 43-49.

1 An earlier version of this paper was presented at the Placenames workshop held at Geoscience Australia, Canberra, 2005, and thanks go to the audience there for useful comment. I thank Eva Schultze-Berndt for unpublished data on Jaminjungan and Ngarinyman. Part of the work on this paper was funded by NSF grant BCS-0902114 ‘Dynamics of Hunter-Gatherer Language Change’. Thanks also to the Australian Institute of Aboriginal and Torres Strait Islander Studies for support for work on the project.

2 The relevant language for the VRD is Wardaman, which is usually classified together with Wagiman and Yangman, outside the district (a family referred to by Evans (2003: 2) as Wardaman/Wagiman. Some authors had included these languages in the Gunwinyguan family (Evans 2003: 13).

3 In the region under discussion both NPN and PN languages have a small number of simple verbs and most verbal concepts are conveyed by complex verbs which consist of an inflecting verb and an uninflecting coverb, which are separate words. For further discussion and examples see below.

4 Merlan (1982) analyses the clausal placenames of Mangarrayi as headless relatives, and cites some parallel examples which are not conventionalised placenames, but these are not locative-marked and are quite different from the locative-marked placenames under discussion here.

5 Karranga was not recorded before it died out in the mid 20th century. It may have been close to Mudburra with heavy Wardaman influence. There is a dialect chain between Gurindji and the languages to the south and west, Kartangarurru, Wanyjirra, Nyininy and Jaru.

6 The proposal of a single Mirndi family relies heavily on shared detailed and irregular morphology. The list of lexical cognates found in both the eastern and western branches is quite small at this stage however (Harvey 2008).

7 The territory of Jiyilawung as described by Mathews’ unidentified correspondent is similar to that now ascribed to the Kajirrabeng/Kajirrawung, so it is possible that this is a name-change which occurred in the early 20th century.

8 E.g. gardaalanji ‘people of Brunette Downs area (Nordlinger ) < gardaala ‘gidgea tree’ and possibly the ethnonym Gudanji < guda ‘rock’. This suffix appears to be complex, with the final element being the masculine suffix –ji as this alternates with feminine –nya; the ‘dweller’ morpheme would then be –iny- probably (Nordlinger does not differentiate nj and ny).

9 A complication is the fact that the north-eastern dialect of Mudburra is known as Kuwirrinyji (Kuwarrangu is also identified as a Jingulu dialect influenced by Mudburra: Pensalfini 2003: 6-7). This may point to a scenario where Wardaman called this Mudburra dialect ‘southern’ and the Mudburra in turn called their Gurindji neighbours ‘southern’ (although they are actually more east according to current positions).

10 See also Walsh (2002) for discussion.

11 In travel narratives it is possible to find sequences like ‘we camped, X, we travelled downstream, Y’, where X and Y are placenames without case-marking. As flagged by the comma, however, this is more like a separate presentational phrase, ‘we camped, that was X’, etc. Unlike in the languages with zero locative it is possible to incorporate the placename in a clause with a verb and mark it overtly locative.

12 There is a possible exception to this generalisation in the Malngin area to the west of Gurindji, where there is a site Ngalja-lu ‘Frog dreaming’. Ngalja is ‘frog’ and –lu is the western equivalent of –rni. Locative on placenames is very rare in the western dialects.

13 There are also remnants of an older class-prefix system in a number of Mirndi languages, both Jaminjungan and Barkly, including jV- masculine, and na- feminine. These are the prefixes adduced as the source of such gender prefixes in the diffusion of much of the subsection system by McConvell (1985). For more details of the old set of prefixes see Harvey (2008). Traces of this system are detectable in placenames with ni-ya- where –ya- is a lenited version of jV-.

14 Given its distribution the root marlan may be originally Pama-Nyungan. Why it was borrowed into Jaminjungan then the prefixed form diffused back into Pama-Nyungan languages is not clear: an explanation in terms of ecological zones is not promising since the tree is found throughout the semi-desert and riverine areas.

15 The form with –pung was recorded independently by Eva Schultze-Berndt as Garnagathardbung.

16 The fact that the subject is dual and the object is reflexive is odd here. If the lice on each woman were considered as separate collectives and ‘lice’ are an inalienable part of the women, this may be explained.

17 Harvey (2002: 38) mentions –m locative suffixation in Miriwung, but only in relation to ‘land tenure names’, which are, he claims, based on plant names. He accords the locative the same headless existential relative interpretation as in his 1999 article.

18 This j/t correspondence is found in other words.

19 It is also possible to add locative to the gerund form of an inflecting verb in –u, but use of a coverb without an inflecting verb is much more common.

20 Note that this case assignment as well as the absence of locative from the object is different from Gurindji; Nash (1986: 234) also notes that younger speakers use ergative instead for subjects in transitive subordinate clauses.

21 Sources on Warlpiri do not give the –ngka allomorph for this type, unlike for the –ngkarni/-rlarni and –ngkajinta/-rlajinta types.

22 Use of inherent locative placenames in the locative function and replacement of the locative by another case marker in different locational functions – which I describe here as the zero-locative pattern – is also reported in South Australia for the PN languages Kaurna and Ngarrindjeri (Hercus et al. 1997).

Previous Next