Skip to content
BY 4.0 license Open Access Published online by De Gruyter Mouton May 18, 2022

Language naming in Indigenous Australia: a view from western Arnhem Land

Jill Vaughan ORCID logo, Ruth Singer ORCID logo and Murray Garde ORCID logo
From the journal Multilingua


Language naming systems are local ways of organising diversity, yet the language names used by linguists are sometimes incommensurable with the lived social reality of speakers. The process of assigning language names is not neutral, trivial or objective: it is a highly political process driven and shaped by understandings of group identity, similarity and difference. Closer attention to local perspectives on language naming offers important insights into ideologies around social and linguistic differentiation. This paper draws together accounts of diverse language naming practices from across Indigenous Australia and applies a close lens to the region of western Arnhem Land. Through an examination of three groups (speakers of Bininj Kunwok, Mawng, and Burarra), we describe the range of strategies speakers use to divide up their local language ecologies, practices for naming lects, and the role of variation in the processes of differentiation. Naming practices between these groups show interesting similarities but also striking differences. We further highlight the interplay between two key processes which characterise local language naming strategies in the region: the ‘erasure’ of difference, typically from the perspective of a politically more powerful group, and the intentional creation of linguistic differentiation, or ausbau.

1 Introduction: why language naming matters

Language names and borders as conceived of by linguists are sometimes incommensurable with the lived social reality of speakers. The process of drawing such borders and assigning language names is not neutral, trivial or objective: it is a highly political process driven and shaped by understandings of group identity, similarity and difference (Ionnàccaro and Dell’Aquila 2001; Léglise and Migge 2006). Typically, a language gains wider public acknowledgement through ‘artefactualisation’ as a dictionary or grammar, the ‘birth certificates’ of a language (Blommaert 2008). However, the linguist’s conception of a discrete language, often an artefact of the language documentation process (a ‘doculect’ (Cysouw and Good 2013)), may not reflect the ways in which speakers divide up their own local language ecology. This ‘doculect’ may be just one of a number of closely related varieties that, through chance or circumstance, happens to have been documented while others have not.

Australia is no exception, with linguists working on Australian languages usually classifying and naming language and codes in a community (or relying on someone else’s classification), and then getting on with whatever else it is that they want to research. It is easy to forget or fail to realise, however, that the people who identify as speakers of Australian Indigenous languages often name and classify codes in very different ways to linguists. This paper focuses on western Arnhem Land, a region of small-scale multilingualism where many small languages are spoken within small communities (Evans 2010; Singer 2018b; Vaughan 2022; Vaughan and Singer 2018). Understanding local perspectives on language naming offers important insights into ideologies and practices that accompany processes of social and linguistic differentiation in small-scale multilingual contexts. We build an understanding of the diverse naming practices in western Arnhem Land through an examination of three groups: speakers of Bininj Kunwok in all its varieties, Mawng speakers at Warruwi community, and speakers of Burarra at Maningrida. We describe the range of strategies speakers use to divide up their local language ecologies, practices for naming lects, and the role of variation in the processes of differentiation and connection. Naming practices between these groups show some interesting similarities, but also striking differences.

In each case, an interplay between two key processes can be observed. First is the ‘erasure’ of difference, typically from the perspective of a politically more powerful group. Irvine and Gal describe erasure of this kind as “the process in which ideology, in simplifying the sociolinguistic field, renders some persons or activities (or sociolinguistic phenomena) invisible” (2000: 38). Erasure is just one of a number of processes which linguistic anthropologists have identified as the means by which nation states use language to consolidate and naturalise their position of power and sociolinguistic distinctiveness from other nations (Garde 2008). The second process is referred to as ‘ausbau’, following Kloss’ distinction between Ausbau and Abstand languages (1967). Garde (2008) describes how through the process of ausbau, ‘intentional linguistic differentiation’ is used to shore up political differences between clans who speak different varieties. These processes of erasure and ausbau, often assumed to be associated only with nation-states, are also relevant to the linguistic anthropology of the much smaller social groups of Arnhem Land. We consider the extent to which these processes account for naming practices in western Arnhem Land.

Language naming systems are local ways of organising diversity. The study of language naming is at the very heart of the concerns of sociolinguistics and linguistic anthropology; linking the social, political, geographic and linguistic. It has a bearing on how people see themselves and others, and how they construct ‘same’ and ‘different’ (Myers 2002). In the contemporary world where diversity is seen by states and institutions as a problem – a “logistical headache” (Migge and Léglise 2013: 1) – studying local perspectives can open our eyes to other ways of understanding and valuing diversity, bringing researchers face-to-face with our own erroneous assumptions. In this paper, we highlight the value of centring speakers’ conceptions of differentiation within the language ecology and the need to attend to the views of all relevant social actors to grasp locally salient sociolinguistic processes (Makoni and Pennycook 2005; Migge and Léglise 2013; Pennycook and Makoni 2019).

1.1 Locals versus linguists?: conflicting motivations in naming practices

In the process of language documentation, linguists need to decide on the name they will use for the language or ‘doculect’ they are focusing on. Sometimes this is straightforward, but very often it is not. The processes linguists and other (community outsider) stakeholders undertake when ‘naming’ a variety may be subject to social and political pressures very different to those acting on local community naming practices. Typically, linguists are guided by a universal approach based on structural differences (i.e. phonology, syntax, lexicon) and degrees of mutual intelligibility between varieties.

Haspelmath (2017) outlines a set of principles for linguists making decisions about naming varieties which drive towards standardisation and uniformity. In doing so he reveals some significant divergences from local community naming practices observed in Australia and around the world. These include his recommendation that each language should have a unique name, and that while sometimes multiple names are in concurrent use for one variety “this cannot be a general solution” (2017: 83). As we will see, it is frequently the case that one variety is referred to locally with various epithets due to politically motivated social processes (e.g., Sutton 1979). We will see examples in Sections 24 of the significant impact linguists and other stakeholders can have on community naming practices, and why their actions ought not to be treated lightly.

Local approaches to language naming and dividing up the linguistic space, on the other hand, often adopt a maximalist perspective on linguistic differentiation. In other words, local models of linguistic boundaries may describe larger numbers of language varieties than the linguist’s dialectology (Garde 2008: 147). This is certainly the case in Indigenous Australia (see, e.g., Miller 1971; Sutton 1978; Wilkinson 1991). Walsh (1997) recognises this in his distinction between what he terms ‘language1a’ for ‘geographical dialects’ and ‘language1b’ for other named varieties or registers within some particular language1a (while ‘language2’ refers to the linguist’s approach based on mutual intelligibility). He observes that Indigenous perspectives on linguistic diversity can be primarily about social groupings and affiliations with the land. These are then extended to describe diversity in linguistic space. Linguists, on the other hand, may start from linguistic features and work upwards from there to make connections with social groups and, through them, to the land.

This is not to say that local perspectives always tend towards the maximalist, however, as speakers may want to either maximise or minimise differences between varieties depending on the context of a conversation (Lüpke 2018). In many communities, this is possible due to inherent flexibility in language naming processes – a range of levels of ‘granularity’ may be available for sociolinguistic group identification (e.g. Garde 2013). This flexibility can be altered or undermined, however, through interaction with linguists’ endeavours or other political processes (e.g., strategic essentialism deployed as a strategy in claims for land rights (Rumsey 1989)).

1.2 Language in the Australian context

A rich literature addresses how languages, dialects, varieties, special signed and spoken registers and other ways of talking can be named in Australia (e.g., Sutton 1979; Walsh 1997; Wood 2016). This literature reveals great diversity in language naming and classification processes across the region. By way of background, different systems of language naming from around Australia are discussed in Section 1.2.1 below. Within the diversity of ways of conceiving of languages, there are two common tendencies that seem to be widespread around Australia. The first is the distinction between ‘owning’ a language and ‘speaking’ a language (Merlan 1981; Rumsey 1989; Sutton 1978, 1997). A person might well speak a language they do not ‘own’, and equally might ‘own’ a language they do not speak (or do not speak fluently). For example, in a language survey carried out by the Warruwi School on South Goulburn Island in the early 1990s, participants were asked to write down their languages. Most wrote down a single language. This is rather surprising given that Warruwi Community, as discussed in Section 3 below, is highly multilingual and most adults speak two to eight Indigenous languages. In the survey, however, people reported their language ownership rather than all the languages they spoke. As Biddle (2012) notes, this is common in language surveys in the Australian context. The language that an Indigenous person owns in Arnhem Land is the language of their patrilineal clan, inherited via their father’s father. Thus the Warruwi survey had not, as intended, collected information on language use in the community, but rather on an aspect of participants’ social and linguistic identity.

The second common tendency found across many communities is the way that language ownership, land ownership and clan membership intersect. Merlan (1981), reviewing relevant literature from around Australia and drawing on her own research in the Roper River region, reconstructs a traditional conceptualisation in which languages are associated primarily with an area of land, and only secondarily with a social group, as shown in Figure 1. It seems likely that this formulation predates White contact, and as Merlan notes there is no “neat line [that] can be drawn anywhere in Australia between ‘traditional’ and ‘post-contact’ Aboriginality” (1981: 133).

Figure 1: 
The indirect relationships between individuals and ‘their’ languages (based on Evans 2003b; Merlan 1981).

Figure 1:

The indirect relationships between individuals and ‘their’ languages (based on Evans 2003b; Merlan 1981).

Although this conception of language identity is referred to in the literature as ‘language ownership’ because of the authority it bestows on the ‘owner’, it is not meant to imply that a person or clan owns a language directly. Rather, the clan owns an area of land, a clan estate, which has some language that belongs to it. Therefore that clan has authority over that language and responsibility for it (Figure 1). The idea that the language belongs to the land, rather than the clan directly, is supported by creation myths from quite disparate areas in which creation spirits ‘planted’ languages in the land during the Dreamtime. In doing so, they created an association between specific languages and specific areas of land (see Evans 2010; Merlan 1981; Sutton 1997).

1.2.1 Language naming in Australia

The literature on language naming in Australia attests to the confusion experienced by researchers trying to analyse data on language names recorded from speakers. The fact that researchers for the longest time were searching for ‘tribes’ which shared a common language – based on assumptions about traditional life informed by linguistic nationalism (Irvine and Gal 2000) – made the study of language names more difficult as these tribes did not exist. An early tendency among researchers was to erroneously assume that Indigenous Australians were organised into groups of around 500 who spoke a single language (Merlan 1981; Sutton 1991). In fact, Indigenous people tended to move around in smaller groups which were multilingual or multi-dialectal, often both. The primary unit of social organisation in much of Australia was the small patrilineal clan. Groups of people who lived together were made up of people from multiple clans, because people married members of other clans (Sutton 2003).

Walsh (1997) provides evidence that in pre-contact times there were often very fine-grained systems for naming linguistic varieties, right down to varieties spoken by 30 speakers, or even fewer. Linguists have tended to focus on broader measures in drawing boundaries between varieties, focusing on the level at which typological profiles diverge and lexical differentiation makes mutual intelligibility difficult. Local naming patterns in Australia, though, tend to take a fine-grained view of variation, often reflecting ideology as much as practice. It is likely, however, that in some cases what have been treated as ‘language names’ have in fact been more like momentary or ‘picturesque’ descriptions of a variety rather than proper-name labels. Although linguistic and dialectal differences and similarities are frequently recognised, this does not imply that there were always uniform labels for distinct speech forms. It is possible that some, or even many, varieties had no name at all (Sutton 1979: 90). Early researchers often recorded fine-grained terms, but no overarching term that would group all the varieties perceived by linguists as similar enough to be categorised as a single language. The names that missionaries, linguists and anthropologists used to refer to what they perceived as a single ‘language’ have been taken up in many areas. This is the case for the language name Murrinhpatha, spoken in Wadeye (NT), which Walsh (1997) discusses as a set of languages earlier referred to by speakers using more fine-grained terms.

Systems of language naming in Indigenous Australia are quite variable. In some regions, such as eastern Arnhem Land and Cape York Peninsula (Figure 2), linguistic ideologies exist that designate a unique linguistic variety to each clan, or land-owning descent group (e.g., Schebeck 1968; Sutton 1978; Wilkinson 1991). At Cape Keerweer (western Cape York), for example, “all clans have at least slightly different dialects” (Sutton 1978: 183) or ‘clanlects’. In practice, clanlects perceived to be similar in this region are often referred to by the same name (1978: 180), although from the linguist’s perspective the named varieties are in some cases very similar. In the Western Desert in the centre of Australia, on the other hand, the notion of the patrilineal clan is not clearly applicable. “[D]ialect variations … [have] no direct territorial significance” (Berndt 1959: 93) but are instead more related to kin connections.

Figure 2: 
Key regions and locations in Section 1.2.1 (map: Jill Vaughan).

Figure 2:

Key regions and locations in Section 1.2.1 (map: Jill Vaughan).

Across Australia, language names are derived from a variety of sources. Very commonly, languages are named for properties of their speakers rather than properties of language use (Walsh 1997). In some regions, such as among the Yolŋu Matha – speaking clans of eastern Arnhem Land, each clanlect simply shares the name of the clan that its speakers belong to. While alternate sign languages are ubiquitous, they are usually referred to just by the word meaning sign (v.) or sign language (Green and Jorgensen forthcoming). Language varieties may also be tied to landscape features, such as Kuku Yalunyu, an epithet used in various ways to refer to the varieties spoken by people associated with the same river in Cape York (Wood 2016). Another widespread naming strategy draws on ‘shibboleth’ lexical forms which distinguish varieties from each other. For example, the varieties referred to by the term ‘Western Desert language’ are quite similar in structure but often very different in vocabulary. When Miller (1971) investigated linguistic variation at Warburton mission, one way that speakers classified the varieties was through reference to the demonstrative ‘this’. So the western variety associated with the mission is referred to as Ngaanya-tjarra (this-HAVING) while the variety to the east is known as Ngaatja-tjara (this-HAVING). This is similar to naming practices in north-east Arnhem Land; demonstrative terms are also used to distinguish varieties at a higher level than clanlects (Morphy 1977).

Although language names and their meanings are sometimes quite clear-cut and fixed, they may also be context-bound or emergent. Language labels may be ‘shifters’ (in Jakobson’s (1957) sense), as the meaning may change depending on who uses the term and where they are located (e.g., in the landscape, within a social network, in historical time). Miller considered language naming in the Western Desert to be largely contextual:

The primary purpose of the dialect names is, I would suggest, a means of identifying ‘us’ versus ‘them’. […] It is useful to have more than one set of contrasts, so that one can take a narrow or broad view, depending on which the situation calls for. (Miller 1971: 73)

Multiple systems may also be in operation, even if they are apparently conflicting because the meaning of the names is tied to the context of their use. There is no reason to assume that in the pre-contact era everything was neat and orderly and that the more contextual systems represent some kind of decay. In many cases greater ‘orderliness’ seems to have developed through the pressures of interaction with White society (e.g., Walsh 1997).

1.2.2 Western Arnhem Land: some background

Western Arnhem Land, and northern Australia more generally, has long been recognised as a site of intensive linguistic diversity (e.g. Lüpke 2016). Figure 3 shows how a single language family, Pama-Nyungan (represented in white), covers most of Australia.

Figure 3: 
Map of Australian language families showing how family-level diversity is concentrated in the north of the continent (Evans 2005).

Figure 3:

Map of Australian language families showing how family-level diversity is concentrated in the north of the continent (Evans 2005).

Figure 4: 
Map of the languages of Arnhem Land (compiled by Mark Harvey).

Figure 4:

Map of the languages of Arnhem Land (compiled by Mark Harvey).

The map Figure 3 shows how all the other language families are found in a fairly small area in the north of Australia. In western Arnhem Land, languages of three different language families (Iwaidjan, Gunwinyguan and Maningridan) are still spoken. Western Arnhem Land is used here to refer to an area of cultural continuity. There are long-standing cultural and linguistic differences between eastern and western Arnhem Land which are frequently noted by researchers and Arnhem Landers themselves (Elwell 1982; Singer and Harris 2016). Arnhem Land was set aside as an Indigenous reserve in the 1930s and has been recognised as Indigenous-owned land since the 1970s. It is not possible for outsiders to access the area without a permit, and the region is only accessible by sea and by air for much of the year due to seasonal flooding. Arnhem Land was never successfully settled by farmers so Indigenous groups have been able to remain close to their ancestral lands.[1] Earlier, people were nomadic and moved around in small bands, and now they live in settlements known as remote communities, most former missions.

Geographically the area we refer to as western Arnhem Land is bounded by the East Alligator River to the west and the Arafura Sea to the north (see Figure 4). The southern boundary is formed by an area with few settlements, between the Arnhem Land plateau and remote communities such as Weemol and Bulman in south-eastern Arnhem Land. The eastern boundary is less clear, but we include the remote community of Maningrida as part of western Arnhem Land. Geographically, it is effectively a meeting place between east and west. There is also significant interaction between western and eastern Arnhem Land at Warruwi Community where many eastern Arnhem Landers have settled.

In western Arnhem Land, there are many small languages associated with coastal areas, such as Mawng. The larger dialect assemblage referred to by linguists as Bininj Kunwok is associated with the area inland on the rocky Arnhem Land plateau. Burarra, also a relatively large language, is associated with the coastal plains east and west of the Blyth River in north central Arnhem Land. While all the languages of the area are small by world standards, we can distinguish between the larger languages that have a thousand or more speakers such as Burarra and Bininj Kunwok, and the smaller languages that have several hundred speakers such as Mawng. It is important to remember that what linguists treat as larger languages, such as Bininj Kunwok, tend to be viewed at a finer grain by speakers, such as at the level of geographic dialect or clanlect, as we will see.

This article compares three cases studies of language naming in western Arnhem Land through focusing on three language groups, one from each of the main language families in the region; Bininj Kunwok (Gunwinyguan), Mawng (Iwaidjan) and Burarra (Maningridan). These studies investigate the lectal landscapes of these groups, in particular how speakers themselves view the diversity in their languages. Among speakers of the Bininj Kunwok dialect group, there is an ideology that every clan speaks slightly differently. Section 2 considers not only clanlects of Bininj Kunwok but also higher-level groupings identified by speakers, that linguists have referred to as ‘dialects’. In Section 3, we look at the Mawng language, a small language where there is an ideology of uniformity amongst speakers. They view the Mawng language as spoken today as pretty much the same across the 400 speakers of the language. However, they recall specific varieties of Mawng that are said to have been spoken in the past. Finally, we consider the Burarra language, where speakers report a number of different dialects associated with specific areas of land. These have been recorded by previous researchers in quite different ways, and in Section 4 we track published artefacts of language naming practices to highlight tensions between local and scholarly perspectives.

The discussions in this article draw on long-term collaborative research by the three authors with three different groups in western Arnhem Land. Each author has carried out extensive language documentation and ethnographic research within a research partnership in which community members are co-researchers (cf. Di Carlo and Good 2020). Ethnographic research has included formal interviews and informal discussions on the topic of dialect groupings, classifications of varieties and perceptions of variation. The authors have also constructed and analysed linguistic corpora using both conversational data and elicited data. Differences in the ways that linguistic varieties are classified by earlier linguists are also compared. This paper takes a linguistic anthropological approach focusing on language ideologies. We are less concerned with actual language practices and more concerned with differing perspectives on classification within Indigenous communities, among linguists and between these two groups.[2]

2 Situated perspectives on variation: ausbau and erasure in the Bininj Kunwok varieties

A chain of mutually intelligible varieties is associated with the higher ‘stone country’ of western Arnhem Land, located inland away from the coast (Garde 2013). These are referred to collectively by linguists as Bininj Kunwok or ‘people’s language’ (bininj: ‘person, Aboriginal’, kunwok: ‘speech, language’). The term bininj is also used to refer to the groups whose language(s) use the word bininj for ‘person’ so Bininj Kunwok can also be understood as ‘the language of the bininj’, which conveys a sense of language ownership. Whilst there is community acceptance of the use of this language name in certain contexts (e.g., language maintenance programs, linguistics publications), it is not used outside these contexts by speakers of Bininj Kunwok.

Speakers use a wide range of terms and classify the variation in Bininj Kunwok right down to the clanlect. Each clan also has a unique language or kundangwok (literally ‘IV[3]-mouth-language’). Speakers describe the distinctiveness of each kundangwok with reference to a species name, compassionate interjection or verb prefix (this process of highly localised ausbau is discussed in detail in Garde 2008). Evans’ (2003a) grammar of Bininj Kunwok classifies the rampant variation in Bininj Kunwok into six geographical ‘dialects’ that divide up the country owned by Bininj Kunwok speakers and their linguistic variation (Figure 5). However, speakers of each Bininj Kunwok dialect classify the other dialects from their own perspective. Sometimes they use terms shared with speakers of other dialects, but many are distinctive. Table 1 sets out these dialect-centric glossonyms. The column on the left is the referring group and varieties referred to are in subsequent columns.

Figure 5: 
Map of Bininj Kunwok dialects. Used with the permission of Bininj Kunwok Regional Language Centre (design: Christine Bruderlin).

Figure 5:

Map of Bininj Kunwok dialects. Used with the permission of Bininj Kunwok Regional Language Centre (design: Christine Bruderlin).

Table 1:

Mid-level or ‘dialect’ terms used by speakers of Bininj Kunwok varieties.

Kundjeyhmi Kunwinjku Kuninjku Mayali (Bolkdjam) Kune Narayek Kundedjnjenghmi Mayali
  1. Kundjeyhmi

  2. an-rayek

  3. woh arri-yime

  4. Mayali

Kunwinjku Bininj Bulkay ? ? Kundedjnjenghmi

Kunwinjku Kundjeyhmi Kunwinjku Kunwinjku (Kunrayek) Kunwinjku (Kunrayek) Kunwinjku (Kunrayek) Kundedjnjenghmi/Kunwinjku Kunwinjku
Kuninjku Kundjeyhmi
  1. Kunrinjku

  2. Kun-kerlk

  3. burrkyak

  1. Kuninjku

  2. Kunrayek

  3. kayakki

Kunrayek kurruh Kun-rayek


Mayali (Bolkdjam) Kundjeyhmi

  1. Kundangburddji-kaberrk

  2. Berreboyen

  3. Kundangyohmi

  4. Kunkerlk

  1. Kuninjku

  2. Nakerlk

  3. Buboyen

  1. Mayali

  2. Kune Dulerayek

  3. Buboyen

Kune Dulekerlk


Kune Narayek Kundjeyhmi

  1. Kundangburddji-kaberrk

  2. Berreboyen

  3. Kundangyohmi

  4. Kunkerlk

  1. Kuninjku

  2. Nakerlk

  3. Buboyen

  1. Kune Nakerlk

  2. Kune Kunkerlk

  3. Buboyen

  1. Kune Narayek

  2. Kune Kunrayek

  3. Buboyen


Kundedjnjenghmi Kundjeyhmi

  1. Kunwinjku

  2. Kunkerlk

  3. Manu

Kuninjku Kune Kune Dedjnjengh

Mayali Mayali Kundjeyhmi
  1. Kunwinjku

  2. Kundangyohmi

  3. Kabarridangyohme

Kuninjku Kune Kune Mayali


The dialect with the largest number of speakers in the chain is Kunwinjku, spoken in the north-west of Bininj Kunwok country. It has grown as a result of becoming the main language of Gunbalanya community (Oenpelli), a community where many people trace their ancestry back to other Bininj Kunwok dialects, and languages of other families. Speakers of the Kunwinjku variety apply the name of their own language as a collective name for all dialects, i.e. the whole Bininj Kunwok chain. From the majority Kunwinjku perspective, almost everyone speaks Kunwinjku. Author Garde once pointed out to a Kunwinjku speaker that people further to the east had their own names for their varieties such as Kune, Kuninjku or Kundedjnjenghmi. The Kunwinjku man admitted that may well be what those people say, ‘but really they all speak Kunwinjku’— a view that effectively erases the linguistic identity of the smaller dialects. Speakers of these minority dialects however all use the higher order glossonym ‘Mayali’ (or Kun-mayali) which also means ‘thought, idea, concept’. Kunwinjku speakers largely reject this term as a higher order collective term in favour of the name of their own variety. Recent research on how Bininj Kunwok is spoken (Marley 2020), finds that speakers draw on variants associated with more than one dialect in constructing their multifaceted identities. This may represent recent developments as the communities where children grow up are now just as significant as the geographical region of their clan estate.

3 ‘We all speak one Mawng now’: memories of lost variation

This section explores contemporary perspectives on variation in Mawng. Mawng is spoken at Warruwi Community, South Goulburn Island where most children grow up speaking the language (Singer 2018b; Singer and Harris 2016). Mawng speakers see their language today as uniform and lacking in variation. However, they recall a number of distinct varieties that were spoken in the past such as Mayinjinaj Mawng, Ngurtikin Mawng and Majakurtu Mawng. These varieties are named after social groupings associated with three geographical regions that came together to form Warruwi Community. Talking about the varieties of Mawng with Elders brings to the fore an awareness of the language knowledge that has been lost. This sense of loss exists alongside strong community pride in the Mawng language. Mawng is one of only two languages of the Iwaidjan language family that are still spoken, out of an original seven (Evans 2000; Mailhammer and Harvey 2018). Mawng has around 400 speakers and has probably always been a small language spoken by multilinguals in a highly multilingual region (Singer 2018b). Adults who identify as Mawng speakers also speak English and often also speak the Kunwinjku variety of Bininj Kunwok, a variety of Yolŋu Matha and other languages such as Burarra.

3.1 Mysterious Manangkardi

In discussions of language at Warruwi, Elders often bring up the Manangkardi language and describe it as the original language of Warruwi. The language is alternatively described as a kind of Mawng or as a separate language to Mawng. In the following quote, Elder Nita Garidjalalug, describes how both Manangkardi and the variety of Mawng originally spoken at Warruwi have been lost (1).

Malalkukuj pata tuwuran kanipa. Malalkukuj (island people) are the people who are from here.
La awunginkangung Manangkardi la Mawng. They used to speak Manangkardi and Mawng.
Ta Mawng ta arnangkat awunginka ng ung. The Mawng they spoke was ‘hard/strong’.
Kayirrk ta kawunginka ta nungmurrunti …upside down Mawng The Mawng they speak today is no good … upside down Mawng
But ngarri kapa warak … kunuka pata taka But we come from over there – what are those people called?
Mayinjinaj ngaralk tuka ta whole lot karrkinka ta Mawng, Mayinjinaj ngaralk. Mayinjinaj variety, there’s a whole lot of people who speak Mawng, the Mayinjinaj variety.
Mayinjinaj ngaralk ngarrkinka. Mayinjinaj is the variety we speak.
La ta Mawng wungijalk nungpuran kani Martpalk arlarrarr But the real Mawng from here is gone now.
A Manangkardi awunginka ng ung and Mawng wungijalk kanipa wularrut Ah, they were speaking Manangkardi and real Mawng, here, before.
Tuwarlkparrakanut la kayirrk arlarrarr - ma-nganti inginkay. The old people spoke it but now they are gone – nobody speaks it.
(RS1-361 NG cyclone interview, 28/7/2015, 00:08:40.734 – 00:09:45.593)

Nita Garidjalalug mentions both Manangkardi and Mawng wungijalk[4] ‘real/proper Mawng’ as having been lost. Her discussion conveys a sense of loss of the knowledge of language associated with Warruwi and Weyirra, the two Goulburn Islands. In her account, these were replaced by the Mayinjinaj variety of Mawng, associated with an area of the mainland to the east. This is a common view, but not uncontroversial. It may be mainly those who like Nita identify as Mayinjinaj Mawng that classify the Mawng spoken today as a kind of Mayinjinaj Mawng.

The Manangkardi language has an almost mythical presence at Warruwi because it is so often raised in discussions of the linguistic past, yet there is no trace of it now. Nita Garidjalalug, and other Elders in their sixties had family in their parents’ generation who could understand Manangkardi. They recall hearing people older than their parents speaking to their parents in Manangkardi. However, without clear records of the language, it is difficult to know how similar or different it was to Mawng. For example, in a conversation with linguist Isabel O’Keeffe, two Elders agree that Manangkardi was similar to Mawng but “a little bit different” (O’Keeffe 2016: 429). In an interview with Johnny Namayiwa, the Traditional Owner of Warruwi in (2), Evans suggests the Manangkardi language was mutually intelligible with Mawng.[5] Initially Namayiwa agrees, but then adds he could only understand a “few things”.

Nicholas Evans We were talking before about that Manangkardi language.
That’s really similar to Mawng eh?
Johnny Namayiwa Yes, yes.
Nicholas Evans Like you can understand each other …
Johnny Namayiwa Yes.
Nicholas Evans …when they speak
Johnny Namayiwa Yes.
Nicholas Evans So it’s just like a dialect of [Mawng]?
Johnny Namayiwa Oh few things I understand. Not…
Nicholas Evans Not everything?
Johnny Namayiwa That’s why I was asking about that old man – did you record him?
You didn’t?
(Nick Evans and Johnny Namayiwa, 2/11/2013, 00:01:27.134–00:01:54.678)[6]

It is not clear whether Mawng and Manangkardi were similar enough to be mutually intelligible. Johnny Namayiwa’s comments suggest that perhaps they were but he did not have enough exposure to Manangkardi to build up comprehension. At the end of the quote in (2), he asks Nicholas Evans about his work with Charlie Wardaga at Minjilang. Nicholas Evans worked with this Elder, now deceased, to make the last recordings of the other Iwaidjan languages Ilgar, Garig and Marrku, but unfortunately Evans did not get a chance to record Charlie Wardaga speaking Manangkardi.

There are no clear records of Manangkardi or specific memories of parts of the language among the descendants of speakers. Elder David Manmurulu recalled that an elderly relative who his father George Winunguj cared for spoke Manangkardi to his father. David Manmurulu then recalled a few expressions, but was not completely sure whether they were from Manangkardi or the Ngurtikin variety of Mawng, discussed in the following section.[7] Manangkardi is mentioned in Mawng speaker Lazarus Lamilami’s book Lamilami speaks (Lamilami 1974). Lamilami lists the languages spoken by the other children at Warruwi school when he was a student there (around 1916–1920).[8] He writes “Some of the children spoke Managari,[9] the language of North Goulburn Island” (Lamilami 1974: 95). Lamilami also mentions that students spoke the Maningridan language Ndjébbana, the Iwaidjan language Iwaidja, the Kunwinjku variety of Bininj Kunwok and the Gupapuyngu variety of Yolŋu Matha.

Lamilami described Manangkardi as ‘the language of North Goulburn Island’. There are different views of exactly what area of land the Manangkardi language is associated with. Elders interviewed at Warruwi in the past two decades associate the language with both Warruwi and Weyirra, the two Goulburn Islands. However, Elders who worked with linguist Nicholas Evans at Minjilang (Croker Island) associated Manangkardi with Weyirra (North Goulburn Island) and a part of the mainland coast to the west (Evans 2010; Mailhammer 2021). This is reflected by the placement of the language name Manangkardi in the sea between these two areas in Figure 4 and Figure 7 (based on Evans (2000)). Approaching these differences, remember that multiple perspectives on classification are the norm among speakers of Indigenous Australian languages. Thus accounts of their anchoring in space are expected to vary according to group membership and over time.

3.2 Classifying past varieties of Mawng

In addition to remembering Manangkardi, Elders such as Nita Garidjalalug recall that distinct varieties of Mawng were spoken in the past. Some Mawng speakers categorise Manangkardi as one of these varieties while others classify it as a separate language. This may reflect the fact that different groups always had different perspectives on the relationship between varieties of Mawng and Manangkardi. Missionary linguist Heather Hewett (nee Hinch) worked on the Mawng language at Warruwi from 1964 to 1979. In her earliest report from 1965 to 1966 she mentions four dialects of Mawng.

“Work has commenced on dialect differences in Mawng. There are four main dialects, not differing greatly – but such as they are these differences will need to be defined and the dialect chosen for translation work.” (Heather Hinch report 1965–66, Goulburn Island. Uniting Church in North Australia collection, National Archives of Australia, Darwin).

She followed up in her 1967 report with a statement that there are two dialects of Mawng. In her Mawng dictionary manuscript she labelled 24 words as belonging to one of three varieties: Ngurtikin Mawng, Mayinjinaj Mawng and Malalkukuj Mawng. The small number of dialectal variants recorded by Hewett could reflect that dialect levelling occurred before her work at Warruwi began. Alternatively there may only have ever been small lexical differences between Mawng varieties, like the relatively minor differences between kundangwok clanlects of Bininj Kunwok and some varieties of Burarra.

Interviews with Mawng speakers between 2015 and 2018 revealed similar perspectives to those of Heather Hewett. Most people named three varieties of Mawng but some only named two, while others named four. The two mainland varieties were consistently named Mayinjinaj Mawng and Ngurtikin Mawng, matching what Hewett recorded a generation or two earlier. However, most people interviewed recently used the term Majakurtu Mawng to refer to the varieties of Mawng associated with the two islands, rather than the term Malalkukuj Mawng used by Hewett. The most common way of viewing varieties of Mawng among interviewees is shown in Figure 6.

Figure 6: 
Three varieties model of Mawng – Manangkardi is seen as a separate language.

Figure 6:

Three varieties model of Mawng – Manangkardi is seen as a separate language.

The view of Mawng language varieties represented in Figure 6 is depicted geographically in Figure 7.

Figure 7: 
Areas associated with the three varieties of Mawng today (adapted from Evans 2000).

Figure 7:

Areas associated with the three varieties of Mawng today (adapted from Evans 2000).

The binomial names for varieties of Mawng take the name of a social group associated with an area of land and use it as a modifer. Majakurtu is the name of a group of allied clans (clan aggregate) associated with Warruwi and Weyirra, the two Goulburn Islands. The term Malalkukuj ‘islands’ is synonymous as it refers here to people who are associated with islands. Mayinjinaj is a clan aggregate associated with part of the mainland to the east of Warruwi, which includes both Mawng and Kunbarlang clans. Ngurtikin, in contrast, is the name of a single Mawng clan which is associated with a large area of land and has ‘eastern’ and ‘western’ subgroups (Brown 2016; Gould 2010). The name Manangkardi, in contrast, does not take the form of a binomial – it is never used as a modifier for the word ‘Mawng’.

Given that language naming has always been contextual and changes over time (cf. Wood 2016), there may be no definitive answer to the questions of what area of land the Manangkardi language was associated with pre-contact or what its relationship to Mawng was. However, perhaps Manangkardi was originally associated only with North Goulburn Island, as stated by Elders who linguist Nicholas Evans worked with, while Majakurtu, or Malalkukuj Mawng was associated with South Goulburn Island. The development of Warruwi as an important settlement is likely to have brought with it changing perspectives on the relationship between land and languages in the surrounding region.

3.3 Contemporary Mawng: an ideology of uniformity

From 1973 to 2000 a bilingual Mawng-English program ran at Warruwi School, in which classes for the younger year levels were all taught in Mawng. This raised the question of which Mawng was appropriate for use in teaching. Elder Nancy Ngalmindjalmag, who was then a teaching assistant, made a contribution to the School’s 1989 annual report in which she refers to ‘both dialects’ of Mawng, shown in (3).

“Teaching Maung is very difficult for me at first because of the children using different dialects. So we called a meeting with the old people to talk about how the children talk Maung. We made a decision that all the Maung is the same language now and so we can use both dialects of Maung. But we will need to talk about this more in 1990 to make a strong decision about how Maung should be taught at school.” (DET 1990: 66)

Like others at Warruwi, Nancy Ngalmindjalmag now holds the view that there is only one variety of Mawng. In the quote (4) below, from an interview with her and her aunt Rosemary Urabadi in 2015, they agree on this point.

Ruth Singer: And what about like now, the different families like Mayinjinaj, Ngurtikin or Majakurtu. Do you think they still speak Mawng a little bit differently?
Rosemary Urabadi: Not really
Nancy Ngalmindjalmag: Not really, all the same
Rosemary Urabadi: Still the same. They all we all speak the same.
(RS1-357 NN1 and RU1 Variation interview, 24/7/2015, 00:32:50.492–00:43:25.958)

Over numerous interviews, different Mawng Elders agreed that there is only one Mawng now. The social groups Mayinjinaj, Ngurtikin and Majakurtu are still relevant to people’s identities at Warruwi but the varieties associated with each group are located in the past. Mawng-medium classes were re-introduced into the school curriculum in 2016; Mawng literacy and maths. During this time of renewal in Mawng language teaching, variation in Mawng has not been raised as an issue. Elders are no longer concerned that their grandchildren are being taught the ‘wrong’ variety of Mawng. The ideology of a ‘single Mawng’ is now well-established in the community and Elders are more concerned that some children might not speak Mawng well enough. Community members are well aware that many other small languages such as Kunbarlang and Iwaidja, which were as widely spoken as Mawng a few generations ago, now have few speakers.

We will never know exactly how much variation there was in Mawng in the past and to what extent the variation that existed was tied to land-based social groupings. In addition, there may be more residues of past variation in contemporary Mawng than has been recorded, because little linguistic research has been done with Majakurtu people. What is clear, however, is that there are a wide range of perspectives on past variation in Mawng and the place of Manangkardi.

4 Burarra: artefactualisation and erasure in scholarly accounts

4.1 Burarra and Maningrida community

Speakers of the Burarra language live on, and own, country to the east of Maningrida, a township in north-central Arnhem Land. While Burarra is a non-Pama-Nyungan language, it has long been in contact with Pama-Nyungan languages to the east and is thought to show evidence of this. Some among the Burarra group align socially and culturally with the strongly patrilineal Yolŋu groups of the north-east, while others reflect a localised interweaving of eastern and western Arnhem orientations (Carew 2016).

As is commonplace in the region, Burarra is typically just one code within rich individual multilingual repertoires. The linguistic repertoires of Burarra people vary according to residence patterns, life histories and kinship connections, and frequently include Djinang, Bininj Kunwok varieties (typically Kuninjku and/or Kune), Gurr-goni and Ndjébbana, alongside local varieties of English, and, for some, Kriol.

Many Burarra people now reside in Maningrida community which is located on land belonging to Ndjébbana people. Maningrida was founded as a welfare settlement in the late 1950s, and in the subsequent decades employment and supplies available there attracted many groups from further afield in Arnhem Land. The Burarra in particular migrated in great numbers (Borsboom 1986; Hiatt 1965) and have since this time made up the community’s largest linguistic group, with fewer now living permanently on outstations on traditional country. With the Burarra social orbit now centred in part on Maningrida, and incorporating Darwin (the state capital, a migration destination since the Second World War), the sociolinguistic context of the language community has undergone considerable transformation. There are some 2000 speakers in total, perhaps half of whom are L1 speakers. Much of the L2 speaker community is based in Maningrida where a levelled variety of Burarra, with significant code-switching with English, functions in some settings as a kind of lingua franca (although not as a fully-fledged communilect). Indeed, Maningrida to date remains resolutely multilingual (Vaughan 2019). The evolving nature of the Burarra speaker community, alongside dialect-levelling processes in this urban context as formerly more isolated speakers interact on a regular basis, has had important consequences for the shape of the language and the language ecology among younger/town-based speakers.

4.2 Naming the Burarra varieties

While ‘Burarra’ has come to be used widely as a macro label for all varieties of the language, captured for example by a single ISO639-3 code (bvr), in fact this term masks a number of sociolinguistic complexities. In reviewing linguistic and anthropological work in the region, Carew (2016: 17) concludes that the designation likely originated as an eastern exonym for the language group (see also Borsboom 1978; Glasgow 1994; Mirritji 1976). Viewed from within, however, the sociolinguistic space may be divided in a number of ways, and attested divisions vary according to the identity of the given individual, their interlocutor and/or the particular socio-interactional goals of the communicative act in question.

Other designations for the entire language group and its varieties have been favoured at different times and by different people (both speakers and outsider stakeholders, such as linguists and anthropologists). In his regional survey of linguistic groupings, Languages of Arnhem Land, North Australia (1942), linguist Arthur Capell, treats ‘Burera’ (Burarra), ‘Gudjälavia’ (Gu-jarlabiya) and ‘Gunaidbe’ (Gun-nartpa) as three close dialects, it being “an arbitrary matter which dialect is taken as standard” (1942: 374)[10] (Figure 8). Gu-jarlabiya means something like ‘it moves along steadily’ (from gu- ‘it’ + jarlabiya ‘walk’), while Gun-nartpa is a demonstrative variant (gun- ‘thing of land/neuter noun class’ + -nartpa ‘that (in focus)’) which can function as a shibboleth identifying speakers from clans in the Cadell River region.

Figure 8: 
Attested varieties in Capell (1942).

Figure 8:

Attested varieties in Capell (1942).

These three variety labels reproduced in Capell are replicated in the accounts of several scholars (e.g., Green 1987; Gurrmanamana et al. 2002; Jones and Meehan 1978), but in varied configurations. In each case, however, Burarra is given as a superordinate category encompassing the others as sub-varieties or more minor alternative labels (Figure 9). In her sketch grammar of Burarra, Green (1987: 1) reports the existence of three earlier varieties, but points to extant dialectal differences, especially in vocabulary, between Gu-jarlabiya and Gun-nartpa (see also Armstrong 1967: 13). Green further notes some dialectal variation within Gu-jarlabiya, in the vowels in particular, but does not make any further study of this. Green also uses the term Gu-jingarliya (‘it with tongue’), and observes that, along with Gu-jarlabiya, this label is both interchangeable with Burarra to refer to the entire language group, and available as an endonym to designate the non-Gun-nartpa speakers of Burarra.

Glasgow’s Burarra-Gun-nartpa Dictionary (1994), based on decades of Bible translation and language and literacy work in Maningrida, reflects further subdivisions within the Burarra varieties (Figure 10). She distinguishes Burarra/Gu-jingarliya from Gun-nartpa, and divides the former into two further dialect groups: Gun-narta, also known by the regional tribe name An-barra; and Gun-narda, also known by the regional tribe name Martay. Gun-narta, Gun-narda and Gun-nartpa are regional variants of the same demonstrative, and can function as shibboleth terms. They are socio-territorial identity markers for varieties spoken, respectively, in the coastal region west of the Blyth River, to the east of the Blyth River, and further south in the Cadell River region. An-barra derives from barra ‘base/bottom’, perhaps referring to associated coastal sites that are under water (Carew 2016: 60; Gurrmanamana et al. 2002), while Martay refers to the stringybark flower which blooms in the dry season, an emblem of the Marrangu clan cluster with which this group is affiliated (Elliott 1991). Glasgow describes the varieties as “three very close dialects […] or two, depending on perspective” (1994: 7), but notes that each group is proud to have a distinct identity. In this sense, the set of ‘shibboleth’ language labels (Gun-narta, Gun-narda and Gun-nartpa) reflect the operation of ausbau within the Burarra space, with these local naming practices drawing attention to linguistic differences (where few exist) in order to shore up social distinctions. Glasgow positions Gu-jarlabiya somewhat differently to Green (1987), as an endonym for Gun-nartpa speakers instead. This gives a limited indication of the inherent instability or flexibility which characterises the use of some of these terms.

Figure 10: 

Glasgow (1994).

Figure 10:

Two further subcategories within Gidjiŋali (Gu-jingarliya) are identified within Hiatt’s (1962) work on local organisational structures in Arnhem Land (Figure 11). Alongside An-barra and Martay, he lists Mariŋa (Maringa) and Marawuraba as “four loose common residence groupings” (1962: 280). It is unclear whether Hiatt saw these as dialect groups, and he makes no mention of linguistic distinctions. Maringa refers to groups resident at the north-east edge of Burarra territory, towards Yinangarnduwa (Cape Stewart) and Yan-nhangu country. What Marawuraba refers to is less clear. Information provided to author Vaughan by Burarra speakers in 2016 suggest that it may refer to a set of residence groups within the broader Martay area (see also Armstrong 1967). These groupings are reproduced in Hamilton (1981: 3) and Carew (2016: 16–17), while Elwell (1977: 25) lists all but Marawuraba.

Figure 11: 

Hiatt (1962, 1965.

Figure 11:

In Hiatt’s doctoral thesis, he distinguishes the Gunadba (Gun-nartpa) as a separate “community” (Hiatt 1965: 24). Reporting on her work on child-rearing among the An-barra, Hamilton (1981) replicates a similar schema to Hiatt’s (1962, 1965 (Figure 12), but notes that by that time younger people were including the Gun-nartpa within the broader Burarra category, and that ‘Gidjingali’ was rarely heard in everyday use. Her comments point to a blurring of certain pre-existing localised alignments and identities, but she notes that certain distinctions (such as coastal vs. inland) remained highly salient (Hamilton 1981: 3–4). Carew’s (2016) view is that while Gidjingali was used in anthropological work as a relatively stable label for the An-barra group, it is instead a more socially neutral term that may be used flexibly for various Burarra dialects, including Gun-nartpa (2016: 17–18). Carew provides a useful insight into distinct ways of dividing up the Burarra linguistic space, as her work focuses on the local practices of Gun-nartpa people. She also notes that the terms gun-nyarlkuch (soft speech) and gun-derta (hard speech) are used as secondary terms to distinguish between the Burarra lects, with Gun-nartpa speakers claiming their own variety as gun-nyarlkuch (Carew 2016: 140–1). Recent fieldwork conducted by author Vaughan in Maningrida has found these terms still in limited use, but not in a strictly consistent way. It is difficult to ascertain whether the terms have always been inherently flexible in their deployment, or whether their earlier indexes have weakened in contemporary use.

Figure 12: 

Hamilton (1981).

Figure 12:

Although these labels are understood to refer to linguistic categories, in contemporary usage at least these varieties do not exhibit a significant amount of inter-systemic variation. Lectal variation does exist within most subsystems of the language, especially in the lexicon but also within the morphology, syntax and phonology (although this is more limited and less socially salient), but the social indexicalities of variants can differ between speakers (see Vaughan 2018). Contemporary speakers are able to identify some distinct lexical variants associated with the An-barra, Martay and Gun-nartpa categories especially, although differing alignments of variants with variety labels suggest low levels of ‘linguistic coherence’ (Guy and Hinskens 2016) among the Burarra ‘dialects’. It is opaque to a contemporary analysis whether the indexicalities (and perhaps also isoglosses) of these variables have eroded to some extent over time, or whether instead it is the case that these ‘dialects’ were always social constructs more so than linguistic ones: “notional and ideological reference points” with real social dimensions but not “fully reified as full-blown linguistic systems”, to borrow from Lüpke’s (2015: 5) description of the Casamance region (Senegal).

The variations in Burarra language naming practices documented in each of these sources point to the existence of several co-existing and interacting systems that do not necessarily fit within a neatly hierarchical model. Donaldson (1984) identifies a similar phenomenon in how naming schemas have been represented amongst the Ngiyampaa and their neighbours in central western New South Wales. These variations also reflect the diverse social, cultural and epistemological positions of the scholars and the Burarra speakers they have worked alongside. Sometimes these positionalities are evident – for example Hamilton (1991) worked with An-barra people, while Carew (2016) worked predominantly with the Gun-nartpa group – but very often they are not, and indeed the scholars themselves may not realise how situated their view of the sociolinguistic space is. As Irvine and Gal (2000: 36) note, “there is no ‘view from nowhere’, no gaze that is not positioned”. In delineating language and dialect boundaries, different forms of evidence may be attended to. Researchers and speakers alike are sensitive to distinct networks of knowledge and ideologies in attesting language boundaries – likely navigating conflicting motivations and types of evidence. An obvious example when it comes to language naming and boundaries concerns the frequently encountered conflict between local practices in dividing up the linguistic space, which may in fact reflect social rather than linguistic distinctions, and what constitutes a distinct variety in strictly linguistic terms. The separation of Gun-nartpa as a distinct language from Gu-jingarliya (as in Hiatt 1962; Carew 2016) is an example of local orientations taking precedence over solely linguistic evidence. We may expect that the naming schemas attested reflect a process of negotiation and (often) necessary omission, erasure or simplification in order to represent complex ongoing social processes. These processes are often invisible in the final text.

It is interesting to note the two-way exchange which can characterise codifying language naming practices. Academic scholarship (and other institutions which wield symbolic capital) may first represent observed practices in some way, and then influence them in turn, perhaps reifying some term, or some local means of locating varieties with regards to each other, in a lasting way (see, e.g., Amery 1993 and the case of Dhuwaya in Yirrkala). This may well have been the case at various points within the Burarra space, although it is difficult to be certain. Some likely scenarios are that the ‘artefactualisation’ (Blommaert 2008) of Burarra in the Glasgow (1994) dictionary – a well-known text frequently used in the community – may have further promoted and stabilised the use of ‘Burarra’ (e.g. rather than ‘Gu-jingaliya’). Furthermore, the languages through which Maningrida school’s bilingual program ran until 2008 were labelled as Ndjébbana and Burarra – with ‘Burarra’ typically understood to encompass all dialects, including Gun-partpa.[11] Carew (2016: 17) also points to the adoption of ‘Gidjingali’ to refer to the coastal An-barra group within anthropological work in the region (e.g. Hiatt 1962, 1965; Meehan 1982), a choice that may have exerted some influence on local naming practices.

Tracking academic artefacts of language naming practices over time gives an (albeit limited) insight into a complex and evolving social process, and allows ongoing research to be situated within a diachronic perspective. In the Burarra case, we can observe certain facets of the changing fortunes of different variety labels, and gain a sense of what might have influenced and motivated the shifts. We see, for example, some reflection of the impact of centralisation at Maningrida, following the ‘assimilation and integration’ stage of the region’s post-contact history (Altman 1987: 4), whereby more localised practices have become blurred as a result of social changes brought about by the urban context. The case of Burarra also amply demonstrates that language naming is a highly situated political and social act: any given published schema is fundamentally shaped by the positioning of the researcher, the community perspective they intend to communicate (if any), and their audience. The lens of published representations of language naming is limited, however: as we have seen, they tend not to effectively capture inherent flexibility and flux, and can risk fossilising ongoing processes within a simplified schema.

5 Discussion and conclusions

In this paper, we have outlined several ways in which language naming provides a productive lens into the role of language in social and cultural life, as illustrated by the three case studies drawn from western Arnhem Land. Language naming tells us about how variation is dimensionalised into the geographical and social cosmos (Evans 2018), and about local strategies for dividing up the social space. The three case studies in this paper illustrate how different approaches to language naming can be, even within a relatively small part of Australia. Across all three studies, languages are owned by people and belong to areas of land due to their association with a clan, as described by Merlan (1981) for other groups in the north. However, it is not the case that each clan has its own unique way of speaking (a clanlect) in all three cases. Bininj Kunwok has clanlects that are distinguished by a shibboleth and sometimes have a name, distinct from the clan name. However two Mawng varieties are named for clan aggregates and only one, Ngurtikin Mawng, is associated with a specific clan. Burarra varieties are named for looser ‘residence groups’ or ‘communities’ which consist of people of different clans.

In the examples from Arnhem Land in Sections 24, an exploration of local language naming has illustrated how a range of levels of ‘granularity’ may be available for sociolinguistic group identification. Bininj Kunwok has named dialects and clanlects, which have been interpreted as forming a nested hierarchy. However, as Table 1 shows the number of Bininj Kunwok ‘dialects’ that are named is very much a matter of perspective, i.e. of the language identity of the namer, which belies this neat analysis. Alexandra Marley’s recent account of variation in Bininj Kunwok builds on this situated view, arguing that Bininj Kunwok speakers use linguistic variables from the ‘dialects’ described by earlier linguists in ways that index a range of affiliations, which is more amenable to a third-wave approach from sociolinguistics (Marley 2020). Tamsin Donaldson questions why we should expect Indigenous classifications to create discrete or hierarchical groupings, drawing on Ngiyampaa people’s classification of their languages in western New South Wales (Donaldson 1984). In western Arnhem Land a range of cross-cutting social groups of different scales are simultaneously available and language names are part of this assemblage of categories (Singer 2018a).

The three case studies show how language names in western Arnhem Land emerge from social life, not language form, so claims to speakership are claims of social identity. Accounts of language names that refer to linguistic features are in some sense always post hoc. The fluidity of language identities and the flexibility with which people use language names does not, however render language names or the categories they invoke meaningless. Speakers may make different distinctions depending on their aims in a specific time and context. Diversity is the bedrock of social, cultural and political life in Arnhem Land and skilled social actors index their various memberships and alliances with care. They can also both obscure and create distinctions via language. We have seen much evidence of language naming as a political act, with the twin processes of erasure – the process of simplifying the sociolinguistic field – and ausbau – the creation of intentional linguistic differentiation – characterising many naming frameworks both from community and scholarly perspectives. As Garde (2008) observes, these processes are typically described as the purview of nation states but can be seen here also in operation in small-scale, remote communities.

Linguists laboured for more than a century to relate Indigenous language names to circumscribed categories of linguistic diversity in Indigenous Australia. However, the extent to which languages diverge is not directly relevant to how people use a socially-based assemblage of language names. Migge and Léglise (2013) also found this to be the case, in their study of how the name takitaki is used in relation to creole varieties in French Guiana. Linguistic typology and historical linguistics seek to understand linguistic diversity by describing the degree of divergence between varieties. However, even very similar varieties can be useful for indexing social differentiation. Epps (2021) observes this for Indigenous Amazonia and points to the need for more studies which bring together methods from variationist sociolinguistics, dialectology and multilingualism, such as Marley’s recent work (2020) in order to understand what role divergence plays.

In the three case studies in Sections 24, we have explored the significant role of scholars and other stakeholders in language naming processes, and seen examples of the impact of delineating and ‘artefactualising’ a doculect. In the case of the Burarra varieties, this has led to a proliferation of language naming schemas. Because Indigenous language names are often fluid, flexible and not discrete or hierarchically nested, there is an indeterminacy in how language names and group names are mapped to linguistic units as perceived by linguists. It is difficult to determine whether differences in the naming schemas that linguists come up with reflect differences in their analyses, different views within the community or the changing relationships between people and places in Arnhem Land.

As noted in the introduction, Australian Indigenous people tend to distinguish varieties down to a fine-grained level of analysis, such as clanlects. Linguists tend to be ‘lumpers’, coining names for larger categories such as Bininj Kunwok, which were not previously recognised by Indigenous people. Today Indigenous Australians’ social and political networks extend over a greater area of land than ever before and names for larger categories have been taken up as terms for larger political units, for example in land claims (Rumsey 1989) or, in the case of Bininj Kunwok, for sharing resources for language support. Ultimately language naming reflects what language is understood to be, i.e. what its nature is.

The accounts in this paper show that Indigenous languages are understood as a part of Indigenous people and their social groups and as a part of tracts of land. The diversity of approaches to language naming around the world reflects a diversity of understandings of what language is. Following the ‘ontological turn’ in anthropology, linguists are asking what different groups understand the nature of language to be (Chernela 2018; Demuro and Gurney 2021). It is these different ‘philosophies of language’ that underlie language naming schemas. Like other aspects of Indigenous people’s lives in the north of Australia, understandings of what a language is are also in flux and merit close attention (Hauck 2016).

The picture that emerges from across the Australian region is one of great diversity in naming practices. Assumptions of uniformity can be traced to erroneous methods in language and anthropological documentation. Sutton (1979: 101) gives the example of linguists asking locals “what is your tribe?” and “what is your language called?” and then fitting the responses neatly into a traditional European model of one-to-one language-tribe linguistic nationalism (see also Irvine and Gal 2000). Many scholars have wrongly assumed language to be the sole or primary basis for sociality, rather than one of the many cultural indexes that are available for identity construction – one that often crosscuts key social groupings in Indigenous Australia. We have shown that language naming practices are diverse across the Australian region, but that two key characteristics define most local processes: (i) inherent fluidity and social strategising in the deployment of language names, and; (ii) strong interconnections between language, people and country.

Corresponding author: Jill Vaughan, Research Unit for Indigenous Language, University of Melbourne, Parkville, Australia, E-mail:

Funding source: Australian Research Council

Award Identifier / Grant number: FT200100834

Funding source: Australian Research Council

Award Identifier / Grant number: CE140100041

Funding source: Endangered Languages Documentation Programme

Award Identifier / Grant number: IPF0256


The authors wish to acknowledge that any cultural information provided by the various speech communities in relation to language practices in the article remains the Indigenous Cultural Intellectual Property (ICIP) of the respective speech communities. Ruth Singer would like to thank in particular those Mawng people who discussed the topics of this article including Johnny Namayiwa, David Manmurulu(✝), Rosemary Urabadi and Nancy Ngalmindjalmag. Jill Vaughan would like to thank Abigail Carter, Doreen Jinggarrabarra, Stanley Rankin, Cindy Jin-marabynana, Laurie Guraylayla, Betty Nguraba Nguraba, Rebecca Baker and Joseph Diddo.

  1. Research funding: This research was funded by the Australian Research Council (FT200100834 and CE140100041) and the Endangered Languages Documentation Programme (IPF0256).


Altman, Jon. 1987. Hunter-gatherers today: An Aboriginal economy in North Australia. Canberra: Australian Institute of Aboriginal Studies.Search in Google Scholar

Amery, Rob. 1993. An Australian koine: Dhuwaya, a variety of Yolngu Matha spoken at Yirrkala in North East Arnhem land. International Journal of the Sociology of Language 99. 45–64. in Google Scholar

Armstrong, Gowan. 1967. Social change at Maningrida. Adelaide: (Diploma of Anthropology), University of South Australia.Search in Google Scholar

Berndt, Ronald M. 1959. The concept of ‘the tribe’ in the Western Desert of Australia. Oceania 30(2). 81–107. in Google Scholar

Berndt, Catherine. 1985. Mondalmi: One of the saltwater people. In Isobel White, Diane Barwick & Betty Meehan (eds.), Fighters and singers: The lives of some Australian Aboriginal women, 19–39. Sydney: George Allen and Unwin.10.4324/9781003115632-2Search in Google Scholar

Biddle, Nicholas. 2012. Indigenous language usage. Canberra: Centre for Aboriginal Economic Policy Research Research, School of Social Sciences ANU College of Arts and Social Sciences, Australian National University.Search in Google Scholar

Blommaert, Jan. 2008. Artefactual ideologies and the textual production of African languages. Language & Communication 28(4). 291–307. in Google Scholar

Borsboom, Ad. 1978. Dreaming clusters among Marangu clans. In Lester Richard Hiatt (ed.), Australian aboriginal concepts, 106–120. Canberra: Australian Institute of Aboriginal Studies.Search in Google Scholar

Borsboom, Ad. 1986. The cultural dimension of change: An Australian example. Anthropos 81(4/6). 605–615.Search in Google Scholar

Brown, Reuben. 2016. Following footsteps: The kun-borrk/manyardi song tradition and its role in western Arnhem Land society. Sydney: Sydney University PhD.Search in Google Scholar

Capell, Arthur. 1940. The classification of languages in north and north-west Australia. Oceania 10(3). 241–272. in Google Scholar

Capell, Arthur. 1942. Languages of Arnhem Land, North Australia. Oceania 12(4). 364–392. in Google Scholar

Carew, Margaret. 2016. Gun-ngaypa Rrawa ‘My country’: Intercultural alliances in language research. Melbourne: Monash University PhD thesis.Search in Google Scholar

Chernela, Janet. 2018. Language in an ontological register: Embodied speech in the Northwest Amazon of Colombia and Brazil. Language & Communication 63. 23–32. in Google Scholar

Cysouw, Michael & Jeff Good. 2013. Languoid, doculect, and glossonym: Formalizing the Notion “language”. Language Documentation & Conservation 7. 331–359.Search in Google Scholar

Demuro, Eugenia & Laura Gurney. 2021. Languages/languaging as world-making: The ontological bases of language. Language Sciences 83. 101307. in Google Scholar

DET (Northern Territory Department of Education). 1990. Annual reports from specialist staff in bilingual programs in Northern Territory schools (1989). Darwin (Australia): Northern Territory Department of Education.Search in Google Scholar

Di Carlo, Pierpaolo & Jeff Good (eds.). 2020. African multilingualisms: Rural linguistic and cultural diversity. New York: Lexington books.Search in Google Scholar

Donaldson, Tamsin. 1984. What’s in a name? An etymological view of land, language and social identification from central western New South Wales. Aboriginal History 8(1/2). 21–44.10.22459/AH.08.2011.03Search in Google Scholar

Elliott, Craig. 1991. “Mewal is Merri’s name”: Form and ambiguity in Marrangu cosmology, North Central Arnhem Land. Canberra: Australian National University Master of Arts thesis.Search in Google Scholar

Elwell, Vanessa. 1977. Multilingualism and lingua francas among Australian Aborigines: A case study of Maningrida. Canberra: Australian National University Honours thesis.Search in Google Scholar

Elwell, Vanessa. 1982. Some social factors affecting multilingualism among Aboriginal Australians: A case study of Maningrida. International Journal of the Sociology of Language 36. 83–103. in Google Scholar

Epps, Patience. 2021. Diversifying multilingualism: Languages and lects in Amazonia. International Journal of Bilingualism 25(4). 901–920. in Google Scholar

Evans, Nicholas. 2000. Iwaidjan: A very un-Australian language family. Linguistic Typology 4. 91–142.Search in Google Scholar

Evans, Nicholas. 2003a. Bininj Gun-wok: A Pan-Dialectal Grammar of Mayali, Kunwinjku and Kune, vol. 2. Canberra: Pacific Linguistics.Search in Google Scholar

Evans, Nicholas. 2003b. Context, culture, and structuration in the languages of Australia. Annual Review of Anthropology 32. 13–40. in Google Scholar

Evans, Nicholas. 2005. Australian languages reconsidered: A review of Dixon (2002). Oceanic Linguistics 44(1). 242–286. in Google Scholar

Evans, Nicholas. 2010. Dying words: Endangered languages and what they have to tell us. Malden, MA: Wiley-Blackwell.Search in Google Scholar

Evans, Nicholas. 2018. The dynamics of language diversity. In Rajend Mesthrie & David Bradley (eds.), The dynamics of language: Plenary and focus lectures from the 20th international Congress of linguists, 12–35. Cape Town: Juta and Company (Pty) Ltd.Search in Google Scholar

Garde, Murray. 2008. Kun-dangwok: “clan lects” and Ausbau in western Arnhem land. International Journal of the Sociology of Language 191. 141–169. in Google Scholar

Garde, Murray. 2013. Culture, interaction and person reference in an Australian language. Amsterdam & Philadelphia: John Benjamins.10.1075/clu.11Search in Google Scholar

Glasgow, Kathleen. 1994. Burarra–Gun-nartpa dictionary with English finder list. Darwin: Summer Institute of Linguistics.Search in Google Scholar

Gould, Jacqueline Lisa. 2010. Being in the Black: The business of development in Northern Australia. Canberra: Australian National University PhD thesis.Search in Google Scholar

Green, Rebecca. 1987. A sketch grammar of Burarra. Honours Thesis. Canberra: Australian National University.Search in Google Scholar

Green, Jennifer & Eleanor Jorgensen. forthcoming. Same, similar or different?: Lexical overlap across Australian Indigenous signing.Search in Google Scholar

Gurrmanamana, Frank, Lester R. Hiatt & McKenzie Kim. 2002. People of the Rivermouth: The Joborr texts of Frank Gurrmanamana. Canberra: National Museum of Australia and Aboriginal Studies Press.Search in Google Scholar

Guy, Gregory R. & Frans. Hinskens. 2016. Linguistic coherence: Systems, repertoires and speech communities. Lingua 172–173. 1–9. in Google Scholar

Hamilton, Annette. 1981. Nature and nurture. Canberra: Australian Institute of Aboriginal Studies.Search in Google Scholar

Haspelmath, Martin. 2017. Some principles for language names. Language Documentation & Conservation 11. 81–93.Search in Google Scholar

Hauck, Jan David. 2016. Making language: The ideological and interactional constitution of language in an indigenous Aché community in eastern Paraguay. Los Angeles: University of California PhD thesis.Search in Google Scholar

Hiatt, Lester R. 1962. Local organization among the Australian Aborigines. Oceania 32(4). 267–286. in Google Scholar

Hiatt, Lester R. 1965. Kinship and conflict: A study of an Aboriginal community in northern Arnhem land. Canberra: ANU Press (Australian National University). (accessed 9 December 2020).Search in Google Scholar

Ionnàccaro, Gabriele & Vittorio Dell’Aquila. 2001. Mapping languages from inside: Notes on perceptual dialectology. Social & Cultural Geography 2(3). 265–280.10.1080/14649360120073851Search in Google Scholar

Irvine, Judith T. & Susan Gal. 2000. Language ideology and linguistic differentiation. In Paul V. Kroskrity (ed.), Regimes of language: Ideologies, polities and identities, 35–84. Santa Fe: School of American Research Press.Search in Google Scholar

Jakobson, Roman. 1957. Shifters, verbal categories, and the Russian verb. Cambridge, MA: Harvard University Russian Language Project.Search in Google Scholar

Jones, Rhys & Betty Meehan. 1978. Anbarra concept of colour. In Lester Richard Hiatt (ed.), Australian Aboriginal concepts, 20–29. Canberra: Australian Institute of Aboriginal Studies.Search in Google Scholar

Kloss, Heinz. 1967. Abstand-languages and Ausbau-languages. Anthropological Linguistics 9(7). 29–41.Search in Google Scholar

Lamilami, Lazarus. 1974. Lamilami speaks, the cry went up : a story of the people of Goulburn Islands, North Australia. Sydney: Ure Smith.Search in Google Scholar

Léglise, Isabelle & Bettina Migge. 2006. Language-naming practices, ideologies, and linguistic practices: Towards a comprehensive description of language varieties. Language in Society 35. 313–339. in Google Scholar

Lüpke, Friederike. 2015. Denorthernising multilingualism and multilingualism research framing language in the North: From the monolingual nation state to acknowledging (super) diversity. In Multilingualism in the global south (and beyond) [special issue]. Voices from around the World, vol. 3, 3–5.Search in Google Scholar

Lüpke, Friederike. 2016. Uncovering small-scale multilingualism. Critical Multilingualism Studies 4(2). 35–74.Search in Google Scholar

Lüpke, Friederike. 2018. Multiple choice: Language use and cultural practice in rural Casamance between convergence and divergence. In Jacqueline Knörr & Trajano Wilson Filho (eds.), Creolization and Pidginization in Contexts of postcolonial diversity: Language, culture, identity, 181–208. Leiden: Brill.Search in Google Scholar

Mailhammer, Robert. 2021. English on Croker Island: The synchronic and diachronic Dynamics of contact and variation. Berlin, Boston: De Gruyter Mouton.10.1515/9783110707854Search in Google Scholar

Mailhammer, Robert & Mark Harvey. 2018. A reconstruction of the proto-Iwaidjan phoneme system. Australian Journal of Linguistics 38(3). 329–359. in Google Scholar

Makoni, Sinfree & Alastair Pennycook. 2005. Disinventing and (re) constituting languages. Critical Inquiry in Language Studies: An International Journal 2(3). 137–156. in Google Scholar

Marley, Alexandra. 2020. Kundangkudjikaberrk: Variation and change in Bininj Kunwok, a Gunwinyguan language of northern Australia. Canberra: Australian National University PhD thesis.Search in Google Scholar

Meehan, Betty. 1982. Shell bed to shell midden. Canberra: Australian Institute of Aboriginal Studies.Search in Google Scholar

Merlan, Francesca. 1981. Land, language and social identity in Aboriginal Australia. Mankind 13. 133–148.10.1111/j.1835-9310.1981.tb00716.xSearch in Google Scholar

Migge, Bettina & Isabelle Léglise. 2013. Exploring language in a multilingual context: Variation, interaction and ideology in language documentation. Cambridge: Cambridge University Press.10.1017/CBO9780511979002Search in Google Scholar

Miller, Wick R. 1971. Dialect differentiation in the western desert language. Anthropological Forum 3(1). 61–78. in Google Scholar

Mirritji, Jack. 1976. My people’s life. Milingimbi: Milingimbi Literature Centre.Search in Google Scholar

Morphy, Frances. 1977. language and moiety: Sociolectal variation in a Yu:lngu language of North-East Arnhem land. Canberra Anthropology 1(1). 51–60. in Google Scholar

Myers, Fred R. 2002. Ways of place-making. La Ricerca Folklorica. 101–119. in Google Scholar

O’Keeffe, Isabel. 2016. Multilingual manyardi/kun-borrk: Manifestations of multilingualism in the classical song traditions of western Arnhem Land. Melbourne: University of Melbourne PhD thesis.Search in Google Scholar

Pennycook, Alastair & Sinfree Makoni. 2019. Innovations and challenges in applied linguistics from the global south. London: Routledge.10.4324/9780429489396Search in Google Scholar

Rumsey, Alan. 1989. Language groups in Australian Aboriginal land claims. Anthropological Forum 6(1). 69–79. in Google Scholar

Schebeck, Bernard. 1968. Dialect and social groupings in North east Arnhem land. Canberra: Australian Institute of Aboriginal Studies Unpublished manuscript.Search in Google Scholar

Singer, Ruth. 2018a. The wrong t-shirt: Configurations of language and identity at Warruwi community. The Australian Journal of Anthropology 29(1). 70–88. in Google Scholar

Singer, Ruth. 2018b. A small speech community with many small languages: The role of receptive multilingualism in supporting linguistic diversity at Warruwi Community (Australia). Language & Communication (Indigenous Multilingualisms) 62B. 102–118. in Google Scholar

Singer, Ruth & Salome Harris. 2016. What practices and ideologies support small-scale multilingualism? A case study of Warruwi community, northern Australia. International Journal of the Sociology of Language 241. 163–208.10.1515/ijsl-2016-0029Search in Google Scholar

Sutton, Peter. 1978. Wik: Aboriginal society, territory and language at Cape Keerweer, Cape York Peninsula, Australia. Brisbane: University of Queensland PhD thesis.Search in Google Scholar

Sutton, Peter. 1979. Australian language names. In Stephen Wurm (ed.), Australian linguistic studies, 87–105. Canberra: Pacific Linguistics.Search in Google Scholar

Sutton, Peter. 1991. Language in Aboriginal Australia: Social dialects in a geographical idiom. In Suzanne Romaine (ed.), Language in Australia, 49–66. New York: Cambridge University Press.10.1017/CBO9780511620881.004Search in Google Scholar

Sutton, Peter. 1997. Materialism, sacred myth and pluralism: Competing theories of the origin of Australian languages. In Francesca Merlan, John Morton & Alan Rumsey (eds.), Scholar and sceptic: Australian Aboriginal Studies in Honour of L.R. Hiatt, 211–242. Canberra: Aboriginal Studies Press.Search in Google Scholar

Sutton, Peter. 2003. Native title in Australia: An ethnographic perspective. Cambridge: Cambridge University Press.10.1017/CBO9780511481635Search in Google Scholar

Vaughan, Jill. 2018. “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land. Language & Communication 62. 119–132. in Google Scholar

Vaughan, Jill. 2019. The ordinariness of translinguistics in Indigenous Australia. In Jerry Won Lee & Sender Dovchin (eds.), Translinguistics: Negotiating innovation and ordinariness, 90–103. London/New York: Routledge.10.4324/9780429449918-8Search in Google Scholar

Vaughan, Jill. 2022. Multilingualism. In Claire Bowern (ed.), The handbook of Australian languages. Oxford: Oxford University Press.Search in Google Scholar

Vaughan, Jill & Ruth Singer. 2018. Indigenous multilingualisms past and present. Language & Communication (Indigenous Multilingualisms) 62B. 83–90. in Google Scholar

Walsh, Michael. 1997. How many Australian languages were there? In Darrell Tryon & Michael Walsh (eds.), Boundary rider: Essays in Honour of Geoffrey O’Grady, 393–412. Canberra: Pacific Linguistics.Search in Google Scholar

Wilkinson, Melanie. 1991. Djambarrpuyngu: A Yolngu variety of Northern Australia. Sydney: University of Sydney PhD thesis.Search in Google Scholar

Wood, Ray. 2016. The problem of “tribal names” in eastern Australia: The Kuku Yalanji example. In Jean-Christophe Verstraete & Diane Hafner (eds.), Land and language in Cape York Peninsula and the Gulf country, 337–360. Amsterdam, Netherlands: John Benjamins Publishing Company.10.1075/clu.18.16wooSearch in Google Scholar

Received: 2021-01-15
Accepted: 2022-03-30
Published Online: 2022-05-18

© 2022 Jill Vaughan et al., published by De Gruyter, Berlin/Boston

This work is licensed under the Creative Commons Attribution 4.0 International License.