“We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land

Language & Communication xxx (2018) 1–14

Contents lists available at ScienceDirect

Language & Communication journal homepage: www.elsevier.com/locate/langcom

“We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land Jill Vaughan a, b a b

Department of Language and Literature, Norwegian University of Science and Technology, Trondheim, Norway Research Unit for Indigenous Language, University of Melbourne, Victoria, Australia

a r t i c l e i n f o

a b s t r a c t

Article history: Available online xxx

In Arnhem Land, northern Australia, speakers of the Burarra language live and communicate within a highly multilingual and multilectal language ecology. This paper explores how regional ideologies of socio-cultural distinctiveness and unity are projected into the linguistic space at the level of the language (within Maningrida’s language ecology), as well as at the level of the lect (in terms of dialects and sociolects within the Burarra language). Drawing from current ethnography, naturalistic interactional and elicited language data, and other existing materials, the paper considers how speakers reproduce and evaluate language-internal variation within a linguistically diverse region. These processes are contextualised within the dynamics of long-term ‘egalitarian’ multilingualism which continue to shape contemporary practices and contemporary means of social meaningmaking. Ó 2018 Elsevier Ltd. All rights reserved.

Keywords: Indigenous Australia Burarra Multilingualism Language variation

1. Introduction1 In Arnhem Land, northern Australia, speakers of the Burarra language live and communicate within a highly multilingual and multilectal language ecology. Language practices in the region index a variety of social categories, including patrilineal land-holding units and moiety groupings, and local language ideologies essentialise the connections between individual linguistic varieties and particular tracts of territory. Language affiliations are in large part inherited through kinship networks. In the following quote, Isobel, a young Martay Burarra woman, describes how she connects to languages in her repertoire through her close kin: Jin-ngaypa mununa Martay Burarra jin-gata. Wurra, an-ngaypa jaminya, jin-nigipa mother, jin-gata An-barra. Rrapa ngaypa, like, Martay Burarra ngu-weya ngu-workiya. Rrapa minypa Djinang ng-galiyarra ngu-workiya, ngardawa anngaypa jaminya gu-nika wengga. Rrapa an-ngaypa ninya rrapa bapapa, jungurda apula yerrcha gun-ngayburrpa wengga Yan-nhangu, like gurda gun-gapa east, east side. Well gurdiya minypa, only like gun-ngardapa gun-guyinda marn.gi gun-gata Yan-nhangu aburr-weya, but ng-guna waya ngaypa like marn.gi ngu-nirra [.] Sometimes mix up nyiburr-weya nyiburr-workiya, mix nyiburr-negarra nyiburr-workiya minypa An-barra. Gapala yerrcha aburr-weya aburrworkiya arrburrwa ‘ngika gurda wengga, gun-guna wengga! Gun-guna gun-burral, gun-derta gun-nginyipa.’

E-mail address: [email protected]. Abbreviations: 1: first person exclusive; 12: first person inclusive; A: augmented number (alternative to ‘plural’, used in categories where ‘unit augmented’ and ‘augmented’ oppositions exist); FUT: future; I, II, III, IV: noun class (male, female, edible, land); PL: plural (alternative to ‘augmented’ where there is no opposition between ‘unit augmented’ and ‘augmented’); SG: singular; TF: temporal focus. 1

https://doi.org/10.1016/j.langcom.2017.10.002 0271-5309/Ó 2018 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

2

J. Vaughan / Language & Communication xxx (2018) 1–14

My grandmother, she’s Martay Burarra. But, my grandfather, his mother, she’s An-barra, and I speak Martay Burarra. And, like, I hear and understand Djinang, because that’s my grandfather’s language. And my father and auntie, all my grandfathers, our language is Yan-nhangu, that one over there, east side. Well that one, like, I know only a few words here and there from that Yan-nhangu that they speak, but now I understand that one. [.] We sometimes speak in a mixed-up way, we’re always mixing it (Martay dialect) with An-barra (way of speaking Burarra). The old people are always saying to us, ‘not that language, this language! This one’s the real one (for you). Your one’s a strong one.’ Isobel is multilingual, speaking a number of local Indigenous languages as well as English, but she is also multilectal within her main language, Burarra, having mastery of both the Martay and the An-barra dialects. In drawing upon her linguistic repertoire, Isobel must constantly navigate the ideological constraints that surround the appropriate use of languages and lects in her community. Isobel lives in Maningrida – Arnhem Land’s largest community, and one that represents a concentration of the linguistic diversity of the wider region. Many Indigenous languages are spoken here, alongside increasing use of English and various contact varieties. Interaction is characterised by high levels of individual multilingualism and diverse multilingual strategies (e.g. receptive multilingualism, a broad range of code-mixing practices) – practices which form a symbiotic relationship with the region’s linguistic diversity. In spite of this diversity, a lingua franca has never emerged at Maningrida,2 although the impact of changes in mobility and residence patterns (among other demographic, cultural and linguistic shifts) are observable in both the functioning of language-internal variation, and in the deployment of multilingual repertoires. In this paper, I consider how speakers like Isobel reproduce and evaluate language-internal variation within the highly multilingual language ecology of north-central Arnhem Land. These processes are contextualised within the dynamics of long-term ‘pre-existing’ multilingualism which continue to shape contemporary practices and contemporary means of social meaning-making through lived and performed sociolinguistic boundaries and affiliations. I address how regional ideologies of socio-cultural distinctiveness and unity are projected into the linguistic space at the level of the language (within Maningrida’s language ecology), as well as at the level of the lect (in terms of dialects and sociolects within the Burarra language). This work is based on naturalistic interactional and elicited language data from a number of sources, especially language materials collected since the 1960s in Maningrida and surrounds by missionaries and linguists (in particular Kathy and David Glasgow, Rebecca Green and Margaret Carew), as well as my own more recent data collected over several visits in the last three years.3

2. Arnhem Land, Maningrida and Burarra Arnhem Land is a remote Indigenous-owned territory on the northern coast of Australia’s Northern Territory. Its rich natural environment has for millennia supported a network of diverse cultural groups and, concurrently, a complex multilingual and multilectal language ecology. This language ecology or multilingual ‘regional system’ (e.g. Epps, 2008; Epps, forthcoming) is scaffolded by a range of both long-standing and contemporary practices which contribute to maintaining, and even creating, linguistic diversity. These range from cultural tendencies such as the occurrence of linguistically exogamous marriage, to local language practices such as receptive multilingualism4 and a variety of language mixing strategies (e.g. Elwell, 1977; Evans, 2010; O’Keeffe, 2016; Vaughan, forthcoming). Maningrida is a largely Indigenous community on the region’s central northern coast, and is its most populous at over 2500 inhabitants. Maningrida is a regional hub for a number of outstation communities (small settlements located on traditional homelands) and many community members regularly move between outstations, Maningrida, and Darwin, the state capital 500 km to the west. Founded in the late 1950s as a welfare settlement on the traditional lands of the Ndjébbana-speaking Kunibidji people, Maningrida has since drawn people representing groups from right across the Arnhem region, and today reflects this diversity as one of the most multilingual communities in the world. Thirteen languages from four Australian language families are currently spoken here (Fig. 1) as well as English and Kriol (an English-lexified creole spoken across northern Australia), with alternate sign systems also in frequent use (Carew and Green, 2017). As we will see, however, counting the number of local language varieties is not a matter of objective ‘fact’, but rather is a compromise between language boundaries delineated in descriptive linguistic work in the region and the most commonly attested language categories perceived and deployed in interactions in the community. Any ‘definitive’ list is likely to obscure complexities underpinning language naming strategies and the range of possible ways of delineating local linguistic varieties (Garde et al., 2015). The languages listed in Fig. 1 represent important categories for social identification. All are spoken to some degree in the Maningrida space, most on a daily basis, but the vitality and distribution of the languages vary considerably. Among the larger

2 This is typical of communicative strategies assumed to exist in pre-contact Australia more generally, i.e. that intense multilingualism, rather than reliance on lingua francas, was the norm (Brandl and Walsh, 1982; Dixon, 2002). Post-contact, lingua francas are more commonly attested, alongside widespread language shift. 3 This includes a small corpus of Burarra and several other Maningrida languages, sociolinguistic and linguistic biography interviews, targeted tasks and elicitations (e.g. mapping tasks, discussions about variation and shibboleths arising from naturalistic recordings) and more general ethnographic work in the community. 4 Receptive multilingualism refers to bilingual conversations where each interlocutor uses a different code (e.g. Singer and Harris, 2016; ten Thije and Zeevaert, 2007).

Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

J. Vaughan / Language & Communication xxx (2018) 1–14

3

language groups are Burarra/Gun-nartpa and Bininj Kunwok, which still enjoy robust intergenerational transfer and have large second language speaker communities (Elwell, 1977; Evans, 2003a; Green, 1987). Of the remaining languages listed, there is evidence that children are still learning Ndjébbana (McKay, 2000: 167), Gurr-goni (Green, 2003), Mawng (Singer and Harris, 2016) and many of the Yolngu varieties, with perhaps a small number learning Na-kara (Eather, 1990: 8–9) and Rembarrnga (McConvell et al., 2005). At the other end of the scale, languages like Yan-nhangu, Kunbarlang and Dalabon appear to be critically endangered (Bowern and James, 2010; Cutfield, 2011: 19–22; Kapitonov, forthcoming). Work by Elwell (1977, 1982) and Handelsmann (1996) provides more detailed discussions of Maningrida language demography.

Fig. 1. Genetic groupings of the traditional languages of the Maningrida region.

The traditional languages of the region span the Pama-Nyungan/non-Pama-Nyungan border (e.g. Evans, 2003c; Koch, 2014), which is to say that they represent markedly different genetic groupings and are largely not mutually comprehensible. Speakers of the Burarra language, a focus of this paper, are associated with territory to the east of Maningrida, from the mouth and western bank of the Blyth River, east to Yinangarnduwa (Cape Stewart), and south beyond the Cadell River (Fig. 2). Burarra belongs to the Maningridan subgroup of the Arnhem language group, along with Gurr-goni, Ndjébbana and Nakara. While the Maningridan languages are non-Pama-Nyungan, Burarra in particular shows evidence of long-term contact with Pama-Nyungan languages (Carew, 2016: 59), reflecting the strong social and cultural alignment of many Burarra speakers with the patrilineal Yolngu groups of the north-east. For other speakers, cultural orientations reflect more localised fusions of eastern and western Arnhem Land practices. The Burarra-speaking community covers a large territory and in fact the ‘Burarra’ language category obscures many underlying diversities. As anthropologist Ad Borsboom noted of the Djinangspeaking groups to the south-east, “shared language does not make a unity of these clans” (Borsboom, 1978: 23). There are perhaps 2000 speakers of Burarra in total, around half of whom are not first language speakers. The social orbit of Burarra speakers now centres largely on Maningrida, where they constitute the community’s largest speaker group, and Darwin. In this more urban context, the language has undergone significant transformation, resulting in a newer levelled variety, ‘Maningrida Burarra’ (Carew, 2017). This variety is also shaped by its role as frequent host for English/Kriol lexemes, resulting in stabilised code-switching constructions (Vaughan and Carew, 2016).

3. Multilingualism old and new The multilingual ecology of north-central Arnhem Land long predates the arrival of English and the spread of the contact varieties that subsequently developed. At the time of the arrival of First Fleet in 1788, some 250 distinct languages (over 700 varieties) were spoken across Australia. It is possible to construct a map of the traditional boundaries of territories associated with each of these varieties, as has been done on many occasions (see e.g. Tindale, 1974), but such schemas are more akin to a kind of religious document than to a representation of the actual boundaries of language communities (Sutton, 1991: 50–52). These attested boundaries reflect the direct ideological relationship understood to exist between territory and language, rather than between individuals, social groups and language – which is an indirect, mediated relationship. Particular tracts of land are associated on the one hand with a language, and on the other with a clan or group. These elements are primordially interconnected in origin narratives (such as the Warramurrungunji story of north-west Arnhem Land6), and are continually reproduced in ongoing situated practice and in habitus (Bourdieu,1977). Linguistic distinctions are thus “couched principally in the idiom of local geography” (Sutton, 1991: 49). Individuals are understood to ‘own’ the language of the territory they are

5

On behalf of the people of the Maningrida region, Batchelor Institute, Bawinanga Aboriginal Corporation and Maningrida College. In this narrative a woman, the creator being Warramurrungunji, emerged from the sea and travelled across the land, depositing both the coastal and inland peoples and assigning them all a different language. Versions of the story can be accessed at http://www.eopas.org/transcripts/190 in Amurdak and at http://www.language-archives.org/item/oai:www.mpi.nl:1839_00-0000-0000-0008-156A-D in Iwaidja, both languages of north-west Arnhem Land. 6

Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

4

J. Vaughan / Language & Communication xxx (2018) 1–14

Fig. 2. Language map of the Manayingkarírra (Maningrida) region (Map credit: Brenda Thornley5).

associated with by virtue of their patrilineally-inherited clan membership (Evans, 2007; Merlan,1981; Rumsey,1993), although this can create tensions between rightful ownership, in terms of language ideology, and actual language competence and use. While language categories (at various levels of ‘granularity’) can function as important social identifiers, it is evident that other kinds of local group identities are claimed and reproduced in ways that do not map closely to linguistic groupings. A range of sociocultural practices and features (such as clan networks, totemic emblems, ceremonial connections, see e.g. Cooke and Armstrong, 1998; Keen, 1994) are also drawn upon in identification processes, and these may well cross-cut shared language boundaries. Speakers exert agency over the deployment of the range of identification strategies available (e.g. Garde, 2013), and these social constructions are characterised by inherent flexibility: “sociality [in Arnhem Land] is based on open and flexible networks defined by discourse and action rather than clearly defined groups” (Borsboom, 2015: 321). The terms used to identify language varieties do not always coincide neatly with fixed entities, and they may not be ‘user-neutral’. Indeed, Evans (2003a: 8) recommends that they be treated as a kind of ‘deictic’ to reflect the fact that a term’s referent may shift depending on who uses them and in what context (see also Garde’s work (2008) on shifters and kun-dangwok, a finegrained reference system for patriclan-specific Bininj Kunwok varieties). Furthermore, there is some evidence that language labels may have become more central as identity categories in recent times, perhaps in part as a form of ‘strategic essentialism’ (McElhinny, 1996; Spivak, 1988) within discourses around native title and land rights especially (Rumsey, 1989). This is in contrast with earlier accounts, such as those of anthropologist Les Hiatt who noted that in north-central Arnhem Land Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

J. Vaughan / Language & Communication xxx (2018) 1–14

5

the speakers he worked with never referred to their own social group using a language name, and that he used ‘Gidjingali’ (another term for the Burarra language) to refer to them purely for the sake of convenience (Hiatt, 1965: 1).7 Within the multilingual language ecologies of Australia, individual language repertoires can be extensive. This is certainly the case in the Maningrida region, where speakers commonly have mastery over four to six local languages, as well as English, a range of contact varieties, and local auxiliary sign language. They may ‘hear’ (i.e. have passive competence in) several more. As we saw in Isobel’s statement, language affiliations are inherited through kinship networks (Green, 2003). Individuals marry outside of their clan group and the result is commonly a multilingual pairing (Gurrmanamana et al., 2002; Hiatt, 1965; Singer and Harris, 2016), sometimes structured by customary intermarriage patterns between particular groups (e.g. between the Gun-nartpa-speaking An-nguliny and the Djinang-speaking Wurrkiganyjarr clans (Carew, 2016: 115)). However, since several clans may be associated with a single language, it is also possible for clan exogamy to produce a marriage between speakers of the same dialect, or different dialects of the same language. While clans are typically associated in ideology with a single variety, in reality speakers’ lived experiences are highly multilingual, and marriage is just one of a constellation of practices scaffolding this system. Individuals typically connect to different languages through their parents and grandparents, as well as the broader communities and ceremonial polities they belong to. Patterns of language use vary significantly: speakers may draw on different languages to communicate with different interlocutors, across different contexts, or even within a single interaction to achieve some social interactional goal in ways that are not necessarily predictable. Elwell’s early data from Maningrida (1977, 1982) amply demonstrates the extreme variability inherent in the deployment of linguistic repertoires in the community. She recorded natural, day-to-day interactions at the community supermarket and at the school, and found that speakers drew on a wide range of multilingual resources in their communicative strategies. Table 1 exemplifies the diversity of language choices made by local students (aged 8 to 12) with a range of first languages during several ‘reading periods’ at the school library. Table 1 Code choice across 26 dyadic and multiadic interactions at the school library (Elwell, 1977: 93–98 and Appendix F). Mode of interaction

Languages used

Number of interactions

Single language

Burarra Ndjébbana Gupapuyngu Gumatj English

10 4 4 1 2

Code-mixing

Burarra and Gupapuyngu Ndjébbana, Na-kara and English

1 2

Receptive multilingualism

Ndjébbana and Na-kara Wangurri and Gumatj

1 1

While over a third of the interactions documented occurred in Burarra, reflecting the fact that Burarra is a common L1 and L2 in the community, various other strategies were drawn upon, including two forms of language mixing. Although the codes selected in the data often reflect the main language affiliations of one or more of the interlocutors involved, Elwell found that knowing the details of the individual linguistic repertoires in contact was not sufficient to predict which language would be used in any given interaction. Language choice in this highly multilingual community is strategic, sensitive to complex socioindexical pressures, and not always predictable. Micro-interactional choices such as these are a fundamental part of the scaffolding of practices and ideologies that support the multilingual language ecology of Arnhem Land (see Singer, this volume). Language groups in the region have traditionally been small, but have long existed within a sustaining system akin to what has been termed ‘small-scale’ or ‘egalitarian’ multilingualism in other parts of the world (e.g. François, 2012; Lüpke, 2015, 2016). Such systems are described as featuring multiple languages spoken by relatively small speaker groups (<5000) who readily learn each other’s languages. Crucially, these languages are not subject to a strict social hierarchy, and thus form language ecologies which differ fundamentally from contexts of polyglossia (e.g. Ferguson, 1959; Fishman, 1967) – that is, contexts where the co-existence of languages in a multilingual community depends on domain separation and differing social status of languages (as is familiar from regimented nation-state multilingualism across the world). The ramifications of egalitarian multilingualism are diverse and profound, and include important implications for language contact and change. It is suggested that ecologies of this kind may favour language differentiation, maintenance and grammatical diffusion, and may place limits on lexical borrowing (e.g. Epps, this volume; Evans, 2003b; Garde, 2008), although such claims merit further investigation. In the context of contemporary Maningrida, reflexes of a small-scale egalitarian multilingual ecology are evident. This is an increasingly rare situation worldwide, and is distinct from the vast majority of Indigenous communities in Australia, which

7 This is related to the relatively common situation whereby a linguist (or some other stakeholder) introduces a term that is not in common usage in the language community, or uses an existing term in a new way, to refer to a ‘language’ (in linguistic terms) or other macro linguistic grouping. This is often a “collective label[s] of referential convenience” (Garde, 2008: 148) for a dialect chain, such as in the case of Bininj Kunwok (see Evans, 2003a), or a regional group of related varieties, such as in the case of Yolngu Matha, previously known in the anthropological literature (after William Lloyd Warner) as ‘Murngin’ (see e.g. Schebeck, 2001: 53; Wilkinson, 1991). One possible result of such an act can be the increasing reification and reproduction of these terms within the language community itself. This is certainly the case for Yolngu Matha and may also be for Burarra (Garde et al., 2015).

Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

6

J. Vaughan / Language & Communication xxx (2018) 1–14

have experienced extreme language shift and loss. That is not to say, however, that the language ecology of the Maningrida region is unchanged from the pre-contact era – far from it. Indeed, in some ways the extreme levels of multilingualism readily observable in the community are a modern phenomenon. The density of different cultural and linguistic groups resident in Maningrida represent the urban centralisation of a large region, following the ‘assimilation and integration’ stage of the region’s post-contact history (Altman, 1987: 4). The multilingualism that persists in the community is nevertheless remarkable, and its scale is unique within the Australian context. Interactions comparable to those Elwell described in the late 1970s are still commonplace, although the impact of the increased presence of English, and of the diminished use of certain local languages is evident. Table 2 shows language choices made across 144 clauses/short utterances at a church event in Maningrida. The context is a Good Friday procession of forty or so worshippers from across various language groups, which begins at the church but proceeds around each suburb of the community.8 Table 2 Code choice across 144 clauses/short utterances at a Good Friday church event. Mode of interaction

Languages used

Number of clauses

Single language

English Kuninjku Burarra/Gun-nartpa Djinang Ndjébbana Rembarrnga Shared regional features9

33 29 23 12 2 1 1

Code-mixing

Burarra and English Kuninjku and English Burarra and Ndjébbana Kuninjku and shared Burarra, Kuninjku and English Burarra, Djinang and English

32 7 1 1 1 1

As in Elwell’s library data, Burarra is a key code choice in the Good Friday procession. Burarra is the main language of many speakers featured in the event, and this accounts for at least some of the motivation behind the frequency of its use. Equally, the utterances in Kuninjku, Djinang, Ndjébbana and Rembarrnga were predominantly made by speakers who are patrilineally affiliated with those languages via their clan membership. There are 25 speakers represented in the data above, and almost all (22 in total) used only their main language during the procession, or code-switched between their main language and English/shared features. Only three speakers, two of whom were men in leadership roles in the procession, drew on traditional languages other than their father’s language. One of the leaders of the procession used both Djinang and Burarra in roughly equal measure in his speech, such as in (1); Djinang is his father’s language, but the Martay dialect of Burarra is his grandmother’s language. (1)

Yaw, lim-busi-ban Jesus, lim-busi-ban Jesus! Nguburr-bu barra nguburr-bu barra! yes 1PL-hit-TF 1PL-hit-TF 12A-hit FUT 12A-hit FUT DJINANG BURARRA ‘We’re hitting Jesus, we’re hitting Jesus! We’ll hit (him), we’ll hit (him)!’10 (SR: 20150403-Good_Friday: 11:31–11:36)

It seems to be the case that speakers in this highly performative public context are particularly compelled to use their ideologically-prescribed main language. Interaction here is not dyadic, but rather broadcast, and so speakers need not consider the linguistic repertoire of any particular single interlocutor in their choice of code. The semantic content of the utterances is in many cases partly ‘redundant’11: much of the data consists of the same idea being expressed in several local languages (as in (1)). Another major difference from the Elwell data is the frequent use of two modes in particular: English (as the only language), and Burarra-English code-mixing. English is a common code choice in the community, but is largely limited to interactions with Balandas (non-Indigenous people) and ‘hybrid’ or ‘third space’ domains (e.g. Bhabha, 1994; Soja, 1996) where diverse groups come into contact in intercultural spaces. English also features heavily in the language of the church ‘genre’. A significant proportion of the 33 clauses in English in this data are relatively fixed expressions such as “crucify him!” and “King of King, Lord of Lord!”. Burarra-English code-mixing is also an increasingly central mode in such hybrid contexts (Vaughan, forthcoming), used by

8 The author recorded the procession, and the utterances transcribed reflect those that were in clear hearing distance of the camera. As the communicative context was highly dynamic, with speakers walking and frequently calling out simultaneously, a number of utterances were omitted from the count as they were unclear. 9 This refers to the large number of ‘shared’ words across the region’s languages, many of which were originally borrowed from languages to the north as a result of contact with Maccassan trepang fishermen who visited the region over a long period of time, up until the early twentieth century (Evans, 1992; Walker and Zorc, 1981). 10 In this example, the speaker is playing the role of a jeering member of the crowd which abuses Jesus as he carries the cross to Calvary. 11 Although I take Carew’s view (2016, following Enfield, 2009 that such acts should not in fact be considered as ‘redundancy’, but rather as contributing to the “composite nature of semiotically complex utterances” (2016: 134).

Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

J. Vaughan / Language & Communication xxx (2018) 1–14

7

L1 and L2 speakers of Burarra, especially (but not exclusively) in contexts which involve communication across several speaker groups. The frequency with which these modes feature in the data provides an exemplification of some significant changes within the local language ecology that distinguish it from more ‘traditional’ small-scale egalitarian multilingualism. While reflexes of this system continue to be evident, it is clear that the introduction of English (and English-lexified contact varieties) has to some extent disrupted the operation of the sustainable ‘balance’ between the local traditional languages (a balance which is nevertheless supported by local language ideologies (Singer and Harris, 2016)), and that such changes may intensify over time. There is a sense in which Burarra-English code-mixing could also be considered a kind of highly localised lingua franca in Maningrida (Vaughan, forthcoming), but its role is not widespread or consistent enough for this epithet.12 It has been observed by various scholars that ‘against the odds’ no communilect has ever arisen as a lingua franca between the groups in Maningrida (e.g. Elwell, 1977; McKay, 2000). Instead, Maningrida has remained resolutely multilingual, and this represents one of the few examples worldwide of continuing post-contact egalitarian multilingualism between local traditional languages (see also Carew, 2016; Green, 2003; and O’Keeffe, 2016; Singer and Harris, 2016 for Arnhem Land more generally). 4. Delineating the sociolinguistic space: The Burarra ‘dialects’ Within a diverse language ecology that is characterised by extreme variety in multilingual practice, what can be said about language-internal divisions and variation? Work on the sociopsychological and interactional aspects of multilingualism typically focus on the manipulation and socio-indexicalities of entire codes, rather than of lects and linguistic variants within a single macro code (e.g. Heller, 2007; Pavlenko and Blackledge, 2004; although notable exceptions include Migge and Léglise, 2013; and Stanford and Preston, 2009). In this and the following sections, I consider the questions of what constitutes a (dia) lect, and how language-internal variation functions in the context of multilingual north-central Arnhem Land. I focus on the Burarra language as a case study for exploring these issues, and compare how ideologies favouring diversity play out in both the manipulation of language-internal variation and in multilingual practice. In dividing up the sociolinguistic space of the Burarra language, it is necessary to navigate a range of conflicting perspectives, tensions and varying positionalities. Fig. 3 shows the most commonly reproduced schema for delineating the varieties, with the whole group referred to as Burarra or Gu-jingaliya (meaning ‘it-with-tongue’), and four sub-varieties distinguished: An-barra, Martay, Maringa and Gun-nartpa. These four categories are referred to in local accounts as ‘dialects’ although, as we will see, these ‘dialects’ may not behave as might be expected from descriptions of dialectal variation in other linguistic contexts. These dialects are associated with distinct territories within the wider Burarra region, although understandings of the boundaries of these territories vary, and in any case they are usually not strict physical boundaries in the sense of a national border. Speakers tend to agree that the An-barra variety is associated with land to the west of the Blyth River, Martay with land to its east, Gun-nartpa with country further south, around the Cadell River and Bamboo Creek, and Maringa with coastal territory towards Yinangarnduwa (Cape Stewart). These regional affiliations function as a proxy for connections with landbased patrilineal social groups and ceremonial polities (Carew, 2016: 16; Glasgow, 1994: 7), and can function as central identity categories, reproduced through “communities of descent” (Stanford, 2009). The labels for the dialects also serve as macro labels for the clan groups of those territories, and thus ultimately represent important sociocultural distinctions which have linguistic reflexes. As a regional constellation, these categories align with a local origin story for the dialects, articulated in a narrative Burarra man Noel Cooper collected from his mother (Cooper, 1991). In this retelling, the dialects began life further east at Yurrwi (Milingimbi), but “the tribal fight” scattered speakers to the west, driving them first to Cape Stewart (Maringa dialect) and then on to the Blyth River, forming two distinct dialects on either bank (An-barra and Martay). At a later time, some speakers who had remained at Cape Stewart moved across the floodplains to Ji-balbal and the Cadell River – these were the Gun-nartpa people. Carew (2016: 59) notes that this trajectory is also consistent with recorded oral histories and contact features across the Burarra varieties, although, as we will see, this schema is not a universal one. The macro labels themselves have a somewhat tenuous position: their use as group labels may be a relatively recent phenomenon. As we saw in Section 3, Hiatt observed in the 1960s that the speakers he worked with never used a language label to refer to their social group (Hiatt, 1965), and Carew (2016: 17) surmises that the ‘Burarra’ designation likely originated as an eastern exonym for the language group (see also Borsboom, 1978; Glasgow, 1994; Mirritji, 1976; Warner, 1937). It is plausible that both ‘Burarra’ and ‘Gu-jingaliya’ have benefited from ‘artefactualisation’ processes (Blommaert, 2008), whereby a language is named and delineated in something like a dictionary or a grammar. In the case of ‘Burarra’, Glasgow’s (1994) pan-dialectal dictionary and the labelling of the language in the school’s bilingual program may have further promoted and stabilised the use of that macro term, while the adoption of ‘Gidjingali’ (i.e. Gu-jingaliya) to refer to the coastal An-barra group within anthropological work in the region (e.g. Clunies-Ross, 1978; Hiatt, 1962, 1965; Meehan, 1982) may have exerted some influence on that term’s use. Both scholarly and local accounts about the referents of ‘Burarra’ and ‘Gu-jingaliya’ vary, with some viewing them as equivalent in their ability to designate the entire group, and others using one or the other to refer to all but the Gun-nartpa dialect (Schema B) (see e.g. Glasgow, 1994; Green, 1987). Carew (2016: 17) describes Gu-jingaliya as a “more socially neutral term” that is used variously by different speakers.

12 In Vaughan (forthcoming), I discuss how this speech style might be considered a kind of ‘multilingua franca’ (Makoni and Pennycook, 2012) whereby mixed but varied language use drawing on a wide range of multilingual features is the norm, or even a ‘metrolingua franca’ (Pennycook and Otsuji, 2014) – a multilingua franca of particular spaces which emerge from local contexts of interaction.

Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

8

J. Vaughan / Language & Communication xxx (2018) 1–14

Fig. 3. a: The Burarra varieties (Schema A). b: Schema B. c: Schema C. d: Schema D. e: Schema E.

As noted in Section 3, it is also likely that macro language categories in the Burarra context have become more central as group labels in response to broader discourses around native title and land rights, where asserting a monolithic social identity may have strategic benefit. While these four sub-groupings are reproduced in language ideologies and many local accounts, there are in fact tensions and conflicts in how the Burarra linguistic space is divided up. The differences attested depend in part on which viewpoint is taken, in terms of which of the subgroups a particular individual belongs to. Three alternative delineations are particularly widely attested, shown above as Schemas C, D and E. Schema C represents an ontology at times adopted by Burarra speakers who are typically not closely affiliated with the Gunnartpa sub-group. In Schema B, Gu-jingaliya is used as a meso-level term to group together speakers of the An-barra, Martay and Maringa dialects, but Schema C allows for the distinctions between those three groups to be erased, constituting a single united group in opposition to speakers of the Gun-nartpa dialect. This perspective is most commonly adopted by those associated with coastal territory, or territory on either bank of the Blyth River (i.e. Maringa, An-barra and Martay people), and is useful for speakers when they wish to assert commonality between those sub-communities, and/or distance the Gun-nartpa. Schema C may, however, be dispreferred among Maringa speakers, as the process of conjuring a monolithic Gu-jingaliya group involves minimising the distinctiveness of Maringa. This may be viewed as problematic for those with Maringa Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

J. Vaughan / Language & Communication xxx (2018) 1–14

9

affiliations, as this sub-group is quite commonly erased, or at least not distinguished, by speakers of all the other dialects (Schema D), and is typically subsumed within the Martay category. This may have to do with the perceived lack of linguistic distinctiveness of this group; Maringa is not readily identifiable via invoking a set of lexical shibboleths, unlike the three other dialects which, for example, can be denoted by a regional variant of the altercentric focus demonstrative: gun-narta (Anbarra); gun-narda (Martay); and gun-nartpa. Maringa speakers themselves, however, may assert linguistic distinctiveness; in (2) Maringa woman RB is explaining how Maringa people speak differently from others: (2)

gun-bachirra ngu-ngunyja barra, we talk in saltwater words 1SG-mimic FUT ‘I’ll always mimic the salt water (in my speech), we talk in saltwater words’ IV-salty

Such a statement enables RB to distance Maringa speech from that of inland or ‘freshwater’ speakers (i.e. Martay, Gun-nartpa, and some An-barra speakers13), while not necessitating specific examples of linguistic difference. In this way, the difference between Schemas A/B and Schema D could represent, on the one hand, the operation of a kind of erasure (i.e., “the process in which ideology, in simplifying the sociolinguistic field, renders some persons or activities (or sociolinguistic phenomena) invisible” (Irvine and Gal, 2000: 38)) on the part of non-Maringa speakers to minimise the distinctiveness of Maringa. On the other hand, it could be seen to exemplify the operation of a form of ‘fractal recursivity’ (Irvine and Gal, 2000: 38) on the part of Maringa speakers, whereby perceived sociocultural difference is projected into the linguistic space. Where this process results in the reification or even the deliberate construction of linguistic reflexes of difference, a kind of ausbau effect (Garde, 2008; Kloss, 1967) takes place – that is, the deliberate elaboration of a variety, typically for the purposes of standardisation or nation building. This notion has been largely developed to describe western institutional language development, but as Garde (2008) demonstrates, such processes may be observed in operation among small language groups in regions such as Arnhem Land. Schema E, finally, adopts the perspective of the Gun-nartpa speaker. In her long-term linguistic work with speakers of this group, Carew clearly states that the Gun-nartpa “do not identify as Burarra speakers” (2016: 17), and distinguish themselves as freshwater people fundamentally distinct from the coastal Burarra groups. In this schema, both Gun-nartpa and the coastal dialects are included within the macro Gu-jingaliya label. The coastal/inland distinction is shored up by the cultural allegiances of the Gun-nartpa, who claim closer kinship with other inland groups (such as the Rembarrnga/Kune, Kuninjku and Djinang speaking clans) than with their Burarra neighbours, and also by some significant linguistic differences between their variety and the other Burarra dialects. The case of Gun-nartpa provides an exemplification of the fact that language categories are not the only, nor indeed necessarily the most important, means of social identification, and that primary identity categorisations may well cross-cut language boundaries. These schemas capture aspects of the important distinctions in the ways Burarra people view linguistic and social boundaries within their community. These are idealisations, and various permutations are certainly attested (e.g. Schema E, but with An-barra, Martay and Maringa collapsed into a single category). The formations described here are not static, and typically cannot be categorised as the clear purview of one sub-group or another in any straight-forward way. Instead, they represent available ways of arranging the social world in interaction, and thus individuals may manipulate several schemas across different communicative contexts, with a view to achieving differing social goals. It is clear, in any case, that the delineation of sociolinguistic space is always situated; there is “no view from nowhere, no gaze that is not positioned” (Irvine and Gal, 2000: 36, cf. Nagel, 1986). In the following section these commonly replicated sociolinguistic categories are considered against observed patterns of linguistic variation to investigate the degree to which these salient groupings have reflexes in language practice. 5. Variation and lectal coherence The sub-groups of the Burarra community are demarcated in language practice in a variety of ways. It is not clear, however, that the sets of variants attested for Burarra speakers cohere systematically into clearly identifiable lects, nor that these ‘dialects’ exhibit a significant amount of inter-systemic variation (Guy and Hiskens, 2016). Lectal variation in Burarra exists in some form within most subsystems of the language,14 although local evaluations of the social indexicalities of these variants can differ. Examples of variation documented thus far in work on Burarra (especially Carew (2016, 2017) for Gun-nartpa, Glasgow (1994), Green (1987) and my own ongoing research) are chiefly lexical, but also exist to a lesser extent within the language’s morphology, syntax and phonology. Lexical variation is notable within vocabulary for fauna (especially fish,15 but also, e.g., certain birds, dingos, goannas, wallaroos), flora and body parts. The most socially salient set of lexical variables has already been mentioned: the three altercentric focus demonstratives gun-narta, gun-narda and gun-nartpa. The patterning of these lexemes is well-known and agreed upon, with gun-nartpa even functioning as the main term for that dialect, and with gun-narta and gun-narda serving as alternative dialect names for the An-barra and Martay varieties. More limited variation

13 This dichotomy is highly situated in this case – in other contexts An-barra and Martay speaker are understood to be coastal groups (e.g. when compared to Gun-nartpa speakers, see Carew, 2016: 17). 14 Variation along other parameters (e.g. age, gender) is also attested, but will not be addressed here in any detail. 15 Fish nomenclature in the region is subject to a great deal of variation (Carew, 2016: 145). The same species may be given different names by different speakers, and there also exist ‘dialect polysemes’, whereby the same lexeme may refer to two different species.

Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

10

J. Vaughan / Language & Communication xxx (2018) 1–14

exists within the morphology, syntax and phonology, for example regional differences in noun class membership (Glasgow, 1994) marked by prefixes on nominal words and in the realisation of vowels (Glasgow, 1994: 71-2; Green, 1987: 1). Some of these are exemplified below in Table 3, but more work is needed to attain a thorough account of the nature of Burarra variation. The variables that have been documented (since the 1980s) and discussed with speakers (since 2015) can be roughly allocated into three groups, according to Labov’s framework for categorising variables (e.g. Labov, 1972). A small subset of these variables can be considered ‘stereotypes’, i.e. variables that are overtly recognized, and objects of discussion. The set of demonstratives provide the best examples of this category. A larger set belong to the class of ‘markers’, that is variables that speakers have some awareness of, and are able to manipulate, for example for stylistic purposes. The set of variants for ‘dingo’ (see Table 3) exemplify this group. Finally, there are various variables which could be considered ‘indicators’, in that they correlate relatively stably with dialectal groupings but are below the level of metalinguistic awareness. It is possible that the variants of the verb meaning ‘to tell a story; to cheer’ (Table 3) could belong in this group, but more work is needed to better grasp the distribution of such variables in day-to-day interactions. While the existing documentation noted above and ongoing work on Burarra clearly point to the existence of variables that pattern according to regional dialectal parameters, the set of variables that are oriented to, and that are salient for speakers as recognisable emblems of each group, are very few. Data pooled from these sources capture fewer than fifty lexical variables, and only a handful within the language’s morphology, syntax and phonology (although further data collection may well turn up more variation). Although these varieties have definite social reality for speakers, the variation that defines them appears not to penetrate particularly deeply into the language’s structure or word stock; by and large the linguistic systems are identical. The small subset that functions as stereotypes or shibboleths are linguistic features that have undergone iconisation to become markers of intergroup differentiation, perhaps as a result of the operation of ausbau (Garde, 2008). While a large proportion of the variants documented are strongly indexical of one particular group, others appear to be more flexible and may be recruited and invoked strategically to demonstrate either similarity or difference between groups (see Table 3). It is not the case that the attested variables associated with the Burarra dialects pattern consistently across the dialect categories introduced in Section 4. The observable patterns do, however, lend weight to the linguistic ‘reality’ of some dialects more than others. In Table 3 a subset of the documented variables is given, along with their indexical value in terms of regional affiliations. Indexical evaluations are based in those documented in accounts such as Glasgow (1994), but have been verified or augmented based on observations and discussions during recent fieldwork in Maningrida (2015–2017). This recent work revealed that speakers’ evaluations and attested usages can vary a great deal (across speakers, and even for the same speaker on different occasions16), and so the distributions presented here represent the most frequently attested information.

Table 3 Some examples of the distribution of variation in Burarra. Variants

Meaning

Associated variety/distribution

gun-narta gun-narda gun-nartpa

altercentric focus demonstrative (e.g. ‘that near you’)

An-barra Martay; now in general use in Maningrida Gun-nartpa

an-jingamberr mu-jingamberr gu-jingamberr

Rotten Cheesefruit (Morinda citrifolia)

Martay; now in general use in Maningrida Gun-nartpa Across Burarra varieties

an-gugurkuja an-mugat an-muworduk

dingo

An-barra Martay Gun-nartpa

jin-jipala jin-birniny

crab legs

Across Burarra varieties Martay, especially at Ji-marda outstation

bachikala bamatuka

long pipe

Gun-nartpa; Maringa Martay; now also across other varieties

ngukurorrkurorrk Red Bush Apple (Syzygium ngukchurrga suborbiculare)

Across Burarra varieties Martay (esp. Yilan and Ji-marda); Gun-nartpa

ngoyurra guya

nose; facial features and likeness; An-barra; Martay strength Gun-nartpa

jonama bima

back; spine

Across Burarra varieties Gun-nartpa

gorkorija garkarija

to tell a story; to cheer

An-barra Martay

achila

feminine dative pronoun

Across all dialects but reportedly used in many more constructions among An-barra speakers (e.g. Martay: nguna-buna vs. An-barra: nguna-buna achila ‘she hit me’). Now used widely among younger speakers in Maningrida

16 For example, a Martay Burarra woman (53 years old) initially produced mun-jongorrk for grass fire. Later however, when we had discussed alternative variants and she had highlighted mun-jangarrk instead as the appropriate Martay variant, she said she didn’t know why she had said mun-jongorrk, and that she usually never used that. Such examples highlight tensions between language ideology and actual language use.

Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

J. Vaughan / Language & Communication xxx (2018) 1–14

11

A number of points of interest may be noted in the table. First, there is some support here for the (albeit shallow) linguistic distinctiveness of the dialect groups. Many variants were evaluated as specific to the categories of An-barra, Martay and Gunnartpa, although only one term was understood to index the Maringa variety. Secondly, there are examples where An-barra and Martay align (in opposition to Gun-nartpa), where Martay and Gun-nartpa align, and where Gun-nartpa and Maringa align (in opposition to Martay) (cf. Watson (this volume) on constellations of correlation and variance in Casamance, southern Senegal). Some of these patterns are in line with the delineations described in Section 4, especially where An-barra and Martay pattern together, and the Gun-nartpa-Maringa alignment lends tentative support to the Burarra origin narrative in Cooper (1991). Also of note is the observation that several variants attributed in the Glasgow dictionary to the Martay dialect are now in more general use across all groups living in Maningrida. Similarly, the use of the dative achila appears to have broadened from an An-barra variant to a more widespread distribution. Finally, in a number of the more recent discussions, certain variants have been evaluated as indexing particular outstations rather than a more general dialect category. Such highly localised variation highlights the varying ‘granularities’ at which language practices can take on social meaning (Section 3). The historical trajectories of these indexicalities are unfortunately largely opaque to a contemporary analysis, but it is possible that variation and its recruitment to constructing distinct social groupings in some traditional Australian language ecologies was in earlier times much more fine-grained (e.g. Garde, 2008; Walsh, 1997). The differing alignments of variants with variety labels (or locations) raises the question of to what extent the variants contribute to ‘linguistic coherence’ (e.g., Guy, 2013; Guy and Hiskens, 2016) in the Burarra varieties. Linguistic coherence refers to the extent to which multiple co-existing variables have similar distributions. This can be approached both in terms of the coherence of a language system as well as with regard to patterns of usage within a speech community (Guy and Hiskens, 2016: 1). We can begin to consider the extent of coherence within the Burarra varieties as systems, but the data available currently is not yet adequate to interrogate coherence within the current practices of individual speakers across the community. It seems clear from the data to hand that the isoglosses of the Burarra variables are largely not coincident. That is, although some variables show, for example, a three-way split with a distinct variant for An-barra, Martay and Gun-nartpa, most show a range of different alignments. This points to the conclusion that, according to current evidence, the Burarra dialects are not particularly linguistically coherent.17 The linguistic evidence does offer some limited support to the Burarra variety schemas discussed in Section 4, but also exemplifies isoglosses which cross-cut them. In any case, it is clear that despite somewhat flimsy linguistic scaffolding, these ‘dialects’ continue to be socially vital and constantly reproduced in interaction within and beyond the Burarra speaker community. There are several plausible ways of accounting for this divergence in social and linguistic fact. One explanation could be that the indexicalities (and perhaps also isoglosses) of variables have eroded to some extent over time, and/or have been reworked and recruited at a different level of granularity, thus obscuring earlier formations. Another interpretation is that these varieties are not in fact ‘dialects’ in the sense of how that term is commonly understood: there is no ‘standard’ variety, all are evaluated as equal, and real variation between them is minimal. They are, however, undeniably regional lects of a kind, and are broader than highly localised varieties like clanlects (e.g. Garde, 2008; Wilkinson, 1991). We might describe these varieties instead “as notional and ideological reference points” with real social dimensions, but without being “fully reified as full-blown linguistic systems” – a description that Lüpke (2015: 5) ascribes to the language ecology of the Casamance region in Senegal. Instead, socially important categories (in this case, macro groupings of land-based patrilineal groups) are performed and reproduced through the dimensionalisation and manipulation of linguistic variants, via fractal recursivity. It is important to note that the conclusions drawn here are not based on variation that has emerged organically from a large corpus of naturalistic language use from across the Burarra varieties, but rather rely in part on earlier work (such as Glasgow’s dictionary which drew much data from a bible translation project), and in part on my own contemporary corpus and recent discussions with speakers about where variation exists. Information of the latter kind is certainly fundamentally shaped by language ideologies, and therefore may not represent real language use (although is interesting for other reasons). It is expected that a larger-scale corpus-driven approach – currently in progress – will yield further informative and potentially contrasting results. Since a good deal of the variation that has been documented is at present used by older Burarra speakers who may spend more time on country outside of Maningrida, these lectal systems are now endangered and timely documentation is critical.

6. Concluding comments: the functioning of variation in contemporary egalitarian multilingualism Looking beyond the immediate context of Maningrida and the Burarra social space to further afield in Arnhem Land provides useful context for understanding the processes at play in lectal differentiation (both across and within language boundaries). Language documentation work in Arnhem Land has shown that variation may be dimensionalised along a number of lines, from patriclan-specific interjections and morphology among the Yolngu languages (e.g., Morphy, 1977; Schebeck, 2001; Wilkinson, 1991) and the Bininj Kunwok varieties (Evans, 2003a; Garde, 2008), to moiety-based language

17 This is certainly not a unique situation. Speech communities are not so sociolinguistically coherent as Weinrich, Labov and Herzog’s (1968) ‘orderly heterogeneity’ might imply, and it is common for dialects to show cross-boundary isoglosses. Furthermore, some have questioned whether we can assume any linguistic system to be coherent, as this implies a kind of stasis or inflexibility, cf. work in language and identity on speaker agency in manipulating variation for strategic interactional ends, such as Eckert’s bricolage (2008).

Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

12

J. Vaughan / Language & Communication xxx (2018) 1–14

boundaries in Ndjébbana (McKay, 2000), Djinang/Wurlaki (Elliot, 1991), Gurr-goni (Green, 2003) and the Yolngu languages, as well as to variation dimensionalised according to salient geographical features, like coastal vs. inland (McKay, 2000). The broader driver behind such fine-grained systems of variation is again a kind of ausbau process (as noted by Garde (2008)) – that is, the development of deliberate elaboration to create distance and distinction for important social goals. This is a process normally described for standard languages in larger nation states but seen here at work in smaller languages also. In this context, such processes might operate via the reification of individually-generated variants as markers of social identity, variants which may otherwise have passed into ‘indexical obscurity’ (Evans, 2016). These ‘invisible hand processes’ (Evans, 2003b; Garde, 2008; Keller, 1994) are driven by powerful ideologies that favour diversity and diversification – the same forces that have created and continue to support linguistic diversity at the language level. There is a sense, then, in which we can observe small-scale egalitarian ‘multi-lectalism’ in tandem with that kind of multilingualism, where ‘doing difference’ is achieved with a smaller set of linguistic tools, and capitalises on language-internal variation. This demonstrates that entirely independent linguistic systems are not necessary for these ideologies to succeed in their creation and maintenance of cultural and linguistic diversity. However, as with small-scale egalitarian multilingualism, these processes have been somewhat disrupted by interactions with newer forces in settings like Maningrida, where developments like the production of literacy materials, and the institutional use of local languages has encouraged standardisation and levelling processes to occur, especially in Burarra. Older dimensionalisations are also transformed in this urban space, as what was once more localised variation defined along various lines (e.g. Walsh, 1997) is now also recruited to index a range of ‘supra-local’ categories (Meyerhoff and Niedzielski, 2003) in contemporary contexts, such as in the broadening of the indexical values of certain Martay and An-barra variants (e.g. gun-narda, an-jingamberr). New structures of variation emerge to serve new social goals. The production and evaluation of variation and the interaction of codes, both in the context of Arnhem Land and in highly multilingual contexts more generally, has been acknowledged as providing challenges for the terminological toolbox of the researcher (e.g. Keen, 1995). The treatment of variation within this language ecology means that local perspectives on language differences will typically identify many more language varieties than a traditional dialectology approach would yield (Garde, 2008: 147). It can present a challenge for the researcher to ‘unlearn’ assumptions of how social and linguistic groups interact based in ideologies of linguistic nationalism (Merlan, 1981; Rumsey, 1993), but to do so is essential in gaining an understanding of locally-salient parameters for variation and of “the ideological aspect of analysis [to understand] how people experience the cultural continuities and interruptions” (Silverstein, 1998: 420). Contemporary small-scale multilingual contexts, especially in more urban settings, provide a particular challenge, however, in their confluence of egalitarian ideologies with processes of standardisation and emergent polyglossia. In this paper, I have endeavoured to locate language-internal variation (in the form of dialectal variation) within a broader framework of contemporary, post-contact egalitarian multilingualism, as practiced in north-central Arnhem Land. The role of variation in this context is important to understand both for ongoing local research, and for work in sociolinguistics more broadly. In Maningrida, a newer variety of Burarra is emerging in this highly diverse urban context. This is the outcome of various processes, including levelling of the contributing Burarra dialects and stabilising code-switching practices with English and Kriol. An understanding of the functioning of variation in the older Burarra dialects is crucial in order to explore the fate of dialectal variation in this newer variety, and it provides a rich opportunity to investigate how the socio-indexical functioning and salience of variants might play a role in determining the outcomes of language contact more broadly (Kerswill and Williams, 2002; Trudgill, 1986). Acknowledgements Sincere thanks to Maningrida community, and especially to Abigail Carter, Doreen Jinggarrabarra, Cindy Jin-marabynana, Rebecca Baker, Joseph Diddo, Alistair James, Stanley Djalarra Rankin, Mason Scholes and Jessie Webb. Thanks also to Margaret Carew, Felicity Meakins, Ruth Singer, Rebecca Green and Gillian Wigglesworth for useful conversations along the way, and to the participants in the 2016 American Anthropological Association symposium ‘Evidence of ‘Pre-existing’ Multilingual Ecologies in Contemporary Indigenous Language Ideologies and Practices’, where this special issue began. This work has been funded since 2015 by the Australian Research Council Centre of Excellence for the Dynamics of Language (C.I. Felicity Meakins, University of Queensland, grant CE140100041), the Linguistic Complexity in the Individual and Society project (C.I. Terje Lohndal) at the Norwegian University for Science and Technology, and a University of Melbourne Early Career Researcher Grant (C.I. Jill Vaughan). References Altman, J.C., 1987. Hunter-gatherers Today: an Aboriginal Economy in North Australia. Australian Institute of Aboriginal Studies, Canberra. Bhabha, H., 1994. The Location of Culture. Routledge, London. Blommaert, J., 2008. Artefactual ideologies and the textual production of African languages. Language & Communication 28, 291–307. Borsboom, A.P., 1978. Maradjiri: a Modern Ritual Complex in Arnhem Land, North Australia (Ph.D. thesis). Katholieke Universiteit te Nijmegen. Borsboom, A., 2015. Afterword. In: Toner, P.G. (Ed.), Strings of Connectedness: Essays in Honour of Ian Keen. ANU Press, Canberra, pp. 317–321. Bourdieu, P., 1977. Outline of a Theory of Practice (Translated by Richard Nice). Cambridge University Press, Cambridge. Bowern, C., James, B., 2010. Yan-nhasu language documentation and revitalisation. In: Hobson, J., Lowe, K., Poetsch, S., Walsh, M. (Eds.), Re-awakening Languages: Theory and Practice in the Revitalisation of Australia’s Indigenous Languages. Sydney University Press, Sydney, pp. 361–371.

Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

J. Vaughan / Language & Communication xxx (2018) 1–14

13

Brandl, M.M., Walsh, M., 1982. Speakers of many tongues: toward understanding multilingualism among Aboriginal Australians. Int. J. Sociol. Lang. 36, 71–81. Carew, M., 2016. Gun-ngaypa Rrawa “My Country”: Intercultural Alliances in Language Research (Ph.D. thesis). Monash University, Melbourne. Carew, M., 2017. Gun-nartpa Grammar. Available at: http://call.batchelor.edu.au/gun-nartpa-grammar/. Carew, M., Green, J., 2017. Australian Indigenous sign languages in multilingual contact zones. In: Paper Presented at the International Conference on Minority Languages. Jyväskylä, Finland, 28–30 August 2017. Clunies-Ross, M., 1978. The Structure of Arnhem Land Song Poetry (with Particular Reference to the Wama-Dupan Songs in Gidjingali). Oceania 49 (2), 128– 156. Cooke, P., Armstrong, G., 1998. Ownership and resource use on islands off the Liverpool River, Northern Territory. In: Peterson, N., Rigsby, B. (Eds.), Customary Marine Tenure in Australia. Sydney University Press, Sydney, pp. 285–306. Cooper, N., 1991. A History of the Dialects of the Burarra Language (Written in Burarra and English), Aboriginal Languages Fortnight June 18, 1991. Batchelor College, Maningrida. Cutfield, S., 2011. Demonstratives in Dalabon, a Language of Southwestern Arnhem Land (Ph.D. thesis). Monash University, Clayton. Dixon, R.M.W., 2002. Australian Languages: Their Nature and Development. Cambridge University Press, Cambridge. Eather, B., 1990. A Grammar of Nakkara (Ph.D. thesis). Australian National University, Canberra. Eckert, P., 2008. Variation and the indexical field. J. Sociolinguist. 12 (4), 453–476. Elliott, C., 1991. ’Mewal is Merri’s name’: form and ambiguity in Marrangu cosmology, North Central Arnhem Land (Master of Arts Thesis). Australian National University, Canberra. Elwell, V., 1977. Multilingualism and Lingua Francas Among Australian Aborigines: A Case Study of Maningrida (Honours Thesis). Australian National University, Canberra. Elwell, V., 1982. Some social factors affecting multilingualism among Aboriginal Australians: a case study of Maningrida. Int. J. Sociol. Lang. 36, 83–104. Enfield, N., 2009. The Anatomy of Meaning: Speech, Gesture and Composite Utterances. Cambridge University Press, Cambridge. Epps, P., 2008. A Grammar of Hup. Walter de Gruyter, Berlin/Boston. Epps, P. Amazonian linguistic diversity and its sociocultural correlates. In: Crevels, M., Muysken, P. (Eds.), Language Dispersal, Diversification, and Contact: a Global Perspective. Oxford University Press, Oxford, forthcoming, https://www.researchgate.net/publication/281106457_Amazonian_linguistic_ diversity_and_its_sociocultural_correlates_-_WORKING_COPY. Evans, N., 1992. Macassan loanwords in top end languages. Aust. J. Ling. 12 (1), 45–91. Evans, N., 2003a. Bininj Gun-wok: a Pan-Dialectal Grammar of Mayali, Kunwinjku and Kune (2 vols). Pacific Linguistics, Canberra. Evans, N., 2003b. Context, culture, and structuration in the languages of Australia. Annu. Rev. Anthropol. 32 (1), 13–40. Evans, N., 2003c. The Non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region. Pacific Linguistics, Canberra. Evans, N., 2007. Warramurrungunji undone: Australian languages in the 51st millennium. In: Austin, P., Simpson, A. (Eds.), Language Diversity Endangered. Helmut Buske Verlag, Hamburg. Evans, N., 2010. A tale of many tongues: polyglot narrative in North Australian oral traditions. In: Baker, B.J., Mushin, I., Harvey, M., Gardner, R. (Eds.), Indigenous Language and Social Identity: Papers in Honour of Michael Walsh. Pacific Linguistics, Canberra, pp. 275–295. Evans, N., 2016. Australian Languages. Serious Science (1 November 2016). Available at: http://serious-science.org/australian-languages-7356. Ferguson, C.A., 1959. Diglossia. Word 15, 325–340. Fishman, J., 1967. Bilingualism with and without diglossia; diglossia with and without bilingualism. J. Soc. Issues 23, 29–38. François, A., 2012. The dynamics of linguistic diversity: egalitarian multilingualism and power imbalance among northern Vanuatu languages. Int. J. Sociol. Lang. 214, 85–110. Garde, M., 2008. Kun-dangwok: “clan lects” and Ausbau in western Arnhem land. Int. J. Sociol. Lang. 191, 141–169. Garde, M., 2013. Culture, Interaction and Person Reference in an Australian Language. John Benjamins, Amsterdam & Philadelphia. Garde, M., Singer, R., Vaughan, J., 2015. Language naming in western Arnhem land. In: Paper Presented at the ‘Sociolinguistics of Multilingualism in Aboriginal Australia’ Workshop, Australian Linguistics Society Conference. Western Sydney University, 9–11 December 2015. Glasgow, K., 1994. Burarra-Gun-nartpa Dictionary: with English Finder List. Summer Institute of Linguistics, Darwin. Green, R., 1987. A Sketch Grammar of Burarra. Australian National University, Canberra. Green, R., 2003. Gurr-goni, a minority language in a multilingual community: surviving into the 21st century. In: Blythe, J., McKenna Brown, R. (Eds.), Maintaining the Links: Language, Identity and the Land. Foundation for Endangered Languages, Bath, pp. 127–134. Gurrmanamana, F., Hiatt, L.R., McKenzie, K., 2002. People of the Rivermouth: The Joborr Texts of Frank Gurrmanamana. National Museum of Australia and Aboriginal Studies Press, Canberra. Guy, G.R., 2013. The cognitive coherence of sociolects: how do speakers handle multiple sociolinguistic variables? J. Pragmat. 52, 63–71. Guy, G.R., Hinskens, F., 2016. Linguistic coherence: systems, repertoires and speech communities. Lingua 172–173, 1–9. Handelsmann, R., 1996. Needs Survey of Community Languages: Central Arnhem Land, Northern Territory (Maningrida and Outstations). Report to the Aboriginal and Torres Strait Islander Commission, Canberra. Heller, M., 2007. Bilingualism: A Social Approach. Palgrave Macmillan, Basingstoke. Hiatt, L.R., 1962. Local organization among the Australian Aborigines. Oceania 32 (4), 267–286. Hiatt, L., 1965. Kinship and Conflict: a Study of an Aboriginal Community in Northern Arnhem Land. Australian National University, Canberra. Irvine, J.T., Gal, S., 2000. Language ideology and linguistic differentiation. In: Kroskrity, P.V. (Ed.), Regimes of Language: Ideologies, Polities and Identities. School of American Research Press, Santa Fe, pp. 35–84. Kapitonov, I. Aspects of Kunbarlang Grammar (Ph.D. thesis). University of Melbourne, forthcoming, in prep. Keen, I., 1994. Knowledge and Secrecy in an Aboriginal Religion. Clarenden Press, Oxford. Keller, R., 1994. On Language Change: The Invisible Hand in Language. Routledge, London/New York. Kerswill, P., Williams, A., 2002. ‘Salience’ as an explanatory factor in language change: evidence from dialect levelling in urban England. In: Jones, M.C., Esch, E. (Eds.), Language Change: the Interplay of Internal, External and Extra-linguistic Factors. Mouton de Gruyter, Berlin, pp. 81–110. Keen, I., 1995. Metaphor and the metalanguage: “Groups” in northeast Arnhem Land. American Ethnologist 22 (3), 502–527. Kloss, H., 1967. Abstand-languages and Ausbau-languages. Anthropol. Ling.s 9 (7), 29–41. Koch, H., 2014. Historical relations among the Australian languages. In: Koch, H., Nordlinger, R. (Eds.), The Languages and Linguistics of Australia: a Comprehensive Guide. Walter de Gruyter, Berlin/Boston, pp. 23–89. Labov, W., 1972. Sociolinguistic Patterns. Blackwell, Oxford. Lüpke, F., 2015. Denorthernising Multilingualism and Multilingualism Research Framing Language in the North: from the Monolingual Nation State to Acknowledging (Super)Diversity. Multilingualism in the Global South (and beyond) [Special Issue]. Voices from around the World 3, pp. 3–5 . http:// kups.ub.uni-koeln.de/6502/1/20160117_Layout_Voices_3_Complete_Issue_(3)_(1).pdf. Lüpke, F., 2016. Uncovering small-scale multilingualism. Crit. Multiling. Stud. 4 (2), 35–74. Makoni, S., Pennycook, A., 2012. Disinventing multilingualism: from monological multilingualism to multilingua francas. In: Martin-Jones, M., Creese, A., Blackledge, A. (Eds.), The Routledge Handbook of Multilingualism. Routledge, London, pp. 439–472. McConvell, P., Marmion, D., McNicol, S., 2005. National Indigenous Languages Survey Report. Australian Institute of Aboriginal and Torres Strait Islander Studies, Canberra.

Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002

14

J. Vaughan / Language & Communication xxx (2018) 1–14

McElhinny, B., 1996. Strategic essentialism in sociolinguistic studies of gender. In: Warner, N., Ahlers, J., Bilmes, L., Oliver, M., Wertheim, S., Chen, M. (Eds.), Gender and Belief Systems: Proceedings of the Fourth Berkeley Women and Language Conference. Berkeley Women and Language Group, Berkeley, California, pp. 469–480. McKay, G., 2000. Ndjebbana. In: Dixon, R.M.W., Blake, B.J. (Eds.), Handbook of Australian Languages (Vol. 5). Oxford University Press, South Melbourne. Meehan, B., 1982. Shell Bed to Shell Midden. Australian Institute of Aboriginal Studies, Canberra. Merlan, F., 1981. Land, language and social identity in Aboriginal Australia. Mankind 13, 133–148. Meyerhoff, M., Niedzielski, N., 2003. The globalisation of vernacular variation. J. Sociolinguist. 7 (4), 534–555. Migge, B., Léglise, I., 2013. Exploring Language in a Multilingual Context: Variation, Interaction and Ideology in Language Documentation. Cambridge University Press, Cambridge/New York. Morphy, F., 1977. Language and moiety. Canberra Anthropol. 1 (1), 51–60. Mirritji, J., 1976. My People’s Life. Milingimbi Literature Centre, Milingimbi. Nagel, T., 1986. The View from Nowhere. Oxford University Press, Oxford. O’Keeffe, I., 2016. Multilingual Manyardi/kun-borrk: Manifestations of Multilingualism in the Classical Song Traditions of Western Arnhem Land (Ph.D. thesis). University of Melbourne. Pavlenko, A., Blackledge, A., 2004. Negotiation of Identities in Multilingual Contexts. Multilingual Matters, Clevedon/Buffalo. Pennycook, A., Otsuji, E., 2014. Market lingos and metrolingua francas. Int. Multiling. Res. J. 8 (4), 255–270. Rumsey, A., 1989. Language groups in Australian Aboriginal land claims. Anthropol. Forum 6 (1), 69–79. Rumsey, A., 1993. Language and territoriality in Aboriginal Australia. In: Walsh, M., Yallop, C. (Eds.), Language and Culture in Aboriginal Australia. Aboriginal Studies Press, Canberra, pp. 191–206. Schebeck, B., 2001. Dialect and Social Groupings in Northeast Arnheim Land. LINCOM Europa, München. Singer, R., Harris, S., 2016. What practices and ideologies support small-scale multilingualism? A case study of Warruwi Community, northern Australia. Int. J. Sociol. Lang. 241, 163–208. Silverstein, M., 1998. Contemporary transformations of local linguistic communities. Annu. Rev. Anthropol. 27 (1), 401–426. Soja, E.W., 1996. Thirdspace: Journeys to Los Angeles and Other Real-and-imagined Places. Blackwell, Malden, MA. Spivak, G.C., 1988. Subaltern studies: deconstructing historiography. In: Guha, R., Spivak, G.C. (Eds.), Selected Subaltern Studies. Oxford University Press, London, pp. 3–32. Stanford, J.N., 2009. “Eating the food of our place”: sociolinguistic loyalties in multidialectal Sui villages. Lang. Soc. 38 (3), 287–309. Stanford, J.N., Preston, D.R., 2009. Variation in Indigenous Minority Languages. John Benjamins Publishing, Amsterdam/Philadelphia. Sutton, P., 1991. Language in Aboriginal Australia: social dialects in a geographical idiom. In: Romaine, S. (Ed.), Language in Australia. Cambridge University Press, Cambridge,́ pp. 49–66. ten Thije, J.D., Zeevaert, L. (Eds.), 2007. Receptive Multilingualism: Linguistic Analyses, Language Policies and Didactic Concepts. John Benjamins, Amsterdam. Tindale, N.B., 1974. Aboriginal Tribes of Australia: Their Terrain, Environmental Controls, Distribution, Limits, and Proper Names. Australian National University Press, Canberra. Trudgill, P., 1986. Dialects in Contact. Blackwell, Oxford. Vaughan, J. Translanguaging and hybrid spaces: boundaries and beyond in north central Arnhem Land. In: Mazzaferro, G. (Ed.), Translanguaging in Everyday Practice. Multilingual Education Series, Springer, forthcoming. Vaughan, J., Carew, M., 2016. ‘New’ code-switching and long-term multilingualism at Maningrida (northern Australia). In: Paper Presented at the ‘8th Australianists in Europe’ Workshop. SOAS, London (24–25 September 2016). Walker, A., Zorc, R.D., 1981. Austronesian loanwords in Yolngu Matha of Northeast Arnhem land. Aborig. Hist. 5 (2), 109–134. Walsh, M., 1997. How many Australian languages were there? In: Tryon, D., Walsh, M. (Eds.), Boundary Rider: Essays in Honour of Geoffrey O’Grady. Pacific Linguistics, Canberra, pp. 393–412. Warner, L., 1937. A Black Civilization: A Social Study of an Australian Tribe. Harper & Row, New York. Weinreich, U., Labov, W., Herzog, M.I., 1968. Empirical foundations for a theory of language change. In: Lehmann, W.P. (Ed.), Directions for Historical Linguistics: A Symposium. University of Texas Press, Austin, pp. 95–195. Wilkinson, M., 1991. Djambarrpuyngu: a Yolngu Variety of Northern Australia (Ph.D. thesis). University of Sydney.

Please cite this article in press as: Vaughan, J., “We talk in saltwater words”: Dimensionalisation of dialectal variation in multilingual Arnhem Land, Language & Communication (2018), https://doi.org/10.1016/j.langcom.2017.10.002