http://orcid.org/0000-0002-8879-9798
![Sample our Language & Literature journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5941/TOC-Subject-Sample-2021-Language-Literature.jpg"})
Abstract
This article discusses the distribution and function of a suffix that has been labelled ‘ergative’ in the literature on Dalabon, a Gunwinyguan (non-Pama-Nyungan) language of south-western Arnhem Land. Our first-hand data reveal that although this marker (-yih) more frequently occurs on A arguments of multivalent clauses, it also appears with significant frequency on S arguments of monovalent clauses, particularly with the verb root yin ‘to say, to think, to do’. We explain this non-canonical distribution with a co-dependent analysis of its discourse and pragmatic functions, summarized by the principle ‘mark out the unexpected referent’, following McGregor’s Expected Actor Principle. These functions differ slightly according to clause type. For both types, the marker has a discourse function of ‘mark out the non-topical referent’: either an A argument that sufficiently threatens the construal of local topics, or an S referent after a long period of deferred topichood (particularly speaker referents). The marker also has a correlating pragmatic function of ‘mark out the contrary referent’: either an A participant acting against the motivations and expectations of other (topical) referents (or of the speaker), or an S participant with an unusual stance or speech content.
Notes on Contributors
Ellison Luk is a recent graduate of the University of Sydney and has worked on various Australian Indigenous languages as a research assistant, including the preparation of dictionaries and language archive material. He is currently based in the Catholic University of Leuven (Katholieke Universiteit Leuven), where he is working on a project titled Beyond the Clause: encoding and inference in clause combining. Email: [email protected]
Maïa Ponsonnet is an anthropological linguist currently based at The University of Western Australia in Perth (Linguistics PhD Australian National University, 2014). She has worked extensively with speakers of Indigenous languages in central/western Arnhem Land, in particular Dalabon and Kriol, as well as on Kunwok and Rembarrnga. Email: [email protected]
ORCID
Maïa Ponsonnet http://orcid.org/0000-0002-8879-9798
Notes
* This article is a revision of Ellison Luk's Honours thesis ‘Optionality in Grammatical Structure: The discourse basis of ergative case-marking in Dalabon’. It is dedicated to Maggie Tukumba, who provided most of the data for this article and many others on Dalabon, and who sadly passed away as we were working on our final draft. We are also immensely grateful to the other Dalabon speakers who contributed data to this study. Finally, we extend our thanks to our two anonymous reviewers, and Jean-Christophe Verstraete, who all gave invaluable advice in the preparation of this draft.
1 See Appendix A for list of glossing abbreviations.
2 Also cited as marnû-, where ⟨û⟩ represents [ɨ], which has been treated either as phonemic, or as an allophone of /u/. We neutralize the distinction in this paper, and will use ⟨u⟩ throughout (see Ponsonnet Citation2014a: xxvii).
3 Cutfield undertakes an analysis of discourse–pragmatics in Dalabon (Citation2011: 41–113) on the ordering of nominals at the level of the utterance. What we attempt is different in kind and in scope: we are concerned with nominal case-marking at the level of the discourse and, as such, our frameworks and usage of meta-language will diverge.
4 Evans’ account (Citation2003: 139) of the same cognate marker in Bininj Gun-wok (Dalabon’s closest relative) conservatively adopts the ‘instrumental’ label, but acknowledges an ergative usage (among others), as well as causal usages calqued from Dalabon. He also adopts the ‘instrumental’ label as the primary usage of the suffix, in the latest piece on Dalabon as of the time of writing (Evans Citation2017b). Given their close relatedness, the conditions we postulate for the Dalabon marker are probably similar to those for the Bininj Gun-wok marker, including a third-singular patient animacy alternation in its bound pronominal morphology similar to that of Dalabon (§2.2.1).
5 We define the following thematic roles: Agents as ‘wilful, controlling, [and] instigating participants in states of affairs’, and Patients as ‘strongly affected participants’ (Van Valin Citation2005).
6 According to McGregor’s typology (Citation2010, Citation2013), this would make Dalabon a Type 2 language: the presence of the marker has a coded function of making a referent more prominent, while its absence does not have the coded function of backgrounding it.
7 Maggie Ngarridjan Tukumba (MT), Lily Bennett (LB), Nikipini Daluk (ND) and Queenie Brennan (QB).
8 Data collection was funded by the Australian Institute of Aboriginal and Torres Strait Islanders Studies under the grants G2007/7242 and G2009/7439, and by the Hans Rausing Foundation’s Endangered Language Project under the grant IGS0125 (CI: Maïa Ponsonnet for all grants).
9 Bold indicates reference to jackal or crow. Grey highlighting indicates clauses with topic switch.
10 The names of the characters have been changed to preserve anonymity.
11 Bold type indicates reference to na-Ryan, Bangardi or kirdikird (woman). Grey highlighting indicates clauses not coreferential with na-Ryan (the discourse topic).
12 Bold indicates reference to na-Ryan or nadjamorrwu ‘policemen’. Grey highlighting indicates clauses controlled by nadjamorrwu (as local topic).
13 Bold type indicates reference to murdika ‘car’, Rosita or ‘car tyre’. Grey highlighting indicates topic change.
14 Inanimate As not marked with -yih are uncommon, but given that overt inanimate A arguments are also uncommon, the statistics are not significant.
15 We use ‘serialized’ and ‘serialization’ as theory-neutral terms, to refer to both ‘serial verbs’ or ‘serial clauses’.
16 The fact that the dataset is extracted from a corpus collected in the view to document the expression of emotions in Dalabon is likely to have favoured such occurrences. Nevertheless, given the extent of the corpus (60 transcribed hours, see §1.3), and the extent of topics covered, the bias towards emotions could not possibly have excluded the occurrence of -yih on other verbs as well.
17 The lexicalized compound wunj+mang, literally ‘belongings+get’, has the meaning of ‘to put a spell on someone using their belongings’, referring to the performance of black magic on one’s clothes and other personal effects.
18 Grey highlighting indicates reported speech head clause, grey text indicates reported speech complement clause, bold text indicates kurdang ‘sorcerer’ as topic.
19 For subordination strategies in Dalabon, see Evans (Citation2006).
20 Bold type indicates reference to young man. Grey highlighting indicates emphatic device (overt nominal phrase, negator, intensifying interjection or adverbial).
21 In Gunwinyguan languages, the ‘thematic’ traditionally refers to the monosyllabic verb root that carries the inflection (Evans & Merlan Citation2003; Saulwick Citation2003: 110–158).
22 Semantic motivations similar to what Rumsey describes have been described in other languages with ergative marking in formally monovalent clauses, such as in many Tibeto-Burman languages (Hyslop Citation2010; Willis Citation2011), Gurindji Kriol (Meakins Citation2015) and Kuuk Thaayorre (Gaby Citation2010). Although they provide fine pragmatic analyses, none provide a robust discourse analysis.
23 Speech complements (in Dalabon, as well as around Australia more generally) are difficult to analyze as embodying an O-relation, as they often take the form of finite clauses (with no subordinate marking), attached paratactically to the clause encoding the speech event.
24 Although the thematic di usually forms monovalent verbs, the compound verb njerrh-ye-merey-di attracts transitive person prefixes (see Ponsonnet Citation2014a: 173). This could be due to the presence of ye-, if this is interpreted as a comitative applicative—but it is not clear what the comitative argument would then be, and the form ye- could have other origins (see Evans Citation2006). In any case, irregularities in the valence of thematics are not exceptional in Dalabon.