The Languages and Linguistics of Australia: A Comprehensive Guide 9783110279771, 9783110279696

Citation preview

The Languages and Linguistics of Australia WOL 3

The World of Linguistics

Editor Hans Henrich Hock

Volume 3

The Languages and Linguistics of Australia

A Comprehensive Guide

Edited by Harold Koch Rachel Nordlinger

MOUTON

ISBN 978-3-11-027969-6 e-ISBN 978-3-11-027977-1 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über http://dnb.dnb.de abrufbar. © 2014 Walter de Gruyter GmbH, Berlin/Boston Cover image: uros ravbar/iStock/Thinkstock Typesetting: Dörlemann Satz GmbH & Co. KG, Lemförde Printing: CPI books GmbH, Leck ♾ Gedruckt auf säurefreiem Papier Printed in Germany www.degruyter.com

Table of contents Acknowledgements | vii Preface | ix List of Contributors | xi Maps | xiii

1 Traditional Indigenous languages of Australia 1

2

Harold Koch and Rachel Nordlinger The languages of Australia in linguistic research: context and issues | 3 Harold Koch Historical relations among the Australian languages: genetic classification and contact-based diffusion | 23

3

Janet Fletcher and Andrew Butcher Sound patterns of Australian languages | 91

4

Brett Baker Word structure in Australian languages | 139

5

Rachel Nordlinger Constituency and grammatical relations in Australian languages | 215

6

Claire Bowern Complex predicates in Australian languages | 263

7

Alice Gaby and Ruth Singer Semantics of Australian languages | 295

8

Michael Walsh Indigenous language maintenance and revitalisation | 329

vi 

 Table of contents

2 Post-Contact language varieties 9

Felicity Meakins Language contact varieties | 365

Diana Eades 10 Aboriginal English | 417 Peter Collins 11 Australian English | 449 Language Index | 485 Name Index | 491 Subject Index | 496

Acknowledgements This volume has been a number of years in the making, and we thank all of the many friends and colleagues who have helped us along the way with discussions, suggestions, reviews and advice. In particular, we would like to thank Barry Alpher, Rob Amery, Peter Austin, Juliette Blevins, Mark Harvey, John Ingram, Mary Laughren, Patrick McConvell, William McGregor, David Nash, Pam Peters, Farzad Sharifian, Jeff Siegel, Jane Simpson and Nick Thieberger. Special thanks to Mark Harvey for allowing us to use his map of northern Australian languages (Map 2), and to Claire Bowern for all her help in developing the other maps. Maps 1 and 3 were drawn using data from NSF grant BCS-0844550 Pama-Nyungan and the Prehistory of Australia, awarded to Claire Bowern (Yale University). Thanks too for the help of Joanna Mason with the revised Map 1. We are also extremely appreciative of the financial support provided by the Australian National University’s Project 0107 Aboriginal Language account (Harold Koch and Luise Hercus) and the Australian Research Council (DP0984419 Doing great things with small languages: safeguarding indigenous language material of Australian’s region by clever use of technology (CIs Nick Thieberger and Rachel Nordlinger)). Piers Kelly and Helen Bromhead provided expert and efficient editorial assistance, and we thank them for all their hard work. Our thanks also to the Mouton editorial team, especially Hans Henrich Hock, Uri Tadmor and Katja Lehming for inviting us to develop this volume in the first place, and all their support in seeing it through to completion. Special thanks to Angelika Hermann for production work and to Hilary Faulkner for creating the indices.

Preface The last 40–50 years has seen an enormous surge in work on the languages of Australia; the traditional Indigenous languages, varieties of English, and contact languages such as Australian Kriol. During this time we have also seen a dramatic increase in interest towards Australian languages and their implications for linguistic typology and theory. This volume aims to provide the reader with a comprehensive overview of the key research findings of the last 50 years on the languages and linguistics of Australia. The chapters in this volume, each written by leading researchers in the respective area, provide a starting point for anyone interested in learning about aspects of Australian languages, including Australian English and contact varieties. In writing their chapters, each author was asked to describe the state of the art for the relevant topic in Australian linguistics, including a discussion of what has been accomplished over the last 40–50 years, what the interesting issues are, and which issues are still open or in need of further research. They were also asked to introduce key concepts and terminology and to direct the reader to further references and resources as appropriate. Language names (except in direct quotations) have been regularised to the spelling of AUSTLANG (http://austlang.aiatsis.gov.au/). Abbreviations follow the Leipzig glossing rules wherever possible (http://www.eva.mpg.de/ lingua/resources/glossing-rules.php). The history of Australia has led us to separate this volume into two sections. Chapters 1 to 8 focus on the traditional Indigenous languages of the Australian continent – those that have been spoken by Indigenous communities since before British colonisation in 1788. Chapter 8 discusses the fate of Indigenous languages in the wake of colonisation. Chapters 9 to 11 focus on language varieties that have become part of Australia’s language landscape post-colonisation, namely contact languages such as Australian Kriol (chapter  9) and varieties of English (Aboriginal English in chapter 10, and Australian English in chapter 11). Here is a summary of the chapters in more detail. Chapter  1 (Koch and Nordlinger) provides an overview of Australia’s Indigenous languages, both traditionally and in the present, a summary of the history of their documentation and study, plus a brief survey of several areas of research that have not been covered in detail in this volume. Chapter 2 (Koch) discusses the historical-comparative linguistics of Australian languages, covering issues of genetic relationship as well as contact-induced change and canvassing methodological issues as well as substantive findings. Chapter  3 (Fletcher and Butcher) discusses the findings of articulatory and acoustic phonetics, discussing vowels, consonants, connected speech processes, prosodic features, intonation, and voice quality. The discussion includes the phonetics of creoles and Aboriginal English. Chapter 4 (Baker) is about the phonological and morphological aspects of word structure, including segmental inventories, phonotactics, prosodic vs. syntactic words, minimal words, word prosody (stress), word structure in polysynthetic languages, phonological alter-

x 

 Preface

nations, and reduplication. Chapter 5 (Nordlinger) surveys the syntactic issues of ergativity, nonconfigurationality, noun incorporation and polysynthesis, NP constituency, case and multiple case-marking, and subordination. Chapter  6 (Bowern) is about the phenomenon of complex predicates, bipartite verbal structures consisting of a coverb and a light verb. Five issues are identified: event structure, classification, light verb inventory and stability, the nexus between the two elements, and the word status and etymology of coverbs. This is followed by a discussion of the typology of these constructions, verb serialisation, and historical considerations of complex verb constructions, in particular their reconstructibility and behaviour in language contact. Chapter 7 (Gaby and Singer), on semantics, describes issues of categorisation and classification (ethnobiology, nominal classification), lexical sense relations (polysemy, metonymy and metaphor, relationships between registers, vagueness, lexicography), conceptual domains (kinship, numbers, body parts, emotions, the land and spatial relations), the semantics and pragmatics of grammar, semantic change, pragmatics and interaction. Chapter  8 (Walsh) turns to the consequences of European colonisation, and discusses issues of Indigenous language maintenance and revitalisation. It shows by means of case studies how the languages in different parts of the country have fared. Policies regarding language maintenance are discussed, and programs of language revitalisation are described. Chapter 9 (Meakins), on language contact varieties, describes the development, study, and social role of pidgins (both English-based pidgins and non-English-based pidgins), creoles (Kriol, Torres Strait Creole), mixed languages (Gurindji Kriol, Light Warlpiri), Indigenous koines, and restructured varieties of traditional languages. Chapter 10 (Eades), on Aboriginal English, describes the various approaches to the description of the varieties of English spoken by Aboriginal people and the Aboriginal ways of using English, and the consequences of their usage in the spheres of education and the law. Finally, chapter 11 (Collins) surveys research on the distinctive aspects of Australian English and its evolution, covering phonology, morphology, grammar, discourse, lexicon, and sociolinguistic issues. It is our hope that this volume will become a valuable resource for students and colleagues, and will help to encourage and stimulate future research on the languages of Australia.

List of Contributors Brett Baker, University of Melbourne, [email protected] Claire Bowern, Yale University, [email protected] Andrew Butcher, Flinders University, [email protected] Peter Collins, University of New South Wales, [email protected] Diana Eades, University of New England, [email protected] Janet Fletcher, University of Melbourne, [email protected] Alice Gaby, Monash University, [email protected] Harold Koch, Australian National University, [email protected] Felicity Meakins, University of Queensland, [email protected] Rachel Nordlinger, University of Melbourne, [email protected] Ruth Singer, University of Melbourne, [email protected] Michael Walsh, Australian Institute of Aboriginal and Torres Strait Islander Studies, [email protected]

Map 1: Languages of Australia (as referred to in this volume), including Pama-Nyungan boundary line

Maps

Adnyamathanha Alyawarr Anguthimri Anmatyerre Arabana-Wangkangurru Arrernte Bardi Bilinarra Buandig Bundjalung Bungandidj (see Buandig) Bunuba Central Victorian Language (Boonwurrung, Woiwurrung) Darkinyung Kalkatungu Karajarri Kaurna Kayardild Kaytetye Kok-Kaper (Koko-Bera) Koko-Bera (see Kok Kaper) Kugu Nganhcara Kuku Yalanji Kuuk Thaayorre Kuyani Lardil Linngithigh Mabuiag (see Kala Lagaw Ya) Maranunggu Martuthunira Martu Wangka Madhi Madhi (Mathi-Mathi) Mbabaram 40 41 42 43 44

34 35 36 37 38 39

12 13 28 29 30 31 32 33

11

1 2 3 4 5 6 7 8 9 10

Meriam Mir Mirning (Mirniny) Mparntwe Arrernte (see Arrernte) Mudburra Muruwari Ngaliwurru Nganyaywana Ngarinyman Ngarnga Ngiyampaa Nhanda Nungali Noongar (Nyungar) Nyangumarta Nyawaygi Nyulnyul Paakantyi Dharug Diyari Djambarrpuyngu Dyirbal Ganggalida (Yukulta) Garrwa Gippsland language (Gunnai and Bidawal) Gooniyandi Gupapuyngu Gurindji Guugu Yimidhirr Jingulu Jiwarli Kala Lagaw Ya Palawa Kani Banyjima (Panyjima)

Key to Map 1 (Note languages not listed here can be found on Map 2)

20 21 22 23 24 25 26 27 61 62

47 48 49 50 51 52 53 54 55 56 57 58 59 60 14 15 16 17 18 19

45 46

Parnkalla (Barngarla) Pintupi Pitjantjatjara Pitta Pitta Thalanyji Tharrkari Umbindhamu (Umpithamu) Umpila Ungarinyin (Ngarinyin) Waanyi Wambaya Wangkangurru (see ArabanaWangkangurru) Wangka-Yutjuru Warlpiri Wargamay Warrnambool language Warumungu Warungu (Warrongo) Wathawurrung Wemba Wemba Western Desert Language (Wati) Western Torres Strait Language (see Kala Lagaw Ya) Western Victorian Language (see Madhi Madhi, Wenba Wemba) Worrorra Yalarnnga Yankunytjatjara Yanyuwa Yawuru Yidiny Yir Yiront Yukulta (see Ganggalida) 83 84 85 86 87 88 89

74 75 76 77 78 79 80 81 82

63 64 65 66 67 68 69 70 71 72 73

Maps   xiii

Map 2: The languages of northern Australia This map is based on the contributions of many researchers, which were collated by Mark Harvey (University of Newcastle), and is available at http://www1.aiatsis.gov.au/aseda/802_Harvey/MH_top%20end.png [Note that Murriny-Patha on this map is spelled Murrinh-Patha in this volume, and Na-kara is spelled Nakkara]

xiv   Maps

Map 3: Language families and subgroups of Pama-Nyungan

Maps   xv

| Part I Traditional Indigenous languages of Australia

Harold Koch and Rachel Nordlinger

1 The languages of Australia in linguistic research: context and issues 1 Background on the Indigenous languages of Australia At the time of colonisation in the late 18th century, Australia was home to 700–800 language varieties, distributed across the continent (and including Tasmania and the Torres Strait Islands), which can be grouped into more than 250 distinct languages, some of which include a number of dialects.¹,² These language varieties were spoken across a population of around one million people (e.g. Butlin 1983), which indicates the enormous linguistic diversity of Indigenous Australia. In many cases small populations (e.g. 40–50 people) maintained distinctive language varieties, and the largest populations speaking a single language variety were probably no bigger than 3000– 4000 people. Linguistic diversity was not necessarily an impediment to communication, however, since Indigenous societies were frequently highly multilingual, with an individual often speaking up to 4–6 languages of the surrounding area. Linguistic diversity, in fact, was valued for its indexical relationship to identity and group membership (Evans 2007). The relationship between language and identity is strong for all human societies, but is particularly so in Indigenous Australia, where language is often related directly to the land. As Rumsey (1993, 2005) explains, in Australia there is a direct relationship between a language and a tract of land; in creation myths it is very common for the ancestors to be described as passing across the land instilling different languages into different areas as they go (Evans 2007: 20). People are then connected to a particular tract of land and, through that connection, to the language associated with that place. Thus the Wambaya people are Wambaya because they are linked to places which are associated with the Wambaya language, and therefore speak Wambaya (see Rumsey 1993, 2005 for discussion). This ideology leads to an important distinction between speaking a language and “owning” a language. A person will “own” the language of the land to which her clan, family or group is connected, even if she doesn’t speak it.

1 We wish to extend our thanks to Jane Simpson for reading an earlier version of this chapter and providing many suggestions and comments that have led to substantial improvements in coverage and content. 2 The number of languages cited in the literature generally ranges from 250–300. Recent work by Claire Bowern (NSF grant 0844550) suggests that the figure might be closer to 350 (270 PamaNyungan, and 80–90 non-Pama-Nyungan (Claire Bowern pers. comm. 28/11/13)).

4 

 Harold Koch and Rachel Nordlinger

Unfortunately, the 225 years since colonisation have taken a devastating toll on the traditional Indigenous languages of Australia. Of the 250 or more distinct languages spoken in 1788, only 15–18 are now being learned by children as their first language. Another 100 or so have only small numbers of elderly speakers remaining, and most have no full or fluent speakers left at all (Marmion, Obata and Troy 2014). Australia has been identified as the country that has experienced the greatest and most rapid loss of languages over the last century, of anywhere in the world (Nettle and Romaine 2000: 9), with grim estimates suggesting that if recent trends of language shift remain unchecked, there may be no speakers of traditional Indigenous languages left at all by the year 2050 (McConvell and Thieberger 2001). The linguistic situation is not unrelated to official language policy, which during most of the course of Australia’s history since colonisation has promoted an unrelenting culture of English monolingualism. Only in recent decades have there been positive government initiatives for the documentation, teaching, and public use of Indigenous languages (see chapter 8 for the general situation, § 3.6 below for a discussion of bilingual education, and chapter  10 for legal contexts). For the turn-ofthe-century situation, see Laughren (2000); for reference to Australian language policy documents see David Nash’s compilation at http://www.anu.edu.au/linguistics/ nash/aust/policy.html. Although it is generally assumed that all Australian languages are ultimately related (except for Meriam Mir from the Eastern Torres Strait, which is clearly Papuan (Piper 2013), and the languages of Tasmania (Bowern 2012)) such relatedness has been more easily established for some languages than for others. Koch (this volume, chapter 2) discusses this issue in more detail. The current common position is that there are about 25 families represented across the continent, with linguistic diversity unevenly distributed such that one single family, Pama-Nyungan, covers seven-eighths of the continent, and the remaining 24 non-Pama-Nyungan families are concentrated in a relatively small part of the north-west (see Map 3, where some of these families are indicated in small caps). While there are a number of phonological, grammatical and typological similarities across Australian languages, many of which are discussed in the chapters in this volume, there is also an enormous amount of differentiation, even among languages that are geographically close. This great range of languages and linguistic structures makes Australia a treasure trove for linguists, and descriptive and analytical work on these languages has proceeded apace over the last 40–50 years. Australian languages have also proven themselves to be particularly interesting for linguistic typology and theory in many respects; this factor has fuelled interest in these languages and is a theme that runs throughout the chapters in this volume. Our knowledge of Australian languages varies enormously across the continent. Some languages ceased to be spoken long before substantial linguistic work could be undertaken. This is the case for the languages of Tasmania, many of the languages of Victoria, and Sydney, for example. Other languages have been the subject of a

The languages of Australia in linguistic research: context and issues 

 5

large amount of linguistic work, by a number of researchers, such as Warlpiri and the Arandic languages in Central Australia. Many languages fall somewhere in between these two extremes, with some amount of language description of varying degrees and quality. A number of works provide overviews of Australian languages and their sociolinguistic, typological and grammatical properties. These include: Blake (1987); Bowern (2013); Dixon (1980, 2002); Evans (2007); Gaby (2008); Walsh (1991); Walsh and Yallop (1993, 2005); Yallop (1982), and also areal sourcebooks and surveys such as McGregor (1988, 2004); Menning and Nash (1981); Thieberger (1993) and Wafer et al. (2009). This volume builds on and complements these works. We have attempted to summarise the developments in Australian linguistics that have taken place since the overviews in Current Trends in Linguistics (Capell 1971, O’Grady 1971, Wurm 1971) and Wurm (1972). To do this, we have focussed on the key areas of historical-comparative linguistics (Koch, Chapter 2); phonetics (Fletcher and Butcher, Chapter 3); (morpho)phonology (Baker, Chapter  4); case, constituency and grammatical relations (Nordlinger, Chapter  5); complex predicates (Bowern, Chapter  6); semantics (Gaby and Singer, Chapter 7); language maintenance and revitalisation (Walsh, Chapter 8); language contact varieties (Meakins, Chapter  9), and Aboriginal English (Eades, Chapter  10). The final chapter completes the picture with a focus on Australian English (Collins, Chapter 11). These chapters provide extensive discussion of the development of research in each of these areas over the last 40–50 years, and reflect many of the key areas of research in the languages and linguistics of Australia during this time. Inevitably, however, there are areas of research that we have not been able to cover for lack of space; § 3 of this chapter attempts to cover these areas in brief and point the reader to some of the relevant literature.

2 History of documentation and study³ The Australian languages first came to the attention of European scholars after the discovery of New South Wales by Captain James Cook in 1770 and the establishment of a British penal colony at Port Jackson (Sydney) in 1788. Cook’s voyages yielded a wordlist of the Guugu Yimidhirr language—including the word kangaroo (Haviland 1974). Wordlists of the Sydney language were collected by a number of officials and naval officers of the first colony. One of these, Lieutenant William Dawes, began a systematic study of the grammar, but his results remained largely unknown until relatively recently (Troy 1992, 1993; http://www.williamdawes.org). The collection of wordlists, most of which used a very unsatisfactory English-based spelling, contin-

3 This section is abbreviated from Koch (2007: 22–24).

6 

 Harold Koch and Rachel Nordlinger

ued for the first century of European settlement. The largest published collection was in E. M. Curr’s (1886–1887) The Australian race, which includes three volumes of lists of up to 120 words for a great many localities of Australia. Many of these were supplied by settlers, policemen, missionaries, etc. For some languages this is the only documentation available. Most of the early attempts to describe the grammar of Australian languages were made by missionaries (Threlkeld 1834, Ridley 1875, Teichelmann and Schürmann 1840, Meyer 1843).⁴ Around the beginning of the twentieth century grammatical sketches were published by the surveyor R. H. Mathews and the physician W. E. Roth (1984), among others. These were typically expressed in terms of the European Traditional Grammar framework, with the result that modern linguists find them unsatisfactory (see Koch 2008). An increase in the amount of documentation as well as in the professional quality of linguistic descriptions followed from: the work of Arthur Capell in the Department of Anthropology at Sydney University from the 1930s; the involvement of linguists of the Summer Institute of Linguistics from the 1950s; funding from the Australian Institute of Aboriginal Studies (AIAS, later AIATSIS) in Canberra from the early 1960s; and the establishing of linguistics departments in Australian universities from the late 1960s and 1970s. Indigenous linguists have played an increasing role in the documentation of their own languages (e.g. Bani and Alpher 1987, Ford and Ober 1991, Henderson and Dobson 1994, Granites and Laughren 2001, Bell 2003, Turpin and Ross 2012). In recent years the Resource Network for Linguistic Diversity has provided extensive support and training for Indigenous communities interested in documenting and revitalising their own languages (see http://www.rnld. org for details). The main avenues of publication of the results of linguistic research have been: Sydney University’s Oceania journal and Oceanic Linguistics monograph series, Monash University’s Linguistic Communications series, AIAS (now Aboriginal Studies Press) in Canberra, Pacific Linguistics at Canberra’s Australian National University,⁵ Dixon and Blake’s five volumes of the Handbook of Australian Languages (Australian National University Press and Oxford University Press), Cambridge University Press, Mouton (de Gruyter), and Lincom Europa. A recent initiative of the last-named publisher is a series Outstanding Grammars from Australia, edited by R. M. W. Dixon, which consists of facsimile copies of thesis-length descriptions of (primarily) Australian languages. A number of dictionaries have been published by the Institute for Aboriginal Development (now IAD Press), Batchelor Press, and Muurrbay Aboriginal Language and Culture Cooperative in Nambucca Heads, N.S.W. For further on the history of documentation of Australian languages the reader is referred to McGregor (2008).

4 A comprehensive history of Australian missions, with especial attention to language matters, is Harris (1990). 5 Since 2012 the Pacific Linguistics series is published by de Gruyter Mouton in Berlin.

The languages of Australia in linguistic research: context and issues 

 7

3 Areas of research not covered in this book It became clear to us at the outset of this project that a single volume would not be enough to include discussion of all of the research that has been undertaken on the languages of Australia in the past 50 years. There are, therefore, a number of strands of research that we have not been able to cover adequately in this volume. In this section we briefly survey some of these, and point the reader to the main sources of further information.

3.1 Historiography The history of Australian linguistics has only recently begun to be seriously discussed, with studies on particular researchers, languages, or linguistic themes. A “Society for the History of Linguistics in the Pacific” has been established and has held three conferences. McGregor (2008) is a recent volume focussing on the history of research in Australian linguistics, including the contributions made by key individuals, and includes chapters on the history of research in areas not covered in this volume (e.g. Adam Kendon on Australian sign languages). A special issue of Language & History 54(2) contains further contributions on Australian linguistic history (Koch 2011, Wafer and Carey 2011).

3.2 Specialised speech registers Complex systems of linguistic etiquette have long been part of Australian Indigenous culture. It was anthropologists who first drew attention to this fact (e.g. Thomson 1935, Stanner 1937, Sansom 1980, Liberman 1985). Avoidance registers used for communicating with in-laws, especially mothers-in-law, have been a special object of interest to linguists (Dixon 1971, Haviland 1979a, 1979b; Rumsey 1982, McConvell 1982, McGregor 1989, Laughren 2001). Hale (1971) describes some features of registers used in the context of initiation; and Nash and Simpson (1981) discuss the practice of name taboos upon death in central Australian communities. Goddard (1992) describes some special speech styles used among Western Desert people including joking speech styles used among certain kin (see also Garde 1996). Harris’s “Yolngu rules of interpersonal communication” (e.g. 1977) have been much cited, especially in educational contexts. A text-book overview of registers is Alpher (1993).

8 

 Harold Koch and Rachel Nordlinger

3.3 Gesture and sign language Early ethnographers made observations about the systems of gestures that are widespread among Aboriginal people. These are typically referred to as “sign languages”, although they are more appropriately described as an auxiliary system of communication. Older sources are summarised in Umiker-Sebeok and Sebeok (1978). The most comprehensive study is Kendon’s (1988) description of the elaborated system used by Warlpiri women especially when under a speech ban after suffering bereavement. Other relevant publications include Kendon (1995), Cooke (1996), Wilkins (1997), and research into the role of gestures in indicating deixis (Haviland 1993, Wilkins 1999). A recent collaborative project involving Indigenous communities from Central Australia and linguists Margaret Carew and Jenny Green has produced the first online dictionary of sign languages in Central Australia, Iltyem-iltyem (http://iltyemiltyem.com/ sign). A recent major publication in the domain of multimodality in Australian languages is Green (2014), a comprehensive multimodal study of story-telling by Arrernte speakers, which involves correlation of linguistic text, gestures, and sand drawings. Green also has a number of related projects and publications underway, as listed on her webpage: http://languages-linguistics.unimelb.edu.au/academic-staff/jennifergreen.

3.4 Song language “Song language” is the term that has come to be used for the special linguistic characteristics of Indigenous songs. This has been a minor area of research throughout the whole period under study. This research mediates between linguistics and ethnomusicology. Much research has involved collaboration between practitioners of the two disciplines (e.g. Barwick et al. 2009, Black and Koch 1983, Dixon and Koch 1996, and the unpublished volumes by the traditional Wangkungurru elder Mick Maclean and the linguist Luise Hercus (Maclean and Hercus, n.d.), with notations provided by the ethnomusicologist Grace Koch). One active scholar combines training and expertise in both disciplines (Turpin 2005). Song texts are the closest equivalent to poetry that occurs in Australian cultures: the first major study of song texts (Strehlow 1971) compared Arrernte songs to the literature of European languages. Note that a number of studies include terms such as “poems” (Dixon and Duwell 1990), “poetry” (von Brandenstein and Thomas 1974, Dixon and Koch 1996), “poetics” (Turpin 2007, Treloyn 2009), or “literature” (Donaldson 1979). One broad type of song is intimately related to traditional landbased mythology and accompanies sacred and secular ceremonies, combined with dance, choreography, body decoration, etc. Another type provides commentary on everyday events. Research in traditional genres is now possible only for a handful of

The languages of Australia in linguistic research: context and issues 

 9

traditional Aboriginal societies (see Walsh 2007). Collections of studies of song language are: Barwick et al. (1995), Clunies Ross et. al. (1987), Kassler and Stubington (1984), Marett and Barwick (2007), and Turpin et al. (2010). A significant website is that of the Wadeye song database (http://sydney.edu.au/arts/indigenous_song/ wadeye/). Much work on the language of song is undertaken within a broader focus on Indigenous music and performance more generally. The National Recording Project for Indigenous Performance in Australia is a current standout in this area, and has already produced a number of recordings and publications (see: http://www.aboriginalartists.com.au/NRP_publications.htm). Other notable works in this field include Berndt (1976), Anderson (1992), Moyle (1986, 1997), Marett (2005), Marett et al. (2013), Magowan (2007), among others.

3.5 Discourse, pragmatics and interaction The collection of texts has long been a part of language description (the so-called “Boasian trilogy” consists of grammar, dictionary and texts (Evans and Dench 2006)), and most grammatical descriptions of Australian languages include a few illustrative texts, usually narratives. A number of linguists have produced volumes consisting primarily of texts, e.g. Holmer and Holmer (1969), von Brandenstein (1970), Schebeck (1974), Glass and Hackett (1979), Heath (1980), Hercus and Sutton (1986), Dixon (1991) and Austin (1997). However, such studies have generally not discussed discourse or pragmatic structure per se. A small number of studies have treated discourse and pragmatic structure as their primary research topic (e.g. Kilham 1977, McGregor 1987, Swartz 1991, Carroll 1995, Glass 1997 [MA thesis version 1980], Rose 2001, Kim et al. 2001, Mushin 2005, Simpson 2007, Mushin and Baker 2008) or have included discourse structure as part of a grammatical description (e.g. Goddard 1985, McGregor 1990, Patz 2002, Wilkins 1989). Studies that consider the literary and aesthetic qualities of texts include Napaljarri and Cataldi (1994), Cataldi (1996) and Klapproth (2004); and some recent work has looked at narrative in children’s discourse (Bavin 2000, Disbray 2008). For a recent discussion of the research on discourse in Australian languages, see Baker and Mushin (2008). For references to studies on the discourse functions of grammatical features see Nordlinger (this volume, chapter 5) and especially the papers in Mushin and Baker (2008). For prosodic aspects of discourse, see references in Fletcher and Butcher (this volume, chapter 3). Studies of interaction have been even fewer than those of narrative discourse. Pioneering studies were by Liberman (1982, 1985) and Garde (2002, 2013). Recent years have seen an increase in interest in this area (see for example, Moses 2009, Blythe 2009 and the papers in Mushin and Gardner 2010).

10 

 Harold Koch and Rachel Nordlinger

3.6 Language and education There have been numerous studies on language issues in the educational system. Early studies discussed problems of literacy and classroom communication for children who spoke languages other than English. Representative collections are Brumby and Vaszolyi (1977), Eagleson, Kaldor and Malcolm (1982). Many short, language-related articles appeared in teacher journals such as The Aboriginal Child at School. Some international publications are Malcolm (1979), Malcolm (1982), Christie and Harris (1985). The introduction of a bilingual education policy (see, for example O’Grady and Hale 1975) in the Northern Territory led to numerous publications. Overviews include Murtagh (1982), Gale (1990), Hartman and Henderson (1994), Devlin (1995), Harris (1995), Hoogenraad (2001); see also the discussion in Laughren (2000). Particularly important are the discussions by Indigenous educators and linguists such as Raymattja Marika (1999) and Eve Fesl (1993). The large number of vernacular books produced by the Northern Territory bilingual education program, supported by linguists and teacher-linguists, are in the process of being digitised and put on the web by a project at Charles Darwin University, Living Archive of Aboriginal Languages (http://laal.cdu.edu.au/). The subsequent dismantling of the policy also attracted the attention (and concern) of linguists (e.g. Devlin 2009, 2011, Simpson, Caffery and McConvell 2009). The complex interaction between home language and the educational system for Indigenous children has been a topic of recent research by a number of researchers, including Jane Simpson and Gillian Wigglesworth (e.g. Simpson and Wigglesworth 2008, Wigglesworth et al. 2011) and Denise Angelo and colleagues (see, for example, the papers listed at http://www.languageperspectives. org.au/).

3.7 Child language acquisition Until recently not much research has been pursued on children’s acquisition of Australian languages. The principal studies before current studies were Bavin and Shopen’s work on Warlpiri (e.g. Bavin 1990, Bavin and Shopen 1991). Warlpiri is also the subject of a much-cited study of Baby Talk (Laughren 1984). The Aboriginal Child Language Acquisition Projects (phases 1 and 2), led by Jane Simpson and Gillian Wigglesworth, have been concerned with analysing the development of children’s linguistic abilities in the interface between schooling and multilingual home communities (see Simpson and Wigglesworth 2008 and http://languages-linguistics. unimelb.edu.au/current-projects/acla2). More recently, a large-scale project has begun on the acquisition of Murrinh-Patha, with involvement from a team of researchers from the University of Melbourne (http://languages-linguistics.unimelb.edu.au/ projects/lamp).

The languages of Australia in linguistic research: context and issues 

 11

3.8 Language and law Problems of intercultural communication affecting speakers of Indigenous languages in their interaction with the criminal justice system have long been discussed in the literature (Strehlow 1936, Kriewaldt 1960–1962, Nash 1979, Liberman 1981, Walsh 1994, Cooke 1995). One of the issues has concerned the kind of English known or produced by Indigenous people caught up in legal processes. A celebrated case involving the linguist T. G. H. Strehlow in the late 1950s is described in Inglis (1961, see also Eades 2013). Highly significant and influential in the area of Aboriginal English in the legal system has been the work of Diana Eades (see, for example, Eades 1992, 1995, 2008, 2013, and this volume, chapter 10). Linguists have been involved in documenting the relation of Aboriginal groups to their traditional lands, as part of the exercise of granting legal rights in land to groups of Australia’s Indigenous people. From the first settlement on Australia by the British in 1788 until the Aboriginal Land Rights (Northern Territory) Act of 1976—for the Northern Territory—and the Native Title Act of 1993—for Australia as a whole, there was no official recognition of Indigenous ownership of the land. Studies concerned with language issues in relation to Aboriginal language claims and the Native Title Act include Neate (1981), Nash (1984), Koch (1985, 1990, 1991), Simpson (1994), Walsh (1999) and Henderson and Nash (2002).

3.9 Placenames Research for land claims has fostered a renewed appreciation of the importance of territorial affiliation for Aboriginal identity, and has documented thousands of sites. Placename research has become the subject of recent interest by linguists, anthropologists, historians, and especially state Geographical Names Boards, as Indigenous placenames are being increasingly recognised in the public sphere. A number of conferences on Indigenous placenames have been held since the 1990s. Many of the papers from these conferences appear in three collections: Hercus, Hodges, and Simpson (2002), Koch and Hercus (2009), and Clark, Hercus, and Kostanski (in press).

3.10 Kinship terminology The study of kinship terminology has traditionally been the preserve of anthropologists more than of linguists (e.g. Elkin 1938–1940). A major study of kin classification systems by the linguistic anthropologist Scheffler (1978) uses methods from linguistic semantics as well as anthropological kinship theory. A significant contribution by linguists is Heath, Merlan and Rumsey (1982). Grammatical uses of kinship categories are described in this book, as well as in studies such as Hale (1966), Hercus and White

12 

 Harold Koch and Rachel Nordlinger

(1973), Dench (1982) and Evans (2000, 2006). Diachronic aspects of Australian kinship are explored in McConvell, Dousset and Powell (2002) and McConvell, Keen and Hendery (2013). The “Austkin” project has in recent years been compiling a database of Australian kinship terminology (Dousset et al. 2010; http://www.austkin.net). Another topic related to kinship is the study of social category terminology—the majority of Australian social groups have societal divisions called moieties, sections, semi-moieties, or subsections (depending on the number and structure of the social categores they distinguish). These have long been the subject of study by anthropologists (e.g. Radcliffe-Brown 1930–1931). More recently, partially in reaction to von Brandenstein (1982), McConvell (1985) initiated the study of the diachrony of subsections. A current research project nicknamed “Austkin II” aims to map the Australian social category terms and reconstruct their prehistoric spread (see McConvell and Dousset 2012).

4 (Selected) internet resources Aboriginal and Torres Strait Islander Languages [ed. David Nathan]: http://www.dnathan.com/VL/ Anggarrgoon: Australian languages on the web [ed. Claire Bowern]: http://anggarrgoon.wordpress.com/ ASEDA: The Aboriginal Studies Electronic Data Archive: http://aseda.aiatsis.gov.au/asedaDisclaimer.php AUSTLANG: Australian Indigenous Languages Database: http://austlang.aiatsis.gov.au/disclaimer.php Australian languages [ed. David Nash]: http://www.anu.edu.au/linguistics/nash/aust/ Language policies for Australian languages: http://www.anu.edu.au/linguistics/nash/aust/policy.html Living Archive of Aboriginal Languages: a repository of bilingual education materials: http://laal.cdu.edu.au/ OZBIB: A linguistic bibliography of Aboriginal Australia and the Torres Strait Islands: http://ozbib.aiatsis.gov.au/ozbibDisclaimer.php Wadeye song database: http://sydney.edu.au/arts/indigenous_song/wadeye/

References Alpher, Barry 1993 Out-of-the-ordinary ways of using a language. In: Michael Walsh and Colin Yallop (eds.), Language and culture in Aboriginal Australia, 97–106. Canberra: Aboriginal Studies Press. Anderson, Gregory D. 1992 Murlarra: A clan song series from Central Arnhem Land. Sydney: University of Sydney PhD thesis.

The languages of Australia in linguistic research: context and issues 

 13

Austin, Peter 1997 Texts in the Mantharta languages, Western Australia. Tokyo: Institute for the Study of the Languages and Cultures of Asia and Africa, Tokyo University of Foreign Studies. Baker, Brett and Ilana Mushin 2008 Discourse and grammar in Australian languages. In: Ilana Mushin and Brett Baker (eds.), Discourse and grammar in Australian languages, 1–23. Amsterdam: John Benjamins. Bani, Ephraim and Barry Alpher 1987 Garka a ipika: masculine and feminine grammatical gender in Kala Lagaw Ya. Australian Journal of Linguistics 7(2): 189–201. Barwick, Linda, Joe Blythe, Lysbeth Ford, Allan Marett, Michael Walsh, and Nicholas Reid 2009 The Wadeye song database. Sydney: University of Sydney and Wadeye Aboriginal Languages Centre Wadeye Knowledge Centre. Barwick, Linda, Allan Marett, and Guy Tunstill (eds.) 1995 The essence of singing and the substance of song: recent responses to the Aboriginal performing arts and other essays in honour of Catherine Ellis. Sydney: The University of Sydney. Bavin, Edith 1990 Locative terms and Warlpiri acquisition. Journal of Child Language 17: 43–66. Bavin, Edith L. 2000 Ellipsis in Warlpiri children’s narratives: an analysis of frog stories. Linguistics 38(3): 569–588. Bavin, Edith L. and Tim Shopen 1991 Warlpiri in the 80s: An overview of research into language variation and child language. In: Suzanne Romaine (ed.), Language in Australia, 104–117. Cambridge: Cambridge University Press. Bell, Jeanie 2003 A sketch grammar of the Badjala language of Gari (Fraser Island). Melbourne: University of Melbourne MA Thesis. Berndt, Ronald M. 1976 Love songs of Arnhem Land. West Melbourne: Nelson. Black, Paul and Grace Koch 1983 Koko-Bera island style music. Aboriginal History 7(2): 157–172. Blake, Barry 1987 Australian Aboriginal grammar. London: Croom Helm. Blythe, Joseph 2009 Doing referring in Murrinh-Patha conversation. Sydney: University of Sydney PhD thesis. Bowern, Claire 2012 The riddle of Tasmanian languages. Proceedings of The Royal Society of Biological Sciences 279(1747): 4590–4595. Bowern, Claire 2013 Australian languages. Oxford Bibliographies Online. Oxford University Press. http://www.oxfordbibliographies.com/browse?module_0=obo-9780199772810. Brumby, Ed and Eric Vaszolyi (eds.) 1977 Language problems in Aboriginal education. Perth: Aboriginal Teacher Education Program, Mount Lawley College of Advanced Education. Butlin, Noel G. 1983 Our original aggression: Aboriginal populations of south eastern Australia 1788 –1850. Sydney: Allen and Unwin. Capell, A. 1971 History of research in Australian and Tasmanian languages. In: Thomas A. Sebeok (ed.), Current trends in linguistics Volume 8: Linguistics in Oceania, 661–720. The Hague: Mouton. Carroll, Peter John 1995 The old people told us: Verbal art in Western Arnhem Land. Brisbane: University of Queensland PhD thesis. Cataldi, Lee 1996 The end of the dreaming? Understandings of history in a Warlpiri narrative of the Coniston massacre. Overland no. 144: 44–47. Christie, Michael and Stephen Harris 1985 Communication breakdown in the Aboriginal classroom. In: John B. Pride (ed.), Cross-cultural encounters: Communication and mis-communication, 81–90. Melbourne: River Seine. Clark, Ian, Luise Hercus and Laura Kostanski (eds.) in press Indigenous and minority placenames: Australian and international perspectives. (Aboriginal History Monographs) Canberra: ANU Press. Clunies Ross, Margaret, Tamsin Donaldson and Stephen A. Wild (eds.) 1987 Songs of Aboriginal Australia (Oceania Monographs 32) Sydney: University of Sydney.

14 

 Harold Koch and Rachel Nordlinger

Cooke, Michael 1995 Interpreting in a cross-cultural cross-examination: An Aboriginal case study. International Journal of the Sociology of Language 113: 99–111. Cooke, Michael 1996 Yolngu signing: Gestures or language? In: Michael Cooke (ed.), Aboriginal languages in contemporary contexts: Yolngu Matha at Galiwin’ku, 51–64. Northern Territory: Batchelor College. Curr, Edward M. 1886–1887 The Australian race: its origin, languages, customs, place of landing in Australia, and the routes by which it spread itself over that continent. 4 vols. Melbourne: J. Ferres, Government Printer. Dench, Alan 1982 Kin terms and pronouns of the Panyjima language of Northwest Western Australia. Anthropological Forum 5(1): 105–200. Devlin, Brian 1995 The evaluation of bilingual programs in the Northern Territory, Australia, 1973–1993. International Journal of the Sociology of Language 113: 23–36. Devlin, Brian 2009 Bilingual education in the Northern Territory and the continuing debate over its effectiveness and value. Paper presented at an AIATSIS Research Forum, Canberra, June 2009. Available at: http://www.abc.net.au/4corners/special_eds/20090914/language/docs/ Devlin_paper.pdf. Devlin, Brian 2011 The status and future of bilingual education for remote indigenous students in the Northern Territory. Australian Review of Applied Linguistics 34(3): 260–279. Disbray, Samantha 2008 More than one way to catch a frog: A study of children’s discourse in an Australian contact language. Melbourne: University of Melbourne PhD thesis. Dixon, R. M. W. 1971 A method of semantic description. In: Danny D. Steinberg and Leon A. Jakobovitz (eds.), Semantics: An interdisciplinary reader in philosophy, linguistics and psychology, 436–471. Cambridge: Cambridge University Press. Dixon, R. M. W. 1980 The languages of Australia. Cambridge: Cambridge University Press. Dixon, R. M. W. 1991 Words of our country. Stories, place names and vocabulary in Yidiny, the Aboriginal language of the Cairns-Yarrabah region. St. Lucia, Qld: University of Queensland Press. Dixon, R. M. W. 2002 The Australian languages: Their nature and development. Cambridge: Cambridge University Press. Dixon, R. M. W. and Barry J. Blake (eds.) 1979 Handbook of Australian languages Vol. 1. Canberra: Australian National University Press. Dixon, R. M. W. and Barry J. Blake (eds.) 1981 Handbook of Australian languages Vol. 2. Canberra: Australian National University Press. Dixon, R. M. W. and Barry J. Blake (eds.) 1983 Handbook of Australian languages Vol. 3. Canberra: Australian National University Press. Dixon, R. M. W. and Barry J. Blake (eds.) 1991 Handbook of Australian languages Vol. 4: The Aboriginal language of Melbourne and other grammatical sketches. Melbourne: Oxford University Press. Dixon, R. M. W. and Barry J. Blake (eds.) 2000 Handbook of Australian languages Vol. 5. Melbourne: Oxford University Press. Dixon, R. M. W. and Martin Duwell (eds.) 1990 The Honey-Ant men’s love song and other Aboriginal song poems. St Lucia, Qld: University of Queensland Press. Dixon, R. M. W. and Grace Koch 1996 Dyirbal song poetry: The oral literature of an Australian rainforest people. St Lucia, Qld: Queensland University Press. Donaldson, Tamsin 1979 Translating oral literature from Australian Aboriginal languages into English. Aboriginal History 3: 66–83. Dousset, Laurent, Rachel Hendery, Claire Bowern, Harold Koch and Patrick McConvell 2010 Developing a database for Australian Indigenous kinship terminology: The AustKin project. Australian Aboriginal Studies 2010/No. 1: 42–56.

The languages of Australia in linguistic research: context and issues 

 15

Eades, Diana 1992 Aboriginal English and the law: Communicating with Aboriginal English speaking clients: A handbook for legal practitioners. Brisbane: Queensland Law Society. Eades, Diana (ed.) 1995 Language in evidence: Issues confronting Aboriginal and multicultural Australia. Sydney: University of New South Wales Press. Eades, Diana 2008 Courtroom talk and neocolonial control. (Language, Power and Social Process 22) Berlin: Mouton de Gruyter. Eades, Diana 2013 Aboriginal ways of using English. Canberra: Aboriginal Studies Press. Eagleson, Robert D., Susan Kaldor and Ian G. Malcolm (eds.) 1982 English and the Aboriginal Child. Canberra: Curriculum Development Centre. Elkin, A. P. 1938–1940 Kinship in South Australia. Oceania 8(4): 419–452; 9(1): 41–78; 10(2): 198–234; 10(3): 295–349; 10(4): 369–389. Evans, Nicholas 2000 Kinship verbs. In: Petra M. Vogel and Bernard Comrie (eds.), Approaches to the typology of word classes, 103–172. Berlin: Mouton de Gruyter. Evans, Nicholas 2006 Dyadic constructions. In: Keith Brown (ed.), Encyclopedia of language & linguistics, 24–28. Oxford: Elsevier. Evans, Nicholas 2007 Warramurrungunji Undone: Australian languages in the 51st millennium. In: Matthias Brenzinger (ed.), Language Diversity Endangered, 342–373. Berlin: Mouton de Gruyter. Evans, Nicholas and Alan Dench 2006 Introduction: Catching language. In: Felix K. Ameka, Alan Dench and Nicholas Evans (eds.), Catching language: The standing challenge of grammar writing, 1–39. Berlin: Mouton de Gruyter. Fesl, Eve Mumewa D. 1993 Conned! St Lucia: University of Queensland Press. Ford, Kevin and Dana Ober 1991 A sketch of Kalaw Kawaw Ya. In: Suzanne Romaine (ed.), Language in Australia, 118–142. Cambridge: Cambridge University Press. Gaby, Alice 2008 Rebuilding Australia’s linguistic profile: Recent developments in research on Australian Aboriginal languages. Language and Linguistics Compass 2(1): 211–233. Garde, Murray 1996 ‘Saying nothing’: the language of joking relationships in Aboriginal Australia. Darwin: Northern Territory University Graduate Diploma thesis. Garde, Murray 2002 Social deixis in Bininj Kun-wok conversation. St Lucia: University of Queensland PhD thesis. Garde, Murray 2013 Culture, Interaction and Person Reference in an Australian Language: An ethnography of Bininj Gunwok communication (Culture and Language Use 13). Amsterdam: John Benjamins. Gale, Mary-Anne 1990 A review of bilingual education in Aboriginal Australia. Australian Review of Applied Linguistics 13(1): 40–80. Glass, Amee 1997 Cohesion in Ngaanyatjarra discourse. (SIL-AAIB Occasional Papers 4) Darwin: SIL. Glass, Amee and Dorothy Hackett 1979 Ngaanyatjarra texts. Canberra: Australian Institute of Aboriginal Studies. Goddard, Cliff 1985 A grammar of Yankunytjatjara. Alice Springs NT: Institute for Aboriginal Development. Goddard, Cliff 1992 Traditional Yankunytjatjara ways of speaking: A semantic perspective. Australian Journal of Linguistics 12(1): 93–122. Granites, Robert Japanangka and Mary Laughren 2001 Semantic contrasts in Warlpiri verbal morphology: A Warlpiri’s verbal view. In: Jane Simpson, David Nash, Mary Laughren and Barry Alpher (eds.), Forty years on: Ken Hale and Australian languages. (Pacific Linguistics 512), 151–159. Canberra: Australian National University. Green, Jennifer 2014 Drawn from the ground: Sound, sign and inscription in Central Australian sand stories. Cambridge: Cambridge University Press.

16 

 Harold Koch and Rachel Nordlinger

Hale, Kenneth L. 1966 Kinship reflections in syntax: Some Australian languages. Word 22: 318–324. Hale, Kenneth L. 1971 A note on a Walbiri tradition of antonymy. In Danny D. Steinberg and Leon A. Jakobovitz (eds.), Semantics: An interdisciplinary reader in philosophy, linguistics and psychology, 472–482. Cambridge: Cambridge University Press. Harris, John 1990 One blood: 200 years of Aboriginal encounter with Christianity. Sutherland, NSW/ Claremont CA: Albatross Books. Harris, Stephen 1977 Yolngu rules of interpersonal communication. Developing Education 4(5): 23–30. Harris, Stephen 1995 Evolution of bilingual education theory in Northern Territory Aboriginal schools. International Journal of the Sociology of Language 113: 7–21. Hartman, Deborah and John Henderson (eds.) 1994 Aboriginal languages in education. Alice Springs: IAD Press. Haviland, John B. 1974 A last look at Cook’s Guugu Yimidhirr word list. Oceania 44: 216–232. Haviland, John B. 1979a How to talk to your brother-in-law in Guugu Yimidhirr. In: Timothy Shopen (ed.), Languages and their speakers, 161–239. Cambridge: Winthrop Publishers. Haviland, John B. 1979b Guugu Yimidhirr brother-in-law language. Language in Society 8: 365–393. Haviland, John B. 1993 Anchoring, iconicity, and orientation in Gugu Yimithirr pointing gestures. Journal of Linguistic Anthropology 3(1): 3–45. Heath, Jeffrey 1980 Nunggubuyu myths and ethnographic texts. Canberra: Australian Institute of Aboriginal Studies. Heath, Jeffrey, Francesca Merlan and Alan Rumsey (eds.) 1982 The languages of kinship in Aboriginal Australia. (Oceania Linguistic Monograph 24) Sydney: University of Sydney. Henderson, John and Veronica Dobson 1994 Eastern and Central Arrernte to English dictionary. Alice Springs N.T.: Institute for Aboriginal Development. Henderson, John and David Nash (eds.) 2002 Language in native title (Native Title Research Series). Canberra: Aboriginal Studies Press. Hercus, Luise, Flavia Hodges and Jane Simpson (eds.) 2002 The land is a map: Placenames of Indigenous origin in Australia. Canberra: Pandanus Books/Pacific Linguistics. Hercus, Luise and Peter Sutton (eds.) 1986 This is what happened: Historical narratives by Aborigines. Canberra: Australian Institute of Aboriginal Studies. Hercus, Luise A. and Isobel M. White 1973 Perception of kinship structure reflected in the Adnjamathanha pronouns. In: Bernhard Schebeck, Luise A. Hercus and Isobel M. White, Papers in Australian Linguistics No.6. (Pacific Linguistics A-36), 49–72. Canberra: Australian National University. Holmer, Nils M. and Vanja E. Holmer 1969 Stories from two native tribes of eastern Australia. Lund: Carl Bloms Boktryckeri. Hoogenraad, Robert 2001 Critical reflections on the history of bilingual education in Central Australia. In: Jane Simpson, David Nash, Mary Laughren and Barry Alpher (eds), Forty years on: Ken Hale and Australian languages, 123–150. Canberra: Australian National University. Inglis, Ken S. 1961 The Stuart case. Parkville: Melbourne University Press. [reprinted 2002 by Black Inc, Melbourne, Victoria]. Kassler, Jamie C. and Jill Stubington (eds.) 1984 Problems and solutions: Occasional essays in musicology presented to Alice M. Moyle. Sydney: Hale and Iremonger. Kendon, Adam 1988 Sign languages of Aboriginal Australia: Cultural, semiotic and communicative perspectives. Cambridge: Cambridge University Press. Kendon, Adam 1995 Sociality, social interaction and sign language in Aboriginal Australia. In: Brenda Farnell (ed.), Human action signs in cultural context: The visible and invisible in movement and dance, 112–123. London: Scarecrow Press.

The languages of Australia in linguistic research: context and issues 

 17

Kilham, Christine A. 1977 Thematic organization of Wik-Munkan discourse. (Pacific Linguistics B-52) Canberra: Australian National University. Kim, Myung-Hee, Lesley Stirling and Nicholas Evans 2001 Thematic organisation of discourse and referential choices in Australian languages. Discourse and Cognition 8: 1–21. Klapproth, Danièle M. 2004 Narrative as Social Practice: Anglo-Western and Australian Aboriginal Oral Traditions. (Language, Power and Social Process 13). Berlin: Mouton de Gruyter. Koch, Harold 1985 Nonstandard English in an Aboriginal land claim. In John B. Pride (ed.), Crosscultural encounters: Communication and mis-communication, 176–195. Melbourne: River Seine. Koch, Harold 1990 Language and communication in Aboriginal land claim hearings. Australian Review of Applied Linguistics Series S, No. 5: 1–47. Koch, Harold 1991 Language and communication in Aboriginal land claim hearings. In: Suzanne Romaine (ed.), Language in Australia, 94–103. Cambridge: Cambridge University Press. Koch, Harold 2007 An overview of Australian traditional languages. In: Gerhard Leitner and Ian G. Malcolm (eds.), The Habitat of Australia’s Aboriginal Languages: Past, Present, and Future (Trends in Linguistics. Studies and Monographs 179), 23–56. Berlin: Mouton de Gruyter. Koch, Harold 2008 R. H. Mathews’ schema for the description of Australian languages. In: William B. McGregor (ed.), Encountering Aboriginal languages: Studies in the history of Australian linguistics (Pacific Linguistics 591), 179–218. Canberra: Australian National University. Koch, Harold 2011 George Augustus Robinson and the documentation of languages of southeastern New South Wales. Language & History 54(2): 140–163. Koch, Harold and Luise Hercus (eds.) 2009 Aboriginal placenames: Naming and re-naming the Australian landscape. (Aboriginal History Monograph 19) Canberra: ANU E Press. Kriewaldt, Justice M. C. 1960–1962 The application of the criminal law to the Aborigines of the Northern Territory of Australia. University of Western Australia Law Review 5: 1–50. Laughren, Mary 1984 Warlpiri baby talk. Australian Journal of Linguistics 4: 73–88. Laughren, Mary 2000 Australian Aboriginal languages: Their contemporary status and functions. In: R. M. W. Dixon and Barry J. Blake (eds.), The handbook of Australian languages Vol. 5, 1–32. Oxford: Oxford University Press. Laughren, Mary 2001 What Warlpiri ‘avoidance’ registers do with grammar. In: Jane Simpson, David Nash, Mary Laughren, Peter Austin and Barry Alpher (eds.), Forty years on: Ken Hale and Australian languages¸ 199–225. Canberra: Australian National University. Liberman, Kenneth 1981 Understanding Aborigines in Australian courts of law. Human Organization 40: 247–255. Liberman, Kenneth 1982 The organisation of talk in Aboriginal community decision-making. Anthropological Forum 5: 38–53. Liberman, Kenneth 1985 Understanding interaction in central Australia: An ethnomethodological study of Australian Aboriginal people. London: Routledge and Kegan Paul. Maclean, Mick Irinyili and Luise Hercus n.d. The Emu history from Wangkangurru-Arabana country. Unpublished ms. Canberra: Australian National University. Maclean, Mick Irinyili and Luise Hercus n.d. The journey of the Seven Sisters Through the Lake Eyre region. Unpublished ms. Canberra: Australian National University. Maclean, Mick Irinyili and Luise Hercus n.d. The story of Wurru the Crane. Unpublished ms. Canberra: Australian National University. Magowan, Fiona 2007 Melodies of mourning: music and emotion in Northern Australia. Oxford: James Currey Publishers. Malcolm, Ian G. 1979 The West Australian Aboriginal child and classroom interaction: A sociolinguistic approach. Journal of Pragmatics 3: 305–320.

18 

 Harold Koch and Rachel Nordlinger

Malcolm, Ian G. 1982 Speech events in the Aboriginal classroom. International Journal of the Sociology of Language 36: 115–134. Marett, Allan 2005 Songs, Dreamings, and Ghosts: The Wangga of North Australia. Middleton: Wesleyan University Press. Marett, Allan and Linda Barwick (eds.) 2007 Studies in Aboriginal song. [Special issue]. Australian Aboriginal Studies 2007/No. 2. Marett, Allan, Linda Barwick and Lysbeth Ford 2013 For the Sake of a Song: Wangga Songmen and Their Repertories. Sydney: Sydney University Press. Marika, Raymattja 1999 The 1998 Wentworth Lecture. Australian Aboriginal Studies 1999/No. 1: 3–9. Marmion, Doug, Kazuko Obata and Jakelin Troy 2014 Community, identity, wellbeing: The report of the Second National Indigenous Languages Survey. Canberra: AIATSIS. McConvell, Patrick 1982 Neutralisation and degrees of respect in Gurindji. In: Jeffrey Heath, Francesca Merlan and Alan Rumsey (eds.), The languages of kinship in Aboriginal Australia (Oceania Linguistic Monographs 24), 86–106. Sydney: University of Sydney. McConvell, Patrick 1985 The origin of subsections in northern Australia. Oceania 56: 1–33. McConvell, Patrick and Laurent Dousset 2012 Tracking the dynamics of kinship and social category terms with Austkin II. Proceedings of the EACL 2012 Joint Workshop of LINGVIS & UNCLH, 98–107. Avignon, France, April 23 – 24 2012. McConvell, Patrick, Laurent Dousset and Fiona Powell (eds.) 2002 Kinship and change in Aboriginal Australia. [Special issue]. Anthropological Forum 12(2). McConvell, Patrick, Ian Keen and Rachel Hendery (eds.) 2013 Kinship systems: Change and reconstruction. Salt Lake City: University of Utah Press. McConvell, Patrick and Nicholas Thieberger 2001 State of Indigenous languages in Australia – 2001. (Australia State of the Environment Second Technical Paper Series [Natural and Cultural Heritage]) Canberra: Department of the Environment and Heritage. http://www.ea.gov.au/soe/ techpapers/index.html McGregor, William 1987 The structure of Gooniyandi narratives. Australian Aboriginal Studies 1987/ No. 2: 20–28. McGregor, William 1988 Handbook of Kimberley languages. Volume 1. General information. (Pacific Linguistics C-105) Canberra: Australian National University. McGregor, William 1989 Gooniyandi mother-in-law “language”: Dialect, register, and/or code? In: Ulrich Ammon (ed.), Status and function of languages and language varieties, 630–656. Berlin / New York: Mouton de Gruyter. McGregor, William 1990 A functional grammar of Gooniyandi. Amsterdam: John Benjamins. McGregor, William B. 2004 The languages of the Kimberley, Western Australia. London / New York: RoutledgeCurzon. McGregor, William B. (ed.) 2008 Encountering Aboriginal languages: Studies in the history of Australian linguistics. (Pacific Linguistics 591). Canberra: Australian National University. Menning, Kathy and David Nash 1981 Sourcebook for Central Australian languages. Alice Springs: Institute for Aboriginal Development. Meyer, Heinrich August Edward 1843. Vocabulary of the language spoken by the aborigines of the southern and eastern portions of the settled districts of South Australia., preceded by a grammar. Adelaide: James Allen. Moses, Karen 2009 How do dinosaurs hug in the Kimberley? The use of questions by Aboriginal caregivers and children in a Walmajarri community. Melbourne: University of Melbourne PhD thesis. Moyle, Richard 1986 Alyawarra Music: Songs and Society in a Central Australian Community. Canberra: Australian Institute of Aboriginal Studies.

The languages of Australia in linguistic research: context and issues 

 19

Moyle, Richard 1997 Balgo: The musical life of a desert community. Perth, WA: Callaway International Resource Centre for Music Education, University of Western Australia. Murtagh, Edward J. 1982 Creole and English used as languages of instruction in bilingual education with Aboriginal Australians: Some research findings. International Journal of the Sociology of Language 36: 15–34. Mushin, Ilana 2005 Word order pragmatics and narrative functions in Garrwa. Australian Journal of Linguistics 25(2): 253–273. Mushin, Ilana and Rod Gardner (eds.) 2010 Studies in Australian Indigenous conversation. [Special issue]. Australian Journal of Linguistics 30(4). Mushin, Ilana and Brett Baker (eds.) 2008 Discourse and grammar in Australian languages. (Studies in Language Companion Series 104). Amsterdam; Philadelphia: John Benjamins. Napaljarri, Peggy Rockman and Lee Cataldi 1994 Warlpiri dreamings and histories: Yimikirli (The Sacred Literature Series). San Francisco, London, Pymble: Harper Collins. Nash, David 1979 Foreigners in their own land: Aborigines in Court. Legal Service Bulletin 4: 105–107. Nash, David 1984 Linguistics and land rights in the Northern Territory. In: Graham R. McKay and Bruce A. Sommer (eds.), Further Applications of Linguistics to Australian Aboriginal Contexts (Occasional Papers, No. 8, Applied Linguistics Association of Australia), 34–46. Parkville: University of Melbourne. Nash, David and Jane Simpson 1981 “No-name” in Central Australia. In: Carrie S. Masek, Roberta A. Hendrick and Mary Frances Miller (eds.), Papers from the Parasession on Language and Behavior, 165–177. Chicago: Chicago Linguistic Society. Neate, Graeme J. 1981 Legal language across cultures: Finding the traditional Aboriginal owners of land. Federal Law Review 12: 187–211. Nettle, Daniel and Suzanne Romaine 2000 Vanishing voices: The extinction of the world’s languages. Oxford: Oxford University Press. O’Grady, Geoffrey N. 1971 Lexicographic research in Aboriginal Australia. In: Thomas A. Sebeok (ed.), Current trends in linguistics Volume 8: Linguistics in Oceania, 779–803. The Hague: Mouton. O’Grady, Geoff and Ken Hale 1975 Recommendations concerning bilingual education in the Northern Territory. Canberra: The Parliament of the Commonwealth of Australia. Patz, Elizabeth 2002 A grammar of the Kuku Yalanji language of North Queensland. (Pacific Linguistics 527). Canberra: Australian National University. Piper, Nick 2013 A sketch of Meryam Mer. (Outstanding Grammars from Australia 14) Munich: Lincom Europa. Radcliffe-Brown, A. R. 1930–1931 The social organization of Australian tribes. Oceania 1: 34–63, 206–246, 323–341, 426–456. Ridley, William 1875 Kámilarói, and other Australian languages. Sydney: Thomas Richards, Government Printer. Rose, David 2001 The Western Desert code: An Australian cryptogrammar. (Pacific Linguistics 513). Canberra: Australian National University. Roth, W. E. 1984 The Queensland Aborigines. 3 vols. Facsimile edition. Victoria Park, Western Australia: Hesperian Park. Rumsey, Alan. 1982 Gun-gunma: An Australian Aboriginal avoidance language and its social functions. In: Jeffrey Heath, Francesca Merlan and Alan Rumsey (eds.), The languages of kinship in Aboriginal Australia (Oceania Linguistic Monographs 24), 160–181. Sydney: University of Sydney. Rumsey, Alan 1993, 2005 Language and territoriality in Aboriginal Australia. In: Michael Walsh and Colin Yallop (eds.), Language and culture in Aboriginal Australia, 191–206. Canberra: Aboriginal Studies Press.

20 

 Harold Koch and Rachel Nordlinger

Sansom, Basil 1980 The camp at Wallaby Cross: Aboriginal fringe dwellers in Darwin. Canberra: Australian Institute of Aboriginal Studies. Schebeck, Bernhard 1974 Texts on the social system of the Atynyamathanha people with grammatical notes. (Pacific Linguistics D-21). Canberra: Australian National University. Scheffler, Harold 1978 Australian kin classification. Cambridge: Cambridge University Press. Simpson, Jane 1994 Confidentiality of linguistic material: The case of Aboriginal land claims. In: John Gibbons (ed.), Language and the law, 428–439. London and New York: Longman. Simpson, Jane 2007 Expressing pragmatic constraints on word order. In: Annie Zaenen, Jane Simpson, Tracy Holloway King, Jane Grimshaw, Joan Maling and Chris Manning (eds.), Architectures, rules and preferences: Variations on themes by Joan W. Bresnan, 403–427. Stanford: CSLI Publications. Simpson, Jane, Jo Caffery and Patrick McConvell 2009 Gaps in Australia’s Indigenous language policy: Dismantling bilingual education in the Northern Territory. (AIATSIS Research Discussion Paper No. 24). Canberra: Australian Institute of Aboriginal and Torres Strait Islander Studies. Simpson, Jane and Gillian Wigglesworth (eds.) 2008. Children’s language and multilingualism: Indigenous language use at home and school. London / New York: Continuum. Stanner, W. E. H. 1937 Aboriginal modes of address and reference in the Northwest of the Northern Territory. Oceania 7: 300–315. Strehlow, T. G. H. 1936 Notes on native evidence. Oceania 6: 323–335. Strehlow, T. G. H. 1971 Songs of Central Australia. Sydney: Angus and Robertson. Swartz, Stephen M. 1991 Constraints on zero anaphora and word order in Warlpiri narrative text. (SIL-AAIB Occasional Papers No.1). Darwin, Northern Territory: Summer Institute of Linguistics. Teichelmann, C. G. and C. W. Schürmann 1840 Outlines of a grammar, vocabulary, and phraseology of the Aboriginal language of South Australia, spoken by the natives in and for some distance around Adelaide. Adelaide: Thomas and Co. [Reprinted in facsimile, 1982, by Tjintu Books, Largs Bay, Adelaide] Thieberger, Nicholas 1993 Handbook of Western Australian languages south of the Kimberley region. (Pacific Linguistics C-124). Canberra: Australian National University. Thomson, Donald 1935 The joking relationship and organized obscenity in North Queensland. American Anthropologist 37: 460–490. Threlkeld, Lancelot 1834 An Australian grammar comprehending the principles and natural rules of the language as spoken by the Aborigines in the vicinity of Hunter’s River, Lake Macquarie, &c. New South Wales. Sydney: Stephens and Stokes, Herald Office. Treloyn, Sally 2009 Half way: Appreciating the poetics of northern Kimberley song. Musicology Australia 31: 41–62. Troy, Jakelin 1992 The Sydney language notebooks and responses to language contact in early colonial NSW. Australian Journal of Linguistics 12: 145–170. Troy, Jakelin 1993 Language contact in early colonial New South Wales. In: Michael Walsh and Colin Yallop (eds.), Language and culture in Aboriginal Australia, 33–50. Canberra: Aboriginal Studies Press. Turpin, Myfany M. 2005 Form and meaning of Akwelye: A Kaytetye women’s song series from Central Australia. Sydney: University of Sydney PhD thesis. Turpin, Myfany 2007 The poetics of central Australian song. Australian Aboriginal Studies 2007/ No. 2: 100–115. Turpin, Myfany and Alison Ross 2012 Kaytetye to English dictionary. Alice Springs: IAD Press. Turpin, Myfany, Tonya Stebbins and Stephen Morey (eds.) 2010 The language of song. [Special issue]. Australian Journal of Linguistics 30(1). Umiker-Sebeok, D. Jean and Thomas A. Sebeok (eds.) 1978 Aboriginal sign languages of the Americas and Australia, Vol. 2, Part III: Australia. 257–440. New York: Plenum Press.

The languages of Australia in linguistic research: context and issues 

 21

von Brandenstein, C. G. 1970 Narratives from the north-west of Western Australia in the Ngarluma and Jindjiparndi Languages. 3 volumes (with a 7” record). Canberra: Australian Institute of Aboriginal Studies. von Brandenstein, C. G. 1982 The names and substance of the Australian subsection system. Chicago / London: University of Chicago Press. von Brandenstein, C. G. and A. P. Thomas 1974 Taruru: Aboriginal song poetry from the Pilbara. Adelaide: Rigby. Wafer, Jim and Hilary M. Carey 2011 Waiting for Biraban: Lancelot Threlkeld and the ‘Chibcha phenomenon’ in Australian missionary linguistics. Language & History 54(2): 112–139. Wafer, Jim, Amanda Lissarrague and Jean Harkins 2009 A handbook of Aboriginal languages of New South Wales and the Australian Capital Territory. Nambucca Heads NSW: Muurrbay Aboriginal Language and Culture Co-operative. Walsh, Michael 1991 Overview of indigenous languages of Australia. In: Suzanne Romaine (ed.), Language in Australia, 27–48. Cambridge: Cambridge University Press. Walsh, Michael 1994 Interactional styles in the courtroom: An example from Northern Australia. In: John Gibbons. (ed.), Language and the law, 217–233. London / New York: Longman. Walsh, Michael 1999 Interpreting for the transcript: Problems in recording land claim proceedings in northern Australia. Forensic Linguistics 6: 161–195. Walsh, Michael 2007 Australian Aboriginal song language: So many questions, so little to work with. Australian Aboriginal Studies 2007/No. 2: 128–144. Walsh, Michael and Colin Yallop (eds.) 1993, 2005 Language and culture in Aboriginal Australia. Canberra: Aboriginal Studies Press. Wigglesworth, Gillian, Jane Simpson and Deborah Loakes 2011 NAPLAN Language assessments for indigenous children in remote communities: Issues and problems. Australian Review of Applied Linguistics 34(3): 320–343. Wilkins, David P. 1989 Mparntwe Arrernte: Studies in the Structure and Semantics of Grammar. Canberra: Australian National University PhD thesis. Wilkins, David 1997 Handsigns and hyperpolysemy: Exploring the cultural foundations of semantic association. In: Darrell Tryon and Michael Walsh (eds.) Boundary rider: Essays in honour of Geoffrey O’Grady (Pacific Linguistics C-136), 413–444. Canberra: Australian National University. Wilkins, David 1999 Spatial deixis in Arrernte speech and gesture: On the analysis of a species of composite signal as used by a Central Australian Aboriginal group. In: Elizabeth André, Massimo Poesio, and Hannes Rieser (eds.), Proceedings of the Workshop on Deixis, Demonstration and Deictic Belief in Multimedia Contexts), 31–45. Utrecht: Eleventh European Summer School in Logic, Language and Information. Wurm, Stephen A. 1971 Classifications of Australian languages, including Tasmanian. In: Thomas A. Sebeok (ed.), Current trends in linguistics Volume 8: Linguistics in Oceania, 721–778. The Hague: Mouton. Wurm, S. A. 1972. Languages of Australia and Tasmania. The Hague: Mouton. Yallop, Colin. 1982 Australian Aboriginal languages. (The Language Library). London: André Deutsch.

Harold Koch

2 Historical relations among the Australian languages: genetic classification and contact-based diffusion 1 Introduction 1.1 Themes and scope The historical-comparative linguistics of Australian languages is a relatively young field of study.¹ The field has involved a number of premature judgments, changes of position, and disputes about methodology. Especially contestable have been persistent claims about the supposed exceptionality of Australian languages. Another issue has been how typological differences are reconcilable with genetic relationship. Discussions of genetic relationship among the languages have been complicated by the undisputed role of language contact. A major issue since the 1960s has concerned the existence of and justification for a single language family, Pama-Nyungan (PN²), that is claimed to cover the major part of the continent.³ This survey covers predominantly the developments from about 1970 to the present: nevertheless some of the earlier background is given in the next section and elsewhere when necessary.

1.2 Brief overview of the literature In this section I present an overview, largely in chronological order, of the main references on Australian comparative linguistics. Particular issues will be discussed further in later sections. The Vienna-based linguist Wilhelm Schmidt (1919a, 1919b) attempted a classification of Australian languages on the basis of the disparate early documents available early in the twentieth century, which he comprehensively collated. Although his classification methods have not been found to be satisfactory by modern standards

1 I thank Claire Bowern, Simon Greenhill, Robert Mailhammer, Rachel Nordlinger, and an anonymous reviewer for helpful comments on earlier drafts of this chapter. I take full responsibility, however, for the current content. 2 I use the following abbreviations for languages: PN Pama-Nyungan, nPN non-Pama-Nyungan, pPN ProtoPama-Nyungan, and pA Proto Australian. 3 The locations of the Australian language families and the subgroups of Pama-Nyungan that are mentioned in this chapter are shown on Map 3, which was prepared with the help of Claire Bowern.

24 

 Harold Koch

(see Koch 2004a: 18–25), he recognised the genetic unity of all the languages of the southern part of the continent, but claimed that (groups of) the northern languages — in which he included the Arandic languages of central Australia and the languages of Cape York Peninsula — were not related to the southern family or to one another. Arthur Capell from the late 1920s added considerably to the documentation of languages in northern Australia and his overview publications introduced typological parameters such as prefixing morphology prevalent among northern languages vs. purely suffixing morphology elsewhere. Capell also proposed that all the languages of the Australian mainland are likely to be related genetically, and called attention to a number of lexical and grammatical roots that were found in the whole continent, his “Common Australian” (CA) elements. Capell was also concerned about reconciling the typological divergence of Australian languages with their supposed genetic unity, and presented hypothetical scenarios whereby the prefixing structure could be derived historically from a rigidification of preferred word orders of subject and object pronouns with respect to the verb and subsequent morphologisation of the elements (Capell 1956). The 1960s saw a comprehensive survey of languages still spoken and the classification of all Australian languages using the then relatively new technique of lexicostatistics. The results were published in O’Grady, Wurm and Hale (1966), O’Grady, Voegelin and Voegelin (1966)⁴, and slightly revised in Wurm (1971, 1972). In spite of the provisional nature of the OWH classification, and its updated form in Walsh and Wurm (1981), its terminology has largely prevailed ever since, being promulgated in reference works such as Lewis (2009). Another current of research emerged from R. M. W. Dixon, including: studies of the north Queensland rainforest languages (Dixon 1970a), “Proto-Australian laminals” (Dixon 1970c), and a book (Dixon 1980), which long remained the standard authority on Australian languages. This major work combines continent-wide typological comparison with tentative reconstructions of “Proto-Australian” (pA). There is now a consensus, however, that Dixon’s “pA” reconstructions are too dependent on data from PN languages⁵ to be attributed to the ancestor of the nPN languages as well. The 1970s and 1980s saw the description of many more languages of the north and centre of Australia and historical analysis of particular (groups of) languages, especially in northern Queensland and Western Australia, such as Alpher (1972), Sutton (1976), Crowley (1976), Black (1980), Austin (1981a). The historical picture of Australian languages presented in Baldi (1990) is somewhat fractured, with O’Grady (1990d) pursuing the traditional PN view, Dixon (1990) expressing doubts about its genetic status, while Blake (1990b) and Heath (1990)

4 I use the abbreviation OWH to refer to the (identical) classification presented in both sources, and OVV for the O’Grady, Voegelin and Voegelin reference when quoting from it. 5 Dixon has consistently rejected the idea of Pama-Nyungan as a genetic group.

Historical relations among the Australian languages 

 25

demonstrated that Dixon’s (1980) “pA” reconstructions of pronouns and verb inflection respectively fail to account for the facts of nPN languages. Blake (1988, 1990a, 1990b) argued from the evidence of pronoun cognates for a “Northern” or nPN ancestor separate from pPN. Blake (1988, 1990a, 1990b) and Evans (1988) claimed that innovations in the pronoun system, case forms, and phonology define PN as a subgroup of Australian languages. Meanwhile several languages were reclassified from PN to nPN and vice versa. A 1991 conference on Australian archaeology and linguistics produced a number of studies on historical-comparative linguistics in relation to other disciplines dealing with Australian prehistory (McConvell and Evans 1997a). Volumes honouring Geoff O’Grady (Tryon and Walsh 1997) and Ken Hale (Simpson et al. 2001) contain further papers on historical-comparative themes. From the mid 1990s Dixon’s research took on a new direction (Dixon 1997, 2001, 2002). No longer attempting to reconstruct pA or to establish a family tree for Australian languages, Dixon claimed that the predominant mechanism of change was the constant diffusion of traits between adjacent languages, which produced situations of linguistic “equilibrium” and which would have largely obliterated the evidence for the original “punctuation” event which entailed the diversification of the languages. This “Punctuated Equilibrium” model has provoked plentiful reactions (see § 2.9). Meanwhile attempts to establish language families and subgroups by the application of the standard comparative method have continued. The major results are presented in Evans (2003b, on nPN languages) and Bowern and Koch (2004, mostly on PN). Other studies of particular groups are Bowern (2004a, 2004b) on Nyulnyulan, Harvey (2008a) on Mirndi, McGregor and Rumsey (2009) on Worrorran, van Egmond (2012) on Anindhilyakwa — all nPN families — and Laffan (2003) on Waka-Kabi, Barrett (2005) on Maric, Weber (2009) on Marrngu — all PN subgroups. Bowern, Evans and Miceli (2008) includes a number of studies on Australian classification and morphological change/reconstruction. Meanwhile Harvey (2009, 2012) has re-examined the genetic relations of two assumed families. The role of language contact and the study of areal distribution of linguistic features have always been prominent in Australian linguistics. Borrowing plays a large role in the 50% equilibrium model developed by Dixon (see §  3.2.1). Heath’s (1978, 1981) studies of linguistic contact in Arnhem Land have been much cited. McConvell (2010) provides an overview of contact studies.

1.3 Exceptionality A recurrent theme in Australian historical-comparative linguistics has been the supposed exceptionality of the languages. Linguists have differed as to whether special methods are therefore required to deal with this situation, and have devoted much energy to methods other than the classic comparative method.

26 

 Harold Koch

1.3.1 Uniformity of phonology Capell (1956: 4) asserted that “Australian languages … are remarkably homogeneous in phonetic structure and even in phonemic structure”. Voegelin et al. (1963) demonstrated this from the indices of diversity of phonological systems they calculated within samples of Australian languages.⁶ Australianists have usually assumed this uniformity to be inherited, rather than resulting from convergence, as mentioned as a possibility by OVV (1966: 16), with most languages having undergone few sound changes. Linguists have therefore devoted little energy to working out sound correspondences and establishing subgroups on the basis of diachronic phonological innovations — except in a few regions where the phonology has been radically altered.

1.3.2 Lexical diversity OVV (1966: 10) claim that the percentage of cognate vocabulary between dialects of the same language, languages of the same family, and different languages within the same phylum is lower in Australia than in other parts of the world. Because of this, “cognates are insufficient or sparse or simply lacking to connect the different phylic families and to show that by and large their relatively undifferentiated structure represents common retention from the proto phylum” and “[i]t is virtually impossible to calculate regular correspondences among the cognates which support the Australian phylum hypothesis, due to the sparseness of cognates …” (OVV 1966: 16). They used the concept of language phylum to describe situations where language families share sufficient material to suggest a (distant) genetic relation which cannot be demonstrated by regular sound correspondence occurring in a substantial number of cognates, and thus classify Australian languages into an “Australian phylum”. At the lowest level of classification, they find “family-like languages”, whose most peripheral dialects may share as little as 45% of their basic vocabulary, but which “show very similar, if not almost identical, structures” (Wurm 1972: 31). In between they find “phylic language families”, characterised by the degree of structural similarity expected of language families but with lexical scores as low as within phyla rather than families elsewhere.

6 For more recent generalisations on Australian phonemic inventories and phonotactics, see Fletcher and Butcher (this volume, chapter 3) and Hamilton (1996).

Historical relations among the Australian languages 

 27

1.3.3 Rate of lexical replacement Dixon (1970a: 653; 1972: 331) claims as a feature of Australian languages “their apparently high rate of vocabulary replacement”. This “rather radical view” (Wurm 1972: 36) has been questioned on the basis that where early records have been compared to modern sources, e.g. Parnkalla (Barngarla) from 1844 and 1960 (OVV 1966: 26) and Guugu Yimidhirr from 1770 and the 1970s (Haviland 1974), the replacement does not appear to be excessive; cf. Black (1997: 58–60) and Alpher and Nash (1999: 9–11).

1.3.4 Amount and nature of borrowing Claims have been made that the amount of borrowing is higher in Australian languages than elsewhere, that basic vocabulary is borrowed as readily as other vocabulary, that grammatical words including pronouns and bound morphology are easily borrowed, and that structural features such as clitic pronouns (and bipartite verbs) are commonly copied without their lexical content. Also claimed to be unique are the motivations for borrowing, including a taboo on uttering words resembling the names of the recently deceased and the supposed high degree of exogamy and/or multilingualism (cf. § 3).

1.3.5 Applicability of historical methods The supposed special nature of Australian languages has led to the question whether traditional methods of historical linguistics are applicable and to proposals for new models of change and/or new methods of pursuing comparative linguistics. Doubts have been expressed about the regularity of sound change, the usefulness of the comparative method, and the relevance of family tree representations. Capell (1956: 83) questioned the utility of the comparative method, since “words are either fairly obviously cognate as between languages, or equally obviously non-cognate. There has not appeared the same necessity of establishing sound laws to prove connections.” Similarly Boretzky (1984) observed, from Ken Hale’s comparative Arandic wordlists, that words for the same concept are either identical or totally different in phonological form and concluded that this is a feature of “exotic” languages, unlike IndoEuropean languages, and that the comparative method is therefore not applicable to such languages. Hoenigswald (1990: 377–378) countered that this is not a problem with the method, but simply indicates that the lects compared are too close to allow much reconstruction; rather the “yield” of the method is not as great where there are not many cognates of differing phonological shape. More recently, Dixon (2002: 699) has forcefully pronounced on the distinctive nature of Australian languages and the consequent need for special methods.

28 

 Harold Koch

The Australian linguistic area poses problems of investigation and analysis unlike those found anywhere else in the world. The established methods of historical and comparative linguistics, which can be applied so successfully elsewhere, have limited appropriateness in Australia. The special nature of the Australian situation must be acknowledged for real progress to be made in describing the nature of this linguistic situation, and for an understanding to be attained concerning the nature of interrelations between its constituent languages. (Dixon 2002: 699)

The usual task in historical linguistics of distinguishing similarities due to common inheritance from those due to contact effects (as well as those resulting from independent parallel developments) is aggravated for Australia by claims regarding the exceptionality of its linguistic situation. Hence an overview of Australian comparative linguistics needs to devote attention to issues of method. Moreover, since matters of descent versus diffusion are so intertwined, it is necessary to discuss issues of contact-induced change (borrowing, diffusion) alongside those of genetic classification and reconstruction. Accordingly, this survey interposes a section on contact (§ 3) between the sections devoted to classification methodology (§ 2) and substantive issues of classification (§ 4).⁷

2 Genetic classification: methodological issues 2.1 Different bases for classification Koch (2004a) documents the diversity of assumptions and methods used in classifying Australian languages. OWH’s classification has led to some confusion, in that the groupings it established on the basis of lexicostatistical percentages of shared vocabulary introduced terminology (families, subgroup) that was subsequently used as if it denoted entities established by the comparative method. In fact, they presented the classification as a tentative genealogical classification, with the hope that details would be confirmed or disconfirmed by further study according to the comparative method. The comparative method includes criteria for establishing genetic relations at two levels, family and subgroup.⁸ Campbell and Poser (2008 passim) show that the requirements for recognising a language family relationship have always been: shared basic vocabulary, shared grammatical markers (i.e. morphemes), and systematic phonological correspondences. Establishing a subgroup, on the other hand, presupposes a (tentative) reconstruction of a proto-language, then demonstrates, for a

7 Space limitations preclude a discussion of details of the classification of particular languages, the reconstruction of particular proto-languages, and the particular changes that have been discovered. 8 There can furthermore be subgroups within subgroups, and families can be treated as subgroups of higher-level families if this can be justified by common innovations from a higher level.

Historical relations among the Australian languages 

 29

group of languages, that they have undergone a set of identical innovations which are better interpreted as having taken place once in an intermediate common ancestor than having occurred independently or by borrowing. Campbell and Poser (2008: 155) suggest that a premature acceptance of a genetic relation between all the Australian languages has led to a confusion of criteria for establishing families with those for establishing subgroups: families are called subgroups, and common innovations are sought for their proof. “[I]f there is no legitimate Proto-Australian, or if Proto-Australian existed but cannot be reconstructed, then it is in effect impossible to determine what a shared innovation, a departure from Proto-Australian traits, would be … When the assumed genetic unity of all Australian languages is given up (or at least not made the starting assumption), the standard methods … for establishing families among Australian languages can prevail.” Miceli (2004) shows how Australianists have sought to justify PN as a subgroup rather than using the criteria appropriate for establishing a language family. Dixon (1980) attacked the idea of PN as a genetic entity on the basis that no common innovations had been demonstrated — which Bowern (2006: 251) suggests is akin to requiring that the Indo-European family be justified on the basis of innovations from Proto-Nostratic.⁹ Nevertheless Evans (1988) and Blake (1988) took up Dixon’s challenge, arguing that common innovations could indeed be found in the distinctive case allomorphy (Ergative *-ngku ~ *lu and Locative *-ngka ~ *-la), Blake’s set of distinctive personal pronouns, and Evans’ initial laminalisation, whereby nPN word-initial *t and *n are reflected in PN languages by *th and *nh. The same ambivalence about criteria seems to lie behind the doubts expressed by Green and Nordlinger (2004) about the relative absence of “innovations” differentiating the Mirndi “subgroup” from other nPN languages; Mirndi is now accepted as a language family (Harvey 2008a), which needs no further justification. Also confusing is Dixon’s (2002) use of the label “low-level genetic subgroups” for the small language families he recognises, on the grounds that “some of these subgroups may eventually be shown to be linked together in higherlevel groupings” (p. xxiv).

2.2 Use of typology Typological description has always figured prominently in the discussion of Australian languages. Capell (1937: 54) attempted “to indicate a general grouping based on grammatical form”. Capell (1956, 1962) classified languages according to whether they used prefixes or just suffixes, whether they used noun classes or not, and if so whether there were just two classes or many, whether they used bound

9 If Proto-Australian were to be established, however, it would then be appropriate to justify the subgroup status of PN by the criterion of common innovation.

30 

 Harold Koch

person-markers (for subjects and/or objects) or not, and if so, whether these were normally attached to the verb, a “catalyst” (or auxiliary), or the “nucleus” (first word or constituent) of the clause. Capell’s prefixing languages largely correlate with the nPN languages of the lexicostatistical classification (see § 2.3). Wurm’s (1972: 112–151) catalogue of language families, groups, and subgroups includes comments on the structural features of many sets of languages. Dixon’s two overviews (1980, 2002) combine descriptions of typological parameters with attempts at reconstructing their history. Most linguists who have pursued historical studies have been concerned with the practical description and synchronic typology of Australian languages as well. Typology is mainly of interest in this overview insofar as it impacts on genetic classification. In a few instances typological considerations have interfered with genealogical classification.¹⁰

2.3 Lexicostatistical classification A comprehensive classification resulted from an initiative of Carl Voegelin of Indiana University, survey work in the years 1959–1961 by Kenneth Hale, Geoffrey O’Grady, and Stephen Wurm, and collaboration into the mid-1960s by these three researchers plus Carl and Florence Voegelin. The results were published as O’Grady, Wurm, and Hale (1966), O’Grady, Voegelin, and Voegelin (1966), and with some revisions in Wurm (1971, 1972). The results were based on “cognate densities derived from comparison of the hundred items of a Swadesh-type lexical list in pairs of named communalects” (OVV: 23). A hierarchy of terms was based on the percentage of shared vocabulary in adjacent lects or chains of lects, as indicated in Table 1. O’Grady (1960) shows how the Swadesh list was adapted for Australia; O’Grady and Klokeid (1969) demonstrate the classification of twelve communalects from south-western Australia. Hale (1961) shows how the lexical percentages were converted to displays on a map. Wurm (1972: 109) says that “[t]hough the basis of their classification was admittedly lexicostatistical in nature, typological criteria had been taken into consideration in arriving at the results, and had been regarded as decisive in doubtful cases”. O’Grady and Hale (2004) admit to the mistake of “falling into a sort of ‘typological trap’” in classing Yanyuwa, a prefixing PN language, as nPN and the non-prefixing Tangkic languages as a subgroup of the PN family.

10 A study by Reesink et al. (2009) offers the prospect that the computational analysis of statistically defined clusters of structural features can provide fruitful hypotheses regarding the phylogenetics of languages over a large area. For example, their methods find signals of the distinction between Papuan and Australian languages, and between Pama-Nyungan and non-Pama-Nyungan languages (with a few exceptions).

Historical relations among the Australian languages 

 31

Table 1: Criteria for lexicostatistic classification cognate density

classification

71+% 51–70% 26–50% 16–25% under 15%

different dialects of the same language different languages of the same subgroup different subgroups of the same group different groups of the same family different families of the same phylum

The OWH classification was intended to be a tentative phylogenetic schema, to be confirmed or refined by the later application of the comparative method. Its authors made “a plea for the future consideration of types of evidence additional to that of lexicostatistics, in order that a balanced perspective of Australian historical linguistics might be achieved” (O’Grady 1966: 71). Alterations to the classification have been made (Wurm 1971, 1972, Walsh and Wurm 1981) — typically on the basis of advice from linguists with expertise on particular languages — but not necessarily using the same lexicostatistical criteria, with the result that later versions represent a kind of hybrid classification (see Koch 2004a: 34–48). The inadequacies of lexicostatistical classification are well known (cf. Campbell 2004: 201–210). (a) It relies on only one kind of evidence, omitting the evidence of phonological and morphological changes. (b) It does not distinguish between features shared by retention from a remote ancestor and those resulting from common innovation in an intermediate ancestor. (c) It tacitly relies on the “simplifying assumption” (Black 2007a: 78) that the rate of lexical replacement is constant across languages, so that the relative proportion of shared vocabulary inversely correlates with the relative time depth of shared ancestors. (d) The results can be skewed by undetected loanwords: Australianists typically make no effort to eliminate loanwords from the calculations.¹¹ Nevertheless Australianists have continued to use some form of this method — if only for making judgements about which source documents relate to the same lect and/or language. I suggest several reasons for this practice. First, as Breen (2011: 236) claims, “lexicostatistics … is still the only method for any language for which we have no information other than a small vocabulary dominated by nouns” — such as that in Curr (1886–1887).¹² OVV (1966: 23) noted: “Curr effectively predetermined the scope of any widely applicable list in 1886 by publishing a 120-word list obtained from several

11 A notable exception is Alpher and Nash (1999), who painstakingly set out criteria assessing the effects of undetected loanwords and present lexicostatistical findings from which putative loanwards have been eliminated. Black (1997, 2007b) furthermore demonstrates how anomalous lexicostatistical percentages can actually be used to identify cases of heavy borrowing. 12 Even wordlists, however, sometimes contain a few grammatical forms (e.g. ‘I’, ‘you’, ‘who’) that help with the classification of lects (cf. Koch 2011).

32 

 Harold Koch

hundred communalects”. Second, there is a belief that sound changes cannot help to classify languages that have undergone little phonological divergence. A third reason may be a reluctance to posit grammatical innovations, whether in the belief that grammar can be borrowed, or because there is no agreed reconstruction from which to identify innovations. Australian practice of lexicostatistics differs according to the size of wordlists compared, the specific meanings included in the wordlist¹³, whether only “basic” vocabulary is used, whether any grammatical words (e.g. pronouns, demonstratives, interrogatives) are included, and whether an attempt is made to eliminate loanwords. Furthermore, adaptations have been made to the basic lexicostatistical approach, in particular by Breen.

2.4 Breen’s “modified lexicostatistical method” Breen (1990: 154–163; 2011) has proposed a refinement of the lexicostatistical method, which he hopes will help with classification. His lexical comparisons use a 250-word list first used in Breen (1971). He sorts vocabulary into the following semantic domains: verbs, body parts, human classification, inanimate nature, material culture, and fauna. He compares percentages of shared vocabulary in different semantic domains across a number of languages — contiguous and not, closely related genetically and not. In a few test cases the genetic relations are already known on the basis of non-lexical considerations (e.g. similar pronoun systems and phonological correspondences). The novelty of his approach consists in drawing inferences (regarding the genetic relationships and perhaps contact history) from the relationship between the percentages of shared nominals vs. verbs (or possibly more specific semantic domains). He assumes that as genetically related lects diverge, the inventory of verbs changes more slowly than that of nouns (which include words expressing notions that are adjectival in English), which are lost and replaced through borrowing more readily than verbs. The greater the genetic divergence, the lower the ratio of nouns to verbs. Contiguity is important: if a language borrows from related lects this tends to raise the percentage of nouns to a figure closer to that of verbs. Unrelated or distantly related languages that are in a borrowing relationship will have a higher ratio of noun to verb cognates. When Breen applies his approach to some uncertain cases (Kalkatungu and Yalarnnga, Karnic), however, the results are not decisive; his results for Kaytetye and the rest of Arandic, moreover, are at variance with conclusions reached by the criterion of common innovations (Koch 2004b).

13 Alpher and Nash (1999: 53–56) indicate the slightly different basic wordlists used by themselves and other scholars—O’Grady and Klokeid (1969) and (in unpublished wordlists) Ken Hale and Paul Black.

Historical relations among the Australian languages 

 33

2.5 Blake’s “genetic distance” Blake (1988: 3) introduced a concept of genetic distance “based on the quantity of shared roots: the fewer the cognates the greater the genetic distance. It stands as a measure of genetic relatedness irrespective of the locus of innovation … a group of languages can be genetically close relative to others by common innovation or common retention.” Accordingly, “[w]e can often see that languages are genetically close, but it is more difficult to determine whether this proximity is due to innovation or retention” (Blake 1988: 46). For example, Blake and Reid (1998: 10) claim for four southeastern languages [my emphasis]: “On the basis of these shared function forms, which are not found in the Western Victorian language nor in Wathawurrung nor in the Central Victorian Language, we can say that Bunganditj and the Warrnambool language are relatively closely related … such a statement does not imply that they should be subgrouped. The shared forms may be relic forms. With the shared vocabulary it is clear that some of the forms listed above are relics.” Likewise Breen and Blake (2007: 70–71) claim that Kalkatungu and Yalarnnga, two adjacent languages of western Queensland, are “relatively close genetically”: they “are more similar to one another than either is to any other language”, even without counting what they share from diffusion, yet cannot be described as a subgroup, since there is no clear evidence of common innovations. This conceptualisation of genetic proximity, based on a phenetic measure of the relative amount of shared material, which may be shared by common retention, is in fact not genetic proximity as defined by subgrouping procedures used in the comparative method, according to which the most closely related languages are those which share the most recent ancestral proto-language, posited on the basis that they participated in a common set of innovations that took place before innovations that distinguish them. The relative amount of shared features is irrelevant.

2.6 Computational approaches Dobson and Black (1979) describe multidimensional scaling, a technique for representing spatially the relationships between objects, so that the distances between points correspond to measures of the (dis)similarity of the objects. They illustrate the process by applying it to the lexicostatistical percentages of ten lects from the rainforest area near Cairns, Queensland, taken from Dixon (1970a). Nash (2002) applies the technique to the relationships among basic vocabulary lists from 26 wordlists from the south-eastern part of Western Australia. In both cases the configuration which displayed the degrees of difference mirrored the geographical positions of the lects. McMahon and McMahon (2006: 155) re-present Nash’s data, analysed using computational phylogenetic methods, including a Neighbor-Joining tree and Neighbor-Net network, as an illustration of linguistic data in which the effects of contact have largely obliterated tree-like structure.

34 

 Harold Koch

Crowley and Bowern (2010: 151–159) includes a section on “subgrouping computational methods from biology”, describing the concepts behind several of the programs used for historical linguistics. In NeighborNet, differences in character sets (compared features) are computed into distance (i.e. difference) matrices, and the resulting distances between languages are represented in the form of (trees or) network diagrams, where the length of branches is proportional to the degree of difference (reflecting amount of change). They present an illustration (p. 158) of the network resulting from applying the NeighborNet procedures to some 600 lexical, phonological, and morphological character sets of 13 languages of the Karnic subgroup of Pama-Nyungan. McGregor and Rumsey (2009: 10–15) applied NeighborNet procedures to the Worrorran languages of North Kimberley using 105 basic meanings in 13 Worrorran and eight other neighbouring languages. Their results confirmed both the distinctiveness of the Worrorran group and its division into three subgroups. Crowley and Bowern (2010: 159) also briefly describe Bayesian phylogenetics, which is based on replacements and calculates the most probable set of trees, given the data and a model of evolution. This approach is applied in Bowern and Atkinson (2012) to the question of higher-level subgrouping within the PN family (see § 4.6.4). Jones and Laffan (2008) present a mathematical and computational method, which aims to be objective and replicable, for calculating the percentage of divergence between wordlists whose provenance suggests they could possibly be from the same lect. Their illustration applies the method to five wordlists from the greater Sydney region which were used in producing a consolidated description of the Darkinyung language (cf. Jones 2008). This procedure is important for deciding, for the many languages known only through old wordlists, which wordlists represent the same language. With similar but more ambitious aims, Bowern (2012) applied computational phylogenetic methods to the wordlist data on Tasmanian languages presented in Plomley (1976), to ascertain how many languages are represented in the data and how many families they belong to. Her results supported the existence of twelve languages in five separate clusters that have no demonstrable family connections between them.

2.7 Approaches to subgrouping Australianists have used a number of approaches to defining subgroups. Bowern (2005: 347) characterises the “way that Australianists tend to model and provide evidence for genetic relations”: they first present cognate density matrices, then give tables of pronouns and inflections, doing “cognate judgements by eyeballing for similarities”, then discuss some typological phonological features, etc., with no lexical reconstruction. Typology — the observation of obvious parallelisms in phonology and grammar — plays an informal role in the initial stages of subgrouping. The lexicostatistical approach explicitly bases subgrouping on the relative amount of shared (and presumably cognate) vocabulary from a particular list of meanings.

Historical relations among the Australian languages 

 35

What Dixon (2002) requires for establishing his “subgroups” — which are really small genetic families — are essentially the criteria for positing a language family:¹⁴ “For a subgrouping to be validated it is … necessary for a good portion of the proto-language to be reconstructed, together with the systematic changes which have been involved in the development of the modern languages” (Dixon 2002: 659). According to the comparative method, a subgroup is established only by demonstrating a unique set of common innovations, with the point of departure being the proto-language ancestral to the whole family to which these languages belong. The aim of Bowern and Koch (2004), as explained in the introduction and guidelines, was to apply this approach to a number of Australian subgroups. This endeavour, however, is hampered to some extent by the incompleteness of the reconstructed protolanguages available for Australian languages, with the result that unique features of a group of languages may be taken as de facto innovations. The practical value of common innovations in establishing subgroups was challenged by Black (2004), who examined the phonological changes in the so-called Paman languages. He found that, although similar changes had taken place in many of the languages, the details of the changes — their conditioning environments and chronology relative to other changes — rarely matched across the languages sufficiently to be used as unambiguous evidence for particular subgroups. He found the few morphological innovations used to support Alpher’s (1972) “Southwest Paman” subgroup to be equally problematic. His suggested remedy (p. 266) is “to not focus on just the evidence of shared innovations, but to consider all evidence available”, including lexicostatistical scores, since “lexicostatistical evidence of some form can always be found”. Also sceptical of the value of common innovation is Dench (2001), who discusses innovations in phonology, morphophonology, and case-marking patterns in seventeen languages of the Pilbara (usually assigned to the Ngayarta, Kanyara and Mantharta subgroups). He finds (p. 130) that “[n]one of the shared innovations described … can be considered, conclusively, to be innovations arising in a single ancestor” rather than being due to contact, and concludes from this study that “building low level subgroups may be impossible in some cases [in Australia]” (p. 131). Bowern (2009) mentions conflicting shared innovations within the Karnic subgroup, where Pitta Pitta and Wangka-Yutjuru share certain innovations with ArabanaWangkangurru but are grouped by other innovations with the remaining Karnic languages to the exclusion of Arabana-Wangkangurru. These facts imply “that something other than a family tree may be more useful to model the history of these languages” (p. 347). It seems clear that the methods of dialectology need to be applied for at least some Australian linguistic situations. It is well-known that the isoglosses of dialects typically criss-cross one another in ways that cannot be modelled by tree diagrams.

14 This follows from the fact that he does not reconstruct any high-level proto-language.

36 

 Harold Koch

This patterning is sometimes found between languages as well, especially in bi- or multilingual societies (Hock and Joseph 1996: 417), and can make subgrouping unclear where the languages are closely related. Haspelmath (2004: 212) noted with respect to Dench (2001): “It seems clear that the Pilbara languages are all rather closely related genealogically, and the difficulties of subgrouping of closely related groups of languages are well known (recall the difficulties of subgrouping of the Romance languages)”. Australianists might apply the approach of Toulmin (2009) to the Indo-Aryan dialect continuum. He described linguistic innovations in terms of propagation events (cf. Croft 2000), and plotted the range of each such event. Where a relative chronology of innovations can be established, it is possible to perform linguistic reconstruction even if the ranges of the separate innovations do not cluster in a way that can be modelled by a tree diagram.¹⁵ This approach could be applied to the data discussed in Black (2004). Table 2 presents some of the phonological changes from a selection of his “Southwest Pama” data, with the changes given in their inferred chronological order. The shading of cells denotes that the change has occurred in the language of its column. Here it can be seen, for example, that final vowel loss is shared by all four of these languages even though they have been differentiated by the effects of earlier sound changes. ¹⁶¹⁷ Table 2: Southwest Pama innovations Yir Yoront

Koko-Bera

Kok-Nar

Kurrtjar

V>Ø/#CVC_C# i/u>e/o16 Final velarisation17 V>Ø/_#

2.8 Use of grammatical evidence for classification Australianists have been criticised for resorting to morphological comparison before establishing sound correspondences. Campbell and Poser (2008: 159–161) describe the “Oz paradox” in the following terms. Linguists don’t trust vocabulary because they cannot be certain that words have not been borrowed. Loanwords cannot be iden-

15 See also François (in press) for models of language change without tree diagrams. 16 This vowel lowering is conditioned by the nature of the vowel in the next syllable. 17 This process adds a velar consonant after word-final rr or l; the added consonant is k or ng in KokNar but the velar fricative gh in Kurrtjar.

Historical relations among the Australian languages 

 37

tified because the historical reconstruction has not been done which would show which words do not fit the regular sound correspondences (which may not help where languages have similar phonology). They therefore investigate morphological similarities (pronouns, cases, verb affixes) for evidence of relationships. Superficial similarities in sounds and functions, even paradigms, are called “cognates” and described in terms of “reconstruction” — even though their cognation has not been demonstrated by regular sound correspondences and similarities due to borrowing or accident have therefore not been eliminated. Without this demonstration of cognation outsiders are justified in remaining unconvinced, especially since Australianists readily accept instances of the borrowing of bound morphology such as case suffixes and pronominal prefixes. Miceli (2008) describes this unorthodox practice, which she labels “inspectional¹⁸ morphological reconstruction”, as in part a consequence of “an ‘unusual’ pattern of similarity” in the data of Australian languages, where “the vast majority of potential cognates … are identical or near-identical” (p. 212), and the proportion of cognates is lower than expected from experience in other languages of the world. Miceli also suggests (p. 215) that Australian practice has been influenced by Meillet’s (1925, 1967) emphasis on the evidence of “particular facts” as legimitising reliance on “shared aberrancies” and “irregular morphology”.¹⁹ Examples of where grammatical data was used to classify languages before the comparative phonology was fully worked out include: the PN family (Blake 1988; Evans 1988 before Alpher’s 2004 phonology); the linking of Yanyuwa into the Warluwarric subgroup of PN on the evidence of Blake’s (1988) personal pronouns and case suffixes and other grammar by Brammall (1991) before Carew’s (1993) phonological reconstruction; Chadwick’s (1984, 1997) proposal of a discontinuous Mirndi family on the basis of case-gender markers and personal pronouns, strengthened by Green and Nordlinger’s (2004) reconstruction of bound subject and object markers in the verb and case-gender affixes on nominals, before Harvey’s (2008a) lexical comparison — which finds only about 100 lexical cognates reconstructible to Proto-Mirndi; Ian Green’s (2003) establishing of a Southern Daly family based primarily on the evidence of congruent verbal paradigms between Murrinh-Patha and Ngan’gityemerri, even in the face of low cognate densities. In defence of Australianists’ use of morphological evidence — in addition to the low number of cognates and nearly identical phonology — it is worth mentioning Nichols’ (1996: 41) claim that the first evidence for a genetic grouping may come from the grammar: “An initial assumption of relatedness is made on the basis of solid evidence that firmly identifies a unique individual proto-language; that evidence is primarily grammatical and includes morphological material with complex paradigmatic

18 With “inspectional” not meant in a totally literal sense. 19 She cites an unpublished paper by Sheldon Harrison which argues that Meillet has been misinterpreted.

38 

 Harold Koch

and syntagmatic organization”. This assumption of course needs to be confirmed by the discovery of sound correspondences and the reconstruction of details of the proto-language.

2.9 Dixon’s Punctuated Equilibrium²⁰ 2.9.1 General features Dixon (1997) put forward what he hoped would be a comprehensive model of linguistic change, describing both the relations of linguistic divergence typically modelled by family tree diagrams and areal relations typically mapped by isoglosses. Although a primary impetus for designing this model was Dixon’s “experience of trying (over a period of 30 years) to make sense of the linguistic situation in Aboriginal Australia” (Dixon 1997: 5), he suggested that “the punctuated equilibrium model is an appropriate one to describe and explain language development in a global context” (Dixon 1997: 140). In his view divergent and convergent changes (a) occur at different times, (b) take place at different rates, and (c) result from different kinds of non-linguistic factors. Throughout most of human history the kind of linguistic splits that can be captured by tree diagrams typically occurred in bursts of relatively short duration and a much greater proportion of history was characterised by periods of relative equilibrium in which linguistic traits diffused between adjacent languages, which tended to converge on a common structural prototype, with the overlaid layers of diffused material obscuring the original genetic structure. This convergence of linguistic (and cultural) features takes place in a geographical area and in a social situation in which different political groups of roughly similar size live in harmony, sharing a somewhat similar culture, with none exerting dominance over the others. The shorter periods of punctuation stem from non-linguistic factors such as natural events (drought, floods, changing sea levels, volcanic eruptions, disease), material innovations (new tools, weapons, or means of transport, the introduction of agriculture, the domestication of animals), the development of aggressive tendencies (concomitant with the rise of hierarchical leaders or a new religion), the expansion into new territory, etc.

20 This theory, abbreviated PE, is discussed here, just before consideration of borrowing, since it involves issues of both classification and diffusion. Dixon’s earlier 50% Equilibrium Level model is discussed in § 4.2.1.

Historical relations among the Australian languages 

 39

2.9.2 Critique by general historical linguists PE as a general model of language development has been found to be not supported. Campbell (2003: 49–50) finds the social explanations to be “speculative and anthropologically unrealistic”. He shows with examples “that the correlation envisaged by Dixon, which equates equilibrium with convergence, and punctuation with divergence, is not supported — both kinds of change take place in both kinds of situations” (Campbell 2003: 51; cf. Campbell and Poser 2008: 318–325 for further discussion). Crowley (1999: 115) has a similar assessment: “Language change typically involves both divergent and convergent developments taking place at the same time, regardless of whether the situation is one of equilibrium or punctuation.” Watkins (2001: 62) finds that “the Indo-European examples show that both contact-induced linguistic change (i.e. diffusion) and system-internally driven linguistic change can occur with equal abruptness and rapidity — thus both counting as ‘punctuation’”, and that “[t]he language areas involving Indo-European languages have all been characterised by interdiffusion of grammatical features, but in none can we really speak of convergence to a common prototype, in the sense of loss of linguistic identity”. Kuteva (1999) studies the correlations between socio-historical situations and linguistic processes (convergence in a Sprachbund, co-existence, expansion-and-split, language assimilation) and shows that “the same linguistic situations are to be found in both socio-historical equilibrium and socio-historical punctuation” (p. 227). Haspelmath (2004: 214) concludes that “the concepts of equilibrium and punctuation are too general and too vague to be of much use when particular languages and historical situations are examined.”

2.9.3 Application to Australia Dixon (1997, 2001, 2002) applies the PE model to Australian languages, claiming (1997: 68) that “it is the only model able to explain the relationships between languages in Australia”. Both genetic and areal considerations are involved. According to Dixon (1997: 90) it is more likely that modern Australian languages descend from one or more languages of the first inhabitants of the continent, who arrived some 50,000 years ago, than from a single proto-language spoken some 5,000 to 10,000 years ago. Their linguistic history consisted of a super-long period of stasis, during which a constant ebb and flow of mutual influence led to the continuous diffusion of linguistic features, with the result that any original genealogical structure would have been obliterated. The 30 low-level genetic groups that Dixon accepts result from “minor punctuations in quite recent times” (2002: xix), such as the expansion of populations into territory that had been vacated during periods of extended drought (1997: 92). There is “no clear evidence for higher-level genetic grouping” (Dixon 2001: 64), nor can it be established “whether all the languages come from one ancestor, or whether there were several genetic origins” whose traces have “become blurred through tens of millennia

40 

 Harold Koch

of diffusion” (Dixon 2001: 64). Most linguistic features are distributed over a continuous geographical area, and the distributions of different features do not match one another — i.e. there is no bunching of isoglosses. Dixon maps these areal distributions and typically explains each by separate processes of diffusion. Various “small linguistic areas” are explained in terms of a period of diffusion within the area. It is further claimed that the languages of the Australian mainland as a whole constitute a large linguistic area of great time depth, formed during an extended period of equilibrium during which there were no major punctuating events. He concludes from this situation that it is not sensible to try to establish a family tree for all the languages, and that hence the comparative method is of limited applicability in Australia.

2.9.4 Critique of Dixon’s Australian interpretation Dixon’s interpretations have generally not been accepted by Australianists. The one case study that might be considered supportive is Dench (2001), which emphasises the difficulty of separating genealogical from contact-induced causes for the distribution of features among the languages of the Pilbara region. Evans (2005: 264–271) shows that a number of grammatical morphemes (personal pronouns and the ergative suffix *-ngku) do indeed share an isogloss — one corresponding to the boundary between PN and nPN languages — and argues that gradually accumulating evidence from reconstructed grammatical and lexical forms supports both a higher-level genetic grouping of a PN family (Evans 2005: 263–264) and the genetic relatedness of many of the nPN families (Evans 2003a: 17–21, 2005: 274–275). Likewise Haspelmath (2004: 212) observed: “since Australian languages have plenty of grammatical morphemes, which are generally not so easily borrowed [as vocabulary], I do not see why deeper genealogical classification in Australia should be as pointless as Dixon suggests it is.” Several of Dixon’s areal groups have been convincingly demonstrated by other scholars to be genetic groups (Koch 2004b for Arandic; McGregor and Rumsey 2009 for Worrorran/North Kimberley; Bowern 2001, 2006, Hercus and Austin 2004 for Karnic and Yarli in the Lake Eyre Basin). On the question of Australia as a whole being a linguistic area, Haspelmath (2004: 211), argues that Dixon’s (2001) listing of “features that are found in all or most of the languages of the continent” (p. 65) “falls short of demonstrating that Australia is a linguistic area, because linguistic areas need not only be internally coherent, but also distinctive with respect to languages outside the area. Thus, one would have to show that the Australianisms are uncommon in the rest of the world, or at least in adjacent areas such as New Guinea.” Dixon’s use of diffusion has been faulted on methodological grounds. Alpher (2005: 795) complains that in taking diffusion as the default hypothesis for any language change Dixon “often fails to try to identify just which similarities are in fact due to diffusion and which are not.” Sutton and Koch (2008: 483–485) show, for two of his historical explanations (involving personal pronouns and noun-class prefixes),

Historical relations among the Australian languages 

 41

that a better case can be made for common inheritance than for Dixon’s borrowing scenarios. Dixon’s pessimism concerning the applicability of the comparative method in Australia is not widely shared. Evans (2005: 243) argues that “a renewed interest in applying the comparative method by a number of Australianist scholars is actually succeeding in giving a more coherent account than that argued for in AL [= Dixon 2002]” — referring especially to studies in Evans (2003b) and Bowern and Koch (2004) — and that new models or special methods for the investigation of Australian linguistic prehistory are not therefore required. Bowern (2006) addresses Dixon’s use of origin myths to support the stability of language-land associations, his unrealistic demands for justifying a PN family (in terms of subgrouping innovations), and his criticism of the family-tree model as it can be applied in Australia. Alpher (2005: 804) disputes Dixon’s claim that the expansion and split of a language can only be due to a cataclysmic event called a “punctuation”; this would rather have occurred frequently as the consequence of the extinction of clan-level lects and “the subsequent succession to ownership of the land by a neighboring group with a legitimate traditional claim to it (without population shift).” Sutton (in Sutton and Koch 2008: 497–500) disputes Dixon’s claim that no major punctuating events have occurred since the original settlement of Australia some 50,000 years ago, citing the rising of the sea level that separated Australia from New Guinea, which stabilised about 6000 years ago, climate change, and the cultural change involved with the explosive expansion of the PN languages.

3 Borrowing, diffusion, and areal relations Key references on this general topic are Heath’s studies of structural and lexical diffusion in Arnhem Land (Heath 1978, 1981) and surveys by Heath (1979) and McConvell (2010).

3.1 Detectability of loanwords Undetected loanwords can interfere with linguistic classification: they may incorrectly be attributed to proto-languages.²¹ Undetected loanwords inflate lexicostatistical percentages and thus distort any interpretation of language relations derived

21 Greenhill et al. (2009) show, by means of computer simulations, that borrowing levels have to be atypically high before they seriously distort tree construction (i.e. classification) according to Bayesian phylogenetic methods.

42 

 Harold Koch

from these percentages. Alpher and Nash (1999: 37–40) demonstrate this for Kuuk Thaayorre, Yir Yoront, and Kok-Kaper (Koko-Bera). On the other hand, the fact that borrowing has taken place can be detected from skewed patterns of lexicostatistical percentages (see Black 1997, 2007b for demonstrations). The identification of specific loanwords is more difficult: “the similarity of Australian phonological systems makes borrowing harder to detect on phonological grounds. But many studies have been able to overcome this through careful reasoning from sound changes” (Evans 2005: 261, note 31). In the ideal case loanwords fail to manifest the sound correspondences that obtain between true cognates; if these correspondences are known, this failure to conform is an indication of their non-cognate status.

3.2 Borrowing rates 3.2.1 Dixon’s 50% Equilibrium Level model Dixon (1970a: 652–656, repeated in 1972: 331–337, 1980: 245–255, 1997: 26–27, 2001: 84, 2002: 27–30) proposed a model of “lexical diffusion in Australia” which has been influential in discussions of classification. He hypothesised that adjacent unrelated languages may through borrowing increase their cognacy score to about 50%, and that adjacent related languages may through lexical replacement decrease their cognacy score to about 50%. Since this “equilibrium level” (which he broadened to a 40–60% band) could result from either divergence or convergence, such a lexical score cannot safely be used as indication of genetic relation. Rather one must look to grammar to find a better indication of the genetic situation; and within vocabulary, it is claimed, the cognate score for verbs also provides better evidence, since verbs are less easily borrowed than nouns. As a consequence of Dixon’s claims, Australianists have been reluctant to interpret neighbouring languages as being closely related unless their lexical score is above 60%, or either the verb score is significantly higher than that of general vocabulary or the grammar indicates a higher degree of agreement.²² Dixon’s model presupposes that all lexical replacement is the result of borrowing from neighbouring languages. Alpher and Nash (1999: 25), on the basis of a careful study of languages in the south-western part of Cape York Peninsula, estimate that “the contribution of borrowing to lexical replacement in these languages is probably less than 50% at a maximum and most usually well below 50%”. The consequences

22 As an example of this deference to Dixon’s model, Austin et al. (1980: 173) present lexicostatistical counts for Gamilaraay and Yuwaalayaay vs. Ngiyampaa and Wiradjuri (languages of the Central New South Wales subgroup), then comment: “the figures for total vocabulary are at the lower bounds of the ‘equilibrium range’. Grammatical comparison is necessary before the issue can be decided … the figures for the verb category are higher and towards the middle of the ‘equilibrium range’”.

Historical relations among the Australian languages 

 43

of this finding for Dixon’s model is that the equilibrium figure, at which loss of vocabulary items is balanced by acquisition of apparent “cognates” by borrowing, is “approximately one-half of the fraction of replacement due to borrowing, that is, a maximum of about 25%”. “Therefore, geographically contiguous languages with ‘cognation’ percentages in the 20–40% range need not be assumed to have developed apart and come into contact only recently, and claims as to the distance of their relationship can be made (admittedly tentatively) on lexicostatistical grounds alone” (Alpher and Nash 1999: 28). Evans (2005: 258–261) cites numerous examples of adjacent languages whose genetic relationship is known to be distant but which share considerably less than 40%, and argues that a number of Dixon’s examples of “small linguistic areas” adduced in support of a 40–60% level resulting from contact are rather to be interpreted as genetic groupings. Black (2006) explores the implications of Dixon’s model for eleven language situations of less than 40% between neighbours and shows that the model would require an unrealistic picture of relatively recent language movements. Evans (2005: 261) concludes: “It is now time to reassess the 50 percent equilibrium claim and downgrade its status from dogma to apocrypha”. I would add that it is also necessary to reconsider the genetic classification of languages where judgments have been made on the basis of Dixon’s 50% equilibrium model.

3.2.2 Are borrowing rates higher in Australia than elsewhere? We now have available figures on percentages of lexical sharings between adjacent languages for dozens of languages from much of the continent: Blake and Reid (1998: 4–5) for Victorian languages, Evans (2005: 258–259), Black (2006), Harvey (2011). Some of these figures are based on basic vocabulary, but many of Harvey’s 50 languages are based on the total lexicon. These counts are sufficient to establish that “neighbouring Australian languages do not uniformly show high levels of shared vocabulary” (Harvey 2011: 348). Very many of these data sets involve figures far below the 40% predicted by Dixon’s equilibrium theory. The one frequently mentioned exception is Heath’s (1981) case of Ngandi and Ritharrngu: Harvey (2011: 367) recalculates the sharing as 40% based on the total lexicon. Jingulu and Mudburra, for which Pensalfini (2001: 392–395) reported figures of 40%, 56%, and 71% (using different methods) for a Swadesh 200-word list, share 40–43% according to Black’s (2007b) careful study using older data and a 148-item basic wordlist. For Gurindji, McConvell (2009) found about 35% of Haspelmath and Tadmor’s (2009) test list of 1,460 lexical meanings to have been borrowed from other Australian languages and identified the source language in most cases. If loanwords from English are included, the Gurindji rate is 45.6%, which is high by world standards (English is 40%), according to Tadmor (2009: 56). It is probably significant in this case that the Gurindji language (as well as the neighbouring Mudburra) has at some time in the past spread into the territory of

44 

 Harold Koch

nPN languages; McConvell suggests (p. 807) that language shift with loanwords from a substratum may have been a factor in the heavy borrowing. Bowern and Atkinson (2012) claim that “loan levels for Australia as a whole have been overstated … In a survey of 49 Australian languages (Bowern 2011), mean loans in basic vocabulary were found to be 8.7%, with median loans at 5.54% (SD 11.01). 75% of the languages surveyed had loan levels under 12% in basic vocabulary. These loan findings … place Australian loan levels on a par with those found in the rest of the world”.²³ Black (1997: 67) likewise observes: “Although some have suggested that Australian languages change more rapidly or borrow more heavily than languages elsewhere in the world, this is not supported by the available evidence.” Likewise Evans (2005: 261) claims that “the levels of borrowing one finds in Australia are no higher than in such familiar cases as French-English, so that it is unclear why Australia needs to be treated as a ‘special case’; borrowing poses no greater problems to lexicostatistic classifications, or to the comparative method, than it does in IndoEuropean or Austronesian, for example”.

3.3 Borrowing according to kind of vocabulary 3.3.1 Borrowing of basic vocabulary Dixon has repeatedly and emphatically claimed that Australian languages are exceptional in that there is no difference in the rate of borrowing of basic or core versus non-basic vocabulary: “there is no evidence that basic items are less liable to be borrowed than non-basic terms in Australian languages” (1980: 254). Heath’s (1981) discussion of the high level of borrowing between Ritharrngu and Ngandi, where even “a considerable number of 100-word list items have been diffused” (p. 364), would appear to provide support. Dixon’s claim seems to have been accepted at face value by non-Australianists (e.g. Haspelmath 2004: 212).²⁴ Dixon has never supported this claim by the publication of wordlists and his judgements about relations between their items; the claim simply rests on his assertion that he has consistently found similar lexical scores between pairs of languages regardless of size of the list compared. Bowern (2006: 254) has tested this claim by comparing scores from a 100-word Swadesh list of basic vocabulary with the first 100 words of dictionaries, for Bardi and Yawuru (of the Nyulnyulan family) and Karajarri (a language of the remotely related Marrngu subgroup of the PN family, which has nevertheless been in contact

23 For further details see Bowern et al. (2011). 24 “It is my impression that the main difference between Australia and other parts of the world is that ‘core’ vocabulary appears to give very little information about genealogical relatedness, because all kinds of words are very easily borrowed.”

Historical relations among the Australian languages 

 45

with the Nyulnyulan languages). She has found a significant difference between basic and general vocabulary both between closely related languages (Bardi–Yawuru 31% basic vs. 15% general) — showing a higher rate of retention of basic vocabulary — and between distantly related languages (Bardi–Karajarri 8% basic vs. 21% general) — showing a higher rate of borrowing of non-basic items. Black (1997: 62–63) compared the vocabularies of two languages of the Torres Strait Islands, the Australian language Mabuiag (Kala Lagaw Ya) and the Papuan Meriam Mir (where all shared items must be the result of borrowing) and found a clear pattern of difference between the percentages of: (a) closed-class grammatical forms, (b) basic nouns, verbs and adjectives from a 100-item list, and (c) species names and cultural terms. He found a similar pattern obtaining between Ngandi and Ritharrngu, which had been the subject of Heath’s study. According to Evans (2005: 260), “[Black’s] paper demonstrates that in Australia core vocabulary is, in fact, less prone to borrowing than other domains”.²⁵

3.3.2 Borrowing according to word class Do borrowing rates differ according to word class (parts of speech)? Dixon (1970a: 661) claimed, supported by the relative lexical percentages for languages of the Rain Forest region, that “borrowing is likely to involve a higher proportion of nouns than of verbs and adjectives”, suggesting as a reason that verbs, unlike nouns, typically do not occur without inflections. The factor of inflection appears to be significant. In Gurindji, which has an atypically high rate of loanwords, McConvell (2009: 795) found a radical difference between inflected simple verbs (of which there are 30 in the language) and coverbs: “Of the simple verbs only 5% are possibly borrowed, whereas of coverbs around 70% are borrowed.” Coverbs are (basically uninflecting) elements “which are interpreted together with a light verb to yield what might be in other languages, including most Australian languages, a single verb” (p. 798). Since it is the coverb which has the greater semantic content in light verb constructions, it could alternatively be argued that the factor responsible for this imbalance in borrowing behaviour is their semantics rather than their lack of inflection. However, the world sample yielded an average of 31.2% loanwords for nouns vs. 14% for verbs and 15.2% for adjectives and adverbs (Tadmor 2009: 61).

25 Breen (2011) gives counts for Kalkatungu and Yalarnnga (p. 247) and for Warlpiri and Warumungu (p. 263) that show higher figures for larger lists (which contain less basic vocabulary).

46 

 Harold Koch

3.3.3 Borrowing according to semantic domain Most claims about relative borrowability according to semantic domains are based on counts of look-alike words in adjacent languages rather than on detailed studies that identify the fact and direction of borrowing of particular words (e.g. Breen 1990, 2011, Breen and Blake 2007: 71). Bodypart scores are found to correlate with a close genetic relation better than nouns in other semantic domains, and domains of flora and especially fauna “provide most of the common items in most unrelated and contiguous pairs” (Breen 2011: 261). Dixon (2002: 27) asserts: “[r]ight across Australia we find that the names of plants and animals have diffused across a group of contiguous languages.” Two studies have been based on situations where it has been possible to identify actual loanwords,²⁶ and where high levels of borrowing have been reported: Heath (1981: 355) on diffusion between the remotely related Ritharrngu and Ngandi and McConvell (2009) on Gurindji loanwords from nearby nPN languages, related Ngumpin languages, and English/Kriol. Heath (1981: 355) found the following percentages of shared terms,²⁷ most of which result from borrowing, according to semantic domain: 78% major human age/sex terms, 67% cardinal-direction nouns, 65% trees/shrubs, 53% major physical-feature nouns, 35% insects, 28% body-part nouns, 22% kin terms. McConvell’s (2009: 796) breakdown of loanwords into semantic fields includes the following:²⁸ 66.2% physical world, 58.3% warfare and hunting, 55.2% animals, 48.5% food and drink, 46.2% kinship, 42.4% the body. “Wanderwörter”, i.e. words which have been “borrowed from language to language, across a significant geographical area” (Trask 2000: 366), present a special problem, since they can be mistaken for cognates. It would be helpful to know in which semantic domains they are likely to be found. Present indications suggest at least the following: social categories (McConvell 1985), affinal kinship, ceremonies, and material culture.

26 See Nash (1997), Koch (in press) for discussion of loanword identification. 27 These figures represent the percentage of items on Ngandi wordlists that are also found in the corresponding Ritharrngu lists. 28 These figures represent the percentage, within each semantic domain, of Gurindji words which can be identified as loanwords from any other language. The words compared are taken from the 1460 items of the Loanword Typology meaning list.

Historical relations among the Australian languages 

 47

3.4 Borrowing of grammatical words 3.4.1 Pronouns and demonstratives McConvell (2009) found that “function words are borrowed much less than other classes: around 15% as opposed to around 50% for nouns, verbs, and adverbs” (p. 795), and that “[a]ll Gurindji personal pronouns, both free and bound (enclitics), and demonstratives are inherited, not borrowed” (p. 797). Heath (1978: 102–103) found no evidence that demonstratives were borrowed, though they are highly unstable, and found independent pronouns to be resistant to borrowing. Dixon (2002: 293), however, asserts: “In many parts of the world, pronouns are said to be resistant to borrowing. There is no such constraint in Australia.” He claims that diffusion of forms is responsible for the distribution of (PN) 1DU *ngali, and to some extent 2PL *nhurra and 3DU *pula.²⁹ This interpretation is disputed by Alpher (2005), Evans (2005), Sutton and Koch (2008), who rather see these as inherited from pPN, according to the reconstruction of Blake (1988).³⁰ Some plausible examples of borrowed pronouns, however, include: the 3SG.NOM form ngayi in eastern Yolngu languages, apparently borrowed from nPN languages beside inherited suppletive oblique (originally Feminine) stem *nhan- (Blake 1988: 26);³¹ Garrwa and Waanyi nominative forms 3SG.M nyulu, 3DU pula, DU1 ngali, 1PL.INC ngampala — apparently borrowed from adjacent Warluwarric languages, beside the suppletive inherited 3SG.M oblique stem nanga- (Blake 1990a: 61–63); and the 3PL form yalu of most Warluwarric languages, apparently borrowed from Alawa or one of its near relatives (Blake 1988: 21–22).

3.5 Borrowing of bound morphology 3.5.1 Heath’s findings Heath (1978) devotes a chapter to “Direct diffusion of morphemes” in his study of languages of Arnhem Land. He found no evidence for the borrowing of verbal suffixes marking tense, aspect, mood (which involve numerous allomorphs and several inflectional classes), or of verbal pronominal prefixes marking subject and object (p. 101). Morphemes that are borrowed include suffixes marking case (Ergative -dhu,

29 This scenario is in fact required if inheritance from a common proto-language is denied. 30 See Sutton and Koch (2008: 484) for an alternative account of Dixon’s scenario whereby Yidiny ngali has been borrowed from Dyirbal. 31 Blake does not specify a particular source for Yolngu ngayi, but reconstructs *ngaya as the ProtoNorthern 3Sg.F pronoun and cites ngaya forms for Ndjebbana (p. 51), Mangarrayi (p. 53), and the Eastern Mirndi languages (p. 57).

48 

 Harold Koch

Instrumental -mirri), kin dyads -ko?, diminutive -ganang?, prefixes marking ‘comitative’ barta- and ray-, noun compounding elements indicating ‘times’ malk- and ‘side’ bala-. Heath (1978: 104–115) proposes the following characteristics of borrowable morphemes³²: (1) they are pronounceable, consisting of whole syllables; (2) they have clear boundaries (most easily seen where a form containing the morpheme contrasts with one without it); (3) they have a single function (e.g. Dative rather than Dative and Plural); (4) they have a transparent meaning (e.g. one does not have to interpret it differently depending on its morphosyntactic environment). It will be apparent that these characteristics of borrowable morphemes are those that are typical of agglutinative morphological structure. I suggest that the main reason that grammatical elements are more easily borrowed in Australian languages than has been found for Indo-European languages has to do with the agglutinative structure of the former and the fusional structure of the latter. The morphemes which Heath found resistant to borrowing — verb tense–aspect–modality suffixes and verbal prefixes marking subject, object, and sometimes mood — do not have the agglutinative characteristics of the borrowable affixes. In spite of these valuable results, however, some of Heath’s specific claims have been overturned, as indicated in the following two subsections.

3.5.2 Noun class prefixes Heath (1978: 88) suggests that “the spread of noun-class systems over much of northcentral and north-western Australia may well have been largely accomplished through direct diffusion of the actual affixes, rather than through independent developments in each language group”.³³ He argues in particular that the three non-human class prefixes of Warndarrang, (r)a-, wu-, and ma-, were probably borrowed from an earlier stage of its neighbour Wubuy (Nunggubuyu). Evans (2003a: 16) finds Heath’s argument not persuasive, and argues that cognate data from further afield (e.g. Mawng) rather suggests that Warndarrang’s class prefixes were inherited. Harvey (2012) shows that three Warndarrang prefixes can be reconstructed as *na-, *gu-, and *ma- in a high-level proto-language, and for a fourth prefix, (r)a, provides more cognates from Limilngan and Gulumoerrgin (Larrakia), which support an ancestral *dV- prefix. On prefixation see further § 3.9.

32 The world-wide relevance of these structural characteristics is invoked by several contributors to Johanson and Robbeets (2012). 33 Sands’ (1995) study implies rather that classifiers developed separately in different languages by the grammaticalisation of generic nouns, an idea based on claims by Dixon (1970b, 1980: 273).

Historical relations among the Australian languages 

 49

3.5.3 Verb derivational suffixes Heath (1978: 92–96) argues that Ngandi had borrowed from Ritharrngu two derivational suffixes, an inchoative verbalising -thi- (‘become’) and a “thematising augment” -dhu- (which enabled coverbs to be inflected). Alpher, Evans and Harvey (2003), however, find that the Ngandi forms can better be explained as reflexes of Proto-Gunwinyguan *-thi ‘inchoative’ and *thu- ‘tell off ’; these are then both cases “of claimed diffusion turning out to be a case of shared (deep-level) inheritance once comparative work is done in more detail” (p. 339, cf. p. 344).

3.5.4 Paradigms? It is usually taken for granted that paradigms cannot be borrowed. Furthermore the sharing of paradigms, especially ones with idiosyncratic structure, has been taken as the strongest evidence in support of genetic groups such as the Maningrida, Gunwinyguan, and Southern Daly families (Harvey 2011: 352–353). I. Green’s (2003: 155) case for the genetic relation of Murrinh-Patha and Ngan’gityemerri in a Southern Daly family is based primarily on shared paradigms of verbal auxiliaries, whose resemblances can only be explained by a genetic relationship: “Clearly there can be no other credible account of the formal similarities of the two languages. The matching array of both regularities and suppletions could obviously not have arisen by chance. Diffusion is similarly to be dismissed as a possible cause of the resemblances.” Paradigms have been used to recognise genetic groupings in situations of intensive diffusion, as noted by Heath (1978: 145): “For all the respects in which Ng[andi] and Ri[tharrngu] have converged, one glance at the verbal paradigms, for example, tells us immediately that Ng[andi] is closely subgrouped with Nu[nggubuyu], while Ri[tharrngu] belongs with the other Yuulngu languages. Similarly, Nu[nggubuyu] and Wa[rndarrang] share many features, but examination of verbal paradigms and to some extent pronominal prefix paradigms shows that Wa[rndarrang] is closer to Mara and Alawa than to Nu[nggubuyu] and Ng[andi].”

3.6 Indirect diffusion: copying of structure without substance 3.6.1 Grammatical structure 3.6.1.1 Bound pronouns According to Dixon (2002: 292) “[t]he most prevalent kind of diffusion is the borrowing of grammatical categories and patterns (rather than of forms)”. Heath (1978 passim) uses the term “indirect diffusion” to describe the copying of patterns without the actual forms (morphemes or words). Heath (1978: 101–102), Waters (1989: 279–282),

50 

 Harold Koch

and Dixon (2002) interpret the development of obligatory bound pronouns in the Yolngu languages Ritharrngu, Djinang and Djinba as “motivated by diffusional pressure from the adjacent languages … that have bound pronominal prefixes” (Dixon 2002: 339). Mushin and Simpson (2008: 583–584), however, urge caution in accepting contact as the explanation for these developments, citing the presence of reduced pronouns in other Yolngu languages that are not in contact with nPN languages, the use of second position for continuing reference in other Australian languages, and a Djinang-like ordering of clitics in a distant PN language, Kugu Nganhcara: “We conclude … that while diffusion from non-Pama-Nyungan languages may have played a part in the development of obligatory pronominal clitics in Ritharrngu and Djinang, it is not sufficient to explain the separate positional restrictions in the two languages” (p. 284). It is also worth noting that McConvell (2010: 808) found for Gurindji that, while it has borrowed heavily from nearby prefixing nPN languages, its enclitic pronouns “are widespread in western Pama-Nyungan and are inherited from some western Pama-Nyungan proto-language, not borrowed. They are unrelated to the Non-Pama-Nyungan bound pronouns and cannot be seen as products of structural diffusion from Non-Pama-Nyungan languages.” Dixon (2007) makes a case that, in the large multi-dialect language of western Victoria, Wemba Wemba, the system of pronominal clitics of various dialects has been influenced by that of adjacent linguistic neighbours. The same paper describes the pronominal systems of fifteen dialects of the Western Desert language, and finds some correlation with the systems of external linguistic neighbours.

3.6.1.2 Type of verb organisation Many languages of northern Australia have a bipartite verb structure, in which lexical verbal meanings are expressed by a combination of an invariant “coverb” and an inflecting light verb (cf. Bowern, this volume and Dixon 2002: 183–201). Of the PN languages, this pattern is especially prevalent among the Ngumpin languages, which are located in the inland northwest adjacent to nPN languages. The Ngumpin profile of a small number (30–40) of inflecting verbs, compared to their more distant relative Warlpiri, with over 100 monomorphemic verb roots (Nash 2008: 221), is attributed to nPN influence: “Eastern Ngumpin languages have large numbers of coverbs borrowed from nPN neighbours — at least 30–40% in Gurindji and Ngarinyman” (McConvell and Laughren 2004: 172). McConvell (2009: 809–810) finds that both indirect diffusion (of patterns) and borrowing of forms have taken place, and these two processes reinforced one another: “The loose nexus complex verb arrangement allows insertion of invariant verbal forms into a frame with a light verb without the complications which arise when these elements are morphologically bound together. Additionally, the more frequent and less marked this arrangement becomes due to the volume of coverb loans, the more it becomes established as the standard pattern and reduces the numbers of monomorphemic verbs in the languages joining the area”. Dixon’s case

Historical relations among the Australian languages 

 51

for indirect diffusion in the reverse direction cannot be maintained, however. Dixon (2001: 74–75, 2002: 199) suggested that Warlpiri formerly had fewer simple verbs but increased its inventory of simple verbs under diffusional pressure from neighbouring languages in the Western Desert (Wati) and Arandic subgroups. An etymological study of Warlpiri verbs by Nash (2008) shows instead that the majority of Warlpiri verbs are inherited from Proto-Ngumpin-Yapa and higher levels (with many cognates being lost in the Ngumpin languages), and concludes that “we do not need to countenance the kind of ‘areal pressure’ which would cause a language to add simple verbs, many of undetectable origin” (p. 233).

3.6.2 Semantics of grammar One might expect the meaning systems to be more subject to contact influence than formal aspects of grammar. Djinang and Djinba have acquired an unusual system of verbal tense characteristic of the adjacent Maningrida subgroup, whereby a “Contemporary” tense describes events in the present or yesterday, while “Precontemporary” refers to events that happened either earlier today or before yesterday (Dixon 2002: 211). Hercus and White (1973) discuss the areal distribution, in south-central Australia, of non-singular pronouns that signal kinship relations between the referents (thus there are separate pronouns referring to ‘we two mother and child’ and ‘we two father and child’). Since the languages involved include those of the Thura-Yura, Karnic, and Arandic subgroups some diffusion of both the system and some forms must have taken place (cf. Koch 1982 for Arandic forms).

3.6.3 Lexical Semantics Patterns of lexical semantics are widely assumed to be diffusable. Thus Austin, Ellis and Hercus (1976) study a pattern of noun compounding where the addition of the second element indicates a piece of a larger whole (e.g. ‘raindrop’ vs. ‘rain’, ‘mulga seed’ vs. ‘mulga tree’). The same lexicalisation pattern is found in a set of languages of the Lake Eyre region, even though the elements are not cognate and the languages belong to different genetic groups: Diyari and Arabana-Wangkangurru are in different subgroups of the Karnic family, and Kuyani and Adnyamathanha belong to the Thura-Yura family. The authors favour an explanation in terms of semantic diffusion: “there has been linguistic borrowing at some time in the past, not of lexical items but of a semantic concept” (p. 63–64). McConvell (2009: 810) reports that Gurindji has taken on aspects of areal semantic organisation from its nPN neighbours, giving up the inherited polysemies of ‘inside’ and ‘under’ and ‘hit with a spear’ and ‘grind’, and replacing the polysemy of ‘hill’ and ‘stone’ with the areal polysemy of ‘hill’ with ‘head’.

52 

 Harold Koch

3.6.4 Phonology Phonological patterns have been attributed to indirect diffusion. These patterns may involve, inter alia: contrasts in place of articulation, contrasts in manner of articulation, allophonic features, phonotactic possibilities. Waters (1989: 286) attributes the loss of laminodental consonants in Djinang and Djinba (two Yolngu languages), through merger with the laminopalatals, to the influence of neighbouring languages Burarra and Rembarrnga, which do not have this contrast. Heath (1978: 39) claims that the lenition of lenis (voiced) stops to approximants in Wubuy (Nunggubuyu) “was stimulated primarily by contact with Wa[rndarrang]”, bringing “the Nu[nggubuyu] consonantal system into line with the Wa[rndarrang] system”, which had a single series of obstruents. Hercus (1972) finds the development of prestopped allophones of nasals and laterals (e.g. [kudna] ‘excrement’, [ngadli] ‘we two’) to be shared between several genetic groups of south-central Australia, and attributes this sharing to diffusion. Evans (2005: 255) attributes the “substitution of s and z for laminopalatal stops” in the Western Torres Strait language (Kala Lagaw Ya) to areal convergence with the language of the eastern islands. Blake et al. (2011: 7) claim: “The loss of final -k on possessor suffixes appears to be related to a general tendency towards vocalic finals in Mathi-Mathi … possibly under the influence of Paakantyi.”

3.7 Motivations for borrowing 3.7.1 Name taboo Dixon’s early writings posit as a significant cause the widespread practice of tabooing a word that sounded similar to the name of a recently deceased person, assuming that the lexical replacement was often borrowed from an adjacent language (Dixon 1980: 28, 99). The importance of name taboo has later been de-emphasised as it is realised that the taboos do not typically affect all speakers of the language and are of variable duration, with the original term not necessarily being lost (Heath 1981: 361, Alpher and Nash 1999: 8–9). The substitute word, moreover, need not be borrowed, but may be a term specifically used for such occasions (Nash and Simpson 1981), an existing synonym, or another word used with altered semantics. Black (1997: 58) concludes his discussion of name taboo as follows: “Since name taboo was generally temporary, and since many languages used substitutes which were also clearly temporary, this cultural practice alone need not have had much effect on lexical change in Australia”. In Dixon (2002: 27) the claim about tabooing is downgraded but still maintained: “Over a long period … the occasional replacement of a tabooed word by a form from a neighbouring language will gradually add up to a significant vocabulary change.”

Historical relations among the Australian languages 

 53

3.7.2 Multilingualism? Voegelin et al. (1963: 25) assert that “the culture of multilingualism is so widespread in Australia as to be virtually universal”; they suggest this as a possible factor responsible for the great degree of lexical diversity that coexists with relative structural homogeneity (i.e. speakers are accustomed to learning the vocabulary of other dialects and languages). Although they do not explicitly relate this to borrowing, it may be thought to be relevant. Heath (1981: 335) attributed the high degree of borrowing in south-eastern Arnhem Land to the “small size of ethnolinguistic group, absence of strong tribal organization, and extensive cross-linguistic marriage.” Harvey (1997) suggests as factors favouring diffusion: population imbalance (in particular a small group adjacent to larger one), considerable time depth of contact, and continuity of contact, such as reciprocal marriages over many generations. Dixon (2002: 25) cites as factors favouring diffusion “a fair proportion of reciprocal exogamous marriages, rampant multilingualism, and an open attitude to borrowing”. Harvey (2011) argues that multilingualism does not necessarily lead to lexical borrowing and cites situations where lexical change increases differentiation between lects: he claims that where lects already share much vocabulary, differentiation is better effected by means of lexical derivations and neologisms than by borrowing. He also relates the prevalence of borrowing to land ownership. The strong ties that bind together members of a land-owning group operate as a conservative force disfavouring borrowing. Where changes of land tenure took place, new and weaker ties were established between the new owners and their neighbours and these looser networks provided the vehicle for “supra-local convergence” as linguistic elements involved in code-switching were adopted as loanwords into the language — consistent with the social network explanations in sociolinguistics (Milroy and Milroy 1985). Harvey posits that this process accompanied the expansion of PN languages, and cites in support the fact that four pairs of languages across the PN/nPN divide have high rates of lexical sharing: Ritharrngu-Ngandi and Mudburra-Jingulu, each 40%; Karajarri-Yawuru 37%, and Ngarinyman-Jaminjung 26%. There is no doubt that many estates (as locally owned territories are called) have changed their linguistic affiliation over time. Exactly what linguistic traces this leaves should to be explored further. Besides the kind of borrowing mentioned by Harvey, language shift of populations may have played a role in linguistic diffusion. McConvell (2009: 811–812) mentions “the possibility that high numbers of loans in Gurindji are associated with past language shift by non-Pama-Nyungan speakers to Pama-Nyungan: part of the borrowed vocabulary is in fact substratal rather than adstratal, and having both kinds of non-inherited items elevates an aggregated loanword percentage”.

54 

 Harold Koch

3.8 Areal distribution of features It has long been observed that many of the features of Australian languages have a distribution that can be described in terms of geography. There has been a relatively strong tradition in Australian linguistics of plotting the areal distribution of features — apical and laminal consonant contrasts (Dixon 1970c), prestopping of nasals and laterals (Hercus 1972), deletion of initial consonants (Alpher 1976; Sutton 1976; Hercus 1979; Dixon 1980: 198), bound pronouns (Dixon 1980: 364), switch reference marking strategies (Austin 1981b). Dixon (2002) has over twenty maps showing the distribution of phonological and grammatical features. A major problem in historical linguistics has been deciding which of these sets of shared features should be interpreted as the result of contact and which owe their distribution to inheritance within a language family or subgroup.

3.9 Diffusional vs. genealogical explanations What constitutes a diffusional explanation? According to Thomason (2010: 32), “[c] ontact is a source of linguistic change if it is less likely that a particular change would have happened outside a specific contact situation”. Evans (2003a: 15–16) observes: “Throughout the last two decades there has been a tendency to overstate the importance of direct diffusion in Australian linguistics, so that in the complex matter of deciding between diffusion and inheritance the default explanation has been taken to be diffusionist by some authors.” He explains (p. 16, note 8) why this is problematic in terms of scientific heuristics: “if we take the classical inheritance approach as our starting point, it is easy to falsify (e.g. through exceptions to regular correspondence sets), whereas if we take diffusion as the starting point, falsification is much more difficult, since every appeal to shared features can be explained as due to yet more diffusion.” It should therefore be obvious that claims that particular features result from diffusion or borrowing need to be based on solid arguments, and not just assumed from their geographic distribution. Thomason (2010: 34) sets out five criteria that need to be satisfied to establish that a particular innovation results from language-external rather than internal causes: (a) considering the proposed receiving language as a whole, (b) identifying a source language from which the features were transferred, (c) finding features that are shared between the two languages, (d) proving that the features were old within the source language, and (e) proving that the features are innovations within the receiving language (i.e., that they did not exist in this language before the proposed contact).³⁴

34 See also Mailhammer (2013) on principles for identifying contact-induced changes.

Historical relations among the Australian languages 

 55

Two areas where the difference between genetic and diffusional explanations have profound implications for linguistic classification (i.e. how the historical relations between the nPN and the PN languages are understood) involve pronominal prefixes on verbs and prefixed noun-class markers. On verbal prefixing, Dixon (1980: 246) has claimed: “It is likely that prefixing of pronominal and other forms to the verb developed in one or two northern languages and diffused over the whole nonPN region.” Dixon (2002: 408) has a more complex hypothesis: “It is possible that prefixing developed independently at more than one place in the present-day prefixing area, and then spread out by diffusion.” With respect to noun classes, Dixon (2002: 471) hypothesises that “noun classes have developed recently, as an areal phenomenon, within the prefixing area; it is basically the CATEGORY of noun classes that has diffused, with each language developing the actual marking for itself, out of its own internal resources.” Both scenarios have been disputed by a number of Australianists, including Evans (2003a), Harvey (2003b), McGregor (2008) — see § 4.5.

4 Genetic classification: substantive claims 4.1 Are all Australian languages related to one another? Schmidt (1919a) recognised a large genetic grouping, labelled the “South Australian languages”, which covered the southern half of the mainland. Among the northern languages he recognised a number of separate groups but considered them to be unrelated to one another and to the great southern family (cf. § 1.2). Kroeber (1923) claimed that it was more likely that all the languages of the mainland were genetically related; he based his conclusions on the geographical distribution of resemblant and presumably cognate forms for seventeen basic noun concepts, especially body parts. His lexical comparisons, however, are not rigorous enough to inspire confidence. Arthur Capell (1937: 58), who personally added greatly to the knowledge of language in the north-western and north-central parts of the continent, concluded that “the languages of Australia, even including those of the Northern Kimberleys, belong to one family”. In Capell (1956) he attempted to offer proof of the ultimate unity of the Australian languages, citing grammatical elements and about 40 lexical elements that he labelled “Common Australia (CA)” — meaning that they were part of a protolanguage ancestral to most of the languages of the mainland, even if they did not represent the original language brought to the continent.³⁵ The full classification by O’Grady, Wurm and Hale (OWH) (1966) included the claim that all the Indigenous languages of Australia were related in one macro-phylum (or

35 See Koch (2004a: 25–30) for an assessment of Capell’s classification methods and results.

56 

 Harold Koch

super-family) — with the exception of: the languages formerly spoken in Tasmania, Meriam Mir of the eastern Torres Strait Islands (known to be related to adjacent Papuan languages of the Kiwai family), Barbaram [Mbabaram] of the Queensland rainforest, and Aniwan [Nganyaywana] of north-eastern New South Wales. (The last two were later proved by Dixon (1991) and Crowley (1976) respectively to be closely related to neighbouring languages, both having undergone radical sound changes which obscured cognates). OWH used the term “linguistic phylum” for two or more language families that are remotely related (OVV 1966: 9), following the practice of Carl and Florence Voegelin (O’Grady 1979: 112). Dixon (1970c, 1972: ch. 1, 1980) posited the unity of the mainland languages and described them in terms of a language family. Dixon (1980) aimed to be “a tentative survey of our present knowledge of Australian languages, together with some hypotheses concerning the structure of the putative ancestor language, proto-Australian, and the development of modern languages from this base” (p. xiii), and “to provide the beginnings of a proof that all the languages of Australia (except perhaps two or three northern tongues such as Tiwi and Djingili [Jingulu] are genetically related” (p. xiv). The viewpoint represented in Dixon (1997, 2001, 2002), on the contrary, is agnostic regarding the genetic unity of all the Australian languages: “The question of whether all Australian languages go back to a single ancestor is not answerable, because of the great time-depth involved” (Dixon 2002: xix). Evans (2003a, 2005) puts forward some of the kinds of evidence that he considers point to the ultimate unity of all the languages usually considered to belong to the Australian phylum (cf. § 4.4, § 4.5).

4.2 Are the Australian languages related to languages beyond the continent? Nineteenth century amateur collectors of wordlists (Taplin 1879, Curr 1886–1887, Mathew 1899) used their data to draw unverifiable inferences about the historical relations between Australian languages and languages of other parts of the world such as Africa or India. It has not even been possible to demonstrate that the languages of Tasmania are related to those on the mainland, in spite of apparently general typological parallels — similar phonological system and suffixing morphology (Crowley and Dixon 1981: 395). Given that Australia and New Guinea were part of the same land mass, called Sahul, until separated by the sea some 10,000 years ago, it seems reasonable to expect there to have been a relation with Papuan languages, but none has been demonstrated as yet — except for Meriam Mir in the eastern Torres Strait, which is demonstrably related to languages of the Trans-Fly group in adjacent areas of Papua New Guinea (see Evans 2005: 254–256). It is also possible that, given the time depth of linguistic separation, any further genetic links between Australian and Papuan languages will remain impervious to the methods of compara-

Historical relations among the Australian languages 

 57

tive linguistics, whose reach is usually assumed to extend only to 10,000 years at most.³⁶ The languages of Tasmania were included by Greenberg (1971) in his “IndoPacific” genetic macro-family, which he claimed included most of the non-Austronesian languages of the New Guinea region as well as those of the Andaman Islands. Pawley (2009: 167, 173) considers the resemblant forms used as evidence for relating Tasmanian to any of the other languages to be negligible and accidental.

4.3 The Pama-Nyungan question 4.3.1 Positions on Pama-Nyungan The major issue that concerns the internal classification of Australian languages is whether there exists a single large family corresponding to Schmidt’s Southern Australian languages, Capell’s Common Australian, or OWH’s Pama-Nyungan family.³⁷ The Pama-Nyungan family, if it exists, is one of the great language families of the world, since it includes up to 200 languages³⁸ and covers seven-eighths of the Australian mainland. Its presence raises fundamental questions of prehistory concerning how a language family might have spread so widely in a non-hierarchical hunter and gatherer society. The question, furthermore, has been controversial in Australian comparative linguistics, which in itself presents a puzzle to non-Australianists “who are likely to be unaware of the nature of the evidence on which a commonly recognized family such as Pama-Nyungan rests” (Haspelmath 2004: 211). The Pama-Nyungan family was first proposed by Kenneth Hale in 1961, who “identified Pama-Nyungan and characterised it as possibly the largest coherent genetic linguistic construct in Australia” (O’Grady and Hale 2004: 70), and it was used in the full classification of O’Grady, Wurm, and Hale (1966), O’Grady, Voegelin and Voegelin (1966) and Wurm (1971, 1972). The published justification was primarily lexicostatistical: the finding that the level of shared cognates was much higher within

36 Using different methods, Nichols (1997: 168) finds some support for a claim that languages of southern Australia and inland New Guinea reflect “the structural type of the languages spoken by the first humans to set foot on ancient Sahul”, while those of north-western Australia and coastal New Guinea reflect “the linguistic type of a colonising population most of whose entries came before the end of the Ice Age”—a scenario which she admits (p. 166) is hard to fit with a concept of a single Proto-Australian language. But contrast the findings of Reesink et al. (2009) mentioned in footnote 10. 37 This issue has been highlighted as “The ‘Pama-Nyungan’ idea” in Dixon (2001:89–98, 2002: 44– 54) and “The Pama-Nyungan debate” in Evans (2005: 261–272). 38 O’Grady, Voegelin and Voegelin (1966) list 160 [I count 162] languages; Dixon’s classification includes 189; I have inflated his figure since he sometimes puts together as a single language lects that others would see as separate languages.

58 

 Harold Koch

than outside this group of languages. Hale and O’Grady always maintained their belief that a PN genetic unit could eventually be justified by the comparative method (Hale 1982, O’Grady and Tryon 1990, O’Grady and Hale 2004). Dixon (1980: especially ch. 8) rejected the concept of Pama-Nyungan as a genetic group (retaining it only as a typological and areal grouping), and attributed the shared features rather to his Proto-Australian, assuming that the nPN languages had innovated their verb structures (by adding personal prefixes and developing complex verb stems) and lost the inherited Ergative and Accusative case-markers. Various reviewers (Heath 1982, Laycock 1982, O’Grady 1981) pointed out that since the bulk of material on which Dixon based his reconstructions was from the PN languages, his pA features should rather be attributed to a pPN. Dixon (2002: xvii–xviii) admitted this methodological weakness of his 1980 book, but persisted in his rejection of a PN family: “The ‘PamaNyungan’ idea … is totally without foundation and must be discarded if any progress is to be made in studying the nature of the linguistic situation in Australia” (2002: xx); “[t]he putative division between ‘Pama-Nyungan’ and ‘non-Pama-Nyungan’ (either Mark I or Mark II) has had a deleterious effect on the study of Australian languages” (2002: 53). Blake (1988, 1990b) argues the evidence of distinctive case suffixes and personal pronouns in support of PN and discusses the changed classification of certain languages. Evans (1988) more emphatically argues that these facts are indications of shared innovations, and postulates a PN phonological innovation that shifts apical nasals and stops to laminals. Alpher (1990) presents evidence from verb inflection that proves the distinctiveness of PN morphology. Alpher (2004) puts the reconstruction of PN phonology on a sounder base, presenting the sound correspondences that justify the pPN phonemic inventory. Koch (2003) cites the evidence of case inflections in personal pronoun paradigms that justifies a nuanced case morphology reconstructible to pPN. Evans (2005: 261–272), Bowern (2006), and Sutton and Koch (2008: 485–491) provide arguments for the PN hypothesis against the interpretation of Dixon (2002). Miceli (2004) makes the important distinction between attempting to justify PN as a subgroup, by finding common innovations from an earlier (and even more difficult to reconstruct) proto-language, versus establishing it as a family, using the kind of individuating-identifying features that Nichols (1996) advocates. Her own position is that “at present, PN cannot be considered an established genetic entity because the evidence so far proposed in its favour relies too heavily on the assumption that all Australian languages are related.”³⁹ Miceli (in press) suggests as a reason for the varying interpretations of the evidence for Pama-Nyungan the possibility that transmission in the Australian multilingual context may not have resulted in the same tree-like pattern that historical linguists have been accustomed to seeing as the evidence of genetic relationship.

39 A reviewer points out that all that is required to establish the separateness of PN is the existence of at least one other Australian language (or language family) that is not PN but is related to PN.

 59

Historical relations among the Australian languages 

Campbell and Poser’s (2008) world-wide survey of language classification, nevertheless, represents the currently prevailing position as favouring PN genetic entity (pp. 145–161).

4.3.2 Summary of the evidence for Pama-Nyungan The evidence for pPN is of three kinds: lexical, phonological, and morphological. Lexical evidence was the most overt kind of evidence used in the positing of the PN family. Particular weight was put on the contrast between the number of cognates detectable among the PN languages vs. those shared with languages outside the group. The citable pPN lexical reconstructions, however, number only in the hundreds, rather than the thousands O’Grady often expressed the hope of finding. Australian linguistics has suffered from the fact that the actual cognates are rarely provided, and there has been a frustrating lack of readily available cognate databases. Alpher (2004: Appendix 5.1), however, provides some 170 PN cognate sets. Claire Bowern’s PN project at Yale University is building up a database of PN cognates (see http://pamanyungan.sites.yale.edu/). The phonological systems of PN languages are not very different from those of nPN languages, and many languages have nearly identical phoneme systems and phonotactics. There has been a tendency to assume that the proto-system corresponds to the predominant synchronic pattern(s), without demonstrating this by the comparative method. Alpher (2004) is the first attempt to establish the phonology of pPN by an explicit application of the comparative method; i.e. by citing the correspondence sets that emerge from sets of cognates. The pPN phoneme system is taken to include three vowels, with a length contrast in the first syllable, and a consonant inventory much like the maximal system of many PN languages displayed in Table 3. The reconstructed phonotactics is taken to be like the norm for Australian languages: words begin with a single consonant, allow roughly similar word-internal consonant clusters, and end with vowels or a restricted set of (usually sonorant) consonants. Table 2.3: Maximal PN consonant system

stops nasals laterals tap/trill approximants

labial

laminodental

apicoalveolar

retroflex

laminopalatal

velar

p m

th nh lh

t n l rr

rt rn rl

ty ny ly

k ng

r

y

w

An issue has been whether pPN distinguished retroflex and apico-alveolar places of articulation (word-internally only, since few languages allow a contrast initially).

60 

 Harold Koch

O’Grady (1990a: xxi) posits the contrast and Alpher (2004: 112) accepts that it is probable. Dixon’s suggestion (1970c, 1980: 155–156) that retroflexes developed after the vowel /u/ is contradicted by PN cognate sets reflecting *pula ‘(they) two’ and *kuna ‘excrement’, which lack the phonemic retroflex rl and rn.⁴⁰ Plausible pPN words containing retroflex consonants are *ngurlu ‘forehead’ (vs. *kuli ‘angry’, *ngula⁴¹) and *parnti- ‘smell’ (reconstructed in O’Grady 1990c: 251) vs. *wanti- ‘fall’. There is still no consensus on whether pPN had a laminal contrast (cf. Alpher 2004: 112–115). Koch (2003) reconstructs for pPN contrasting pronouns *ngathu ‘1SG.ERG’ and *ngatyu ‘1SG.DAT’, which implies a laminal distinction. Further cognates are required, however, to convince most linguists that this represents a feature of pPN.⁴² It is not yet clear whether the laminal contrast, if it existed in pPN, occurred word-initially.⁴³ Another not fully resolved issue concerns the apical vs. laminal contrast in wordinitial position. Evans (1988) argued that pPN has innovated vis-à-vis nPN languages by changing initial apical stops and nasals to laminals. He allows that pPN had a single apical initial consonant that he represents as RLD. Alpher (2004: 115), however, reconstructs for pPN four contrasting initial apicals, *n, *t, *l and *r, although the number of lexemes instantiating each is very small. Some of his best supported examples are *nuka ‘ankle’, *taaku ‘ground’,⁴⁴ *rirra ‘tooth’, *lulku ‘heart’.⁴⁵ Initial apical consonants have become laminals or merged with one another in a large number of subgroups; these changes could be further exploited as diagnostic innovations of particular subgroups (cf. lateralisation of *r- in Ngumpin-Yapa demonstrated in McConvell and Laughren 2004). The consonant clusters reconstructible for pPN, according to Alpher (2004: 118–119) include: homorganic nasal plus stop, n/l/rr plus p/k/ty, and triconsonantal l plus mp/ngk/nyty and rrngk. Especially significant is the contrast between homorganic nyty and heterorganic nty. The morphological evidence for pPN consists of: case suffixes on nominals, grammatical words (personal pronouns, interrogatives, deictics) and their case paradigms, some nominal derivational suffixes, and distinctive TAM inflections on verbs, which

40 Cf. Dixon (2002: 586), where it is acknowledged that cognates no longer support the earlier claim. 41 Alpher (2004: 503–504) reconstructs this as an adverb ‘by and by’, while Koch (2009: 338–340) treats it as the Locative of a distal demonstrative, with a meaning ‘there’ or ‘then’. 42 Dixon (1980: 341) derives the Dative forms from a reconstructed *ngayku, analysed as the regular (for nouns but not pronouns) Dative suffix *-ku added to a stem *ngay; this *ngayku is supposedly reflected in Dyirbal, Wargamay, Nyawaygi, among others. This solution is also suggested in Alpher, O’Grady and Bowern (2008: 16 note 5), whereas Koch (2003) implicitly derives Wargamay ngayku from *ngatyu by sound change — which admittedly requires further support. 43 Dixon’s (1980: 334, 343) reconstruction of 2DU *NHumpVlV and 2PL *NHurra vs. 2SG *NYun (beside *ngin) implies a possible contrast. 44 With apparent apical-initial cognates in nPN languages (Alpher 2004: 546). 45 Hendrie (1990) likewise reconstructs four contrasting pPN apicals, and presents no less than 96 presumed cognate sets from nine PN languages in support of his reconstructions. His case is hampered, however, by the weakness of many of his etymologies.

Historical relations among the Australian languages 

 61

differ for a number of inflectional classes. The accepted case suffixes on nouns are indicated in Table 4. The original conditioning of the allomorphy of Ergative and Locative is not clear. Many languages support a selection according to the length of the stem, with *-ngku after disyllabic (or bimoraic) stems and *-lu after longer ones. Sands (1996: 12–24) argued instead for an original distribution (like that of the Western Desert language) determined by the proper vs. common status of the noun. The original conditions for the Table 4: Reconstructed Pama-Nyungan case allomorphs for nouns Ergative Locative Nominative Accusative Dative Ablative

*-ngku *-ngka *-Ø *-Ø *-ku *-ngu

*-lu *-la

*-Tu46 *-Ta

*-thu

*-nha47

⁴⁶⁴⁷ allomorph *-thu (emphasised by Sands 1996 and Dixon 2002: 158–159) are unclear. Australianists have speculated about the origins of the formal relations between the allomorphs with ngk vs. l (Hale 1976, summarised in Dixon 1980: 318–320) and the reason for the formal relation between the Ergative and Locative formatives (Dixon 1980: 318). These questions are beyond the scope of comparative PN reconstruction, however. Pronouns differ somewhat in their case inflection: Blake (1988: 27) comments: “There are usually declensional differences between nouns and pronouns, the latter sometimes making a slightly different set of distinctions, often having different markers, and sometimes using an oblique stem for some or all of the non-core cases.” A distinctive set of non-singular personal pronouns (shown in Table 5) has been reconstructed to pPN (Blake 1988, 1990b).⁴⁸

Table 5: PN non-singular personal pronouns

st

1 2nd 3rd

Dual

Plural

*ngali *nhumpala *pula

*ngana *nhurra *thana

46 T is a homorganic stop after a nasal. 47 Occurs on stems denoting persons. 48 These forms have been reconstructed to pA in Dixon (1980); in Dixon (2002) their reflexes are attributed to diffusion.

62 

 Harold Koch

The singular personal pronoun paradigms shown in Table  6 can be reconstructed with some confidence for pPN (Koch 2003, 2013). The distinctiveness of irregular forms from those of the normal noun paradigms (1SG -thu, -y, -tyu, 2SG/3SG.F -a) adds richness to the pPN reconstruction. These differences, however, have been obscured by a tendency to interpret them as reduced (by unsupported sound changes) from combinations of regular stems and regular suffixes: e.g. 1SG-ERG *ngay-thu, 1SG-ACC *ngay-nha, and 1SG-DAT *ngay-ku, 2SG-ACC *ngin-nha (Dixon 1980: 340–341), 2SGDAT *ngin-ku (Dixon 2002: 317), 3SG.F-ACC *nhan-nha (Dixon 2002: 305). Table 6: Reconstructed Pama-Nyungan singular pronouns

Ergative Nominative Accusative Dative

1SG

2SG49

3SG.F

3SG.M

*ngathu *ngay *nganha *ngatyu

*nyuntu *nyun *nyuna *nyunu

*nhantu *nhan *nhana *nhanu

*nhulu *nhu *nhunha *nhungu

Other pronominal paradigms that can be reconstructed for pPN (interrogative-indefinites, anaphorics, and deictics) are shown on Table 7 (see Dixon 1980, 2002 for the interrogatives; Koch 2009 for *pa- and *ngu-).        ⁴⁹ Table 7: pPN pronominal paradigms

Ergative Nominative Accusative Dative Locative

‘who’

‘which’

‘that’

*ngaantu *ngaan *ngaana *ngaanu

*wanhthu *wany *wany

*palu *pa *pa(nha) (*parnu)51 *pala

50

*wanhtha

‘yon’

*ngu

*ngula

⁵⁰ ⁵¹ A few derivational nominal suffixes are reconstructible. Evans (1988: 94) calls attention to a deverbal nominalising suffix *-nhtha. McConvell (2008) reconstructs a kinship suffix *-nhtharr, whose original function remains unclear. Another possibility, *-karra is suggested by cognates meaning ‘one of a pair’ in Dyirbal and other eastern languages (Dixon 1980: 323), dyadic marker in western languages (Dench 1997: 110–

49 The 2SG stem is *ngin in a large block of eastern PN languages; it is unclear whether this or *nyun should be reconstructed. 50 A reflex of the expected *wany-ku is apparent in the Nyawaygi Allative wayngku beside Locative wanytya (cited in Koch 1995: 48). 51 A (human) Dative *parnu is reconstructible to high levels but perhaps not all the way to pPN.

Historical relations among the Australian languages 

 63

112), complementiser for same subject concurrent subordinate clauses in Warlpiri (cf. Simpson 1988), and a manner adverbial suffix in Ngumpin languages.⁵² A considerable number of cognate verbs are reconstructible, and they belong to inflectional classes that appear to be cognate. Inflectional classes differ especially in their characteristic consonants. These have often been described as “conjugation markers”, although the same consonant rarely occurs in all inflectional suffixes. Dixon’s claims (1980, 2002) that these consonants were originally part of the verb root and that they were followed by invariant suffixes is not sustainable (Alpher 1990, Harvey 2008b, Koch 2014). Of special interest are widely attested monosyllabic verb roots with comparable inflections that are irregular within their individual verbal systems. Alpher (1990) justifies an n/rr paradigm (Imperative *-rra, most other inflections including *n) for *ya- ‘go’, *maa- ‘get’, and *thu- ‘put’. Koch (2014) discusses the *ng/w paradigm of *kaa- ‘carry’ and *nhu- ‘give’ and the slightly aberrant *nhaa‘see’, whose paradigm has Imperative *nhaaka (vs. *kaangka and *nhungka). Dixon (1980) had called attention to the recurrent *m of *pu- hit’. The more heterogeneous system argued for by Alpher and Koch makes the inflectional paradigms of pPN verbs more similar to those of nPN languages, including those reconstructed for ProtoGunwinyguan by Alpher, Evans and Harvey (2003) and Proto-Arnhem reconstructed by R. Green (2003). Verb inflection among the PN languages needs more careful study, along with the etymological study of verbs.

4.4 Relation of Pama-Nyungan and non-Pama-Nyungan families A fundamental issue of Australian historical linguists concerns the relationship between the PN and nPN languages. Evans (2003a: 5–11, cf. Evans 2005) describes four models of classification that have been offered in the literature and gives them labels. (1) The “rake model”: in the original classification of OVV — based on lexicostatistics and proposed as a provisional genetic classification to be confirmed by the comparative method — the “Australian phylum” consisted of 29 families, one of which was PN — with no higher-order grouping. (2) The “diffusion model for nonPama-Nyungan”: in Dixon’s (1980) approach, PN is accepted only as a label for a typologically conservative class of languages, which do not form a genetic unit, and the bulk of nPN languages are considered to have undergone structural changes which resulted in a more polysynthetic, head-marking structure, including verb prefixation. These typological changes, however, are not considered to have been common innovations which define subgroups, but to have diffused across the north-central parts of the continent. Evans (2005: 277) claims “that the balance of evidence has

52 Forms like -thi(rr(i)), -tharri, -tyarra mentioned by Dixon (2002: 170) as widespread markers of a comitative relation have yet to be shown to be cognate.

64 

 Harold Koch

now reached the point where the ‘diffusion model’ of similarities between nPN prefix systems can no longer be sustained”, and in this work confines his attention to the next two models. (3) The “binary model” is implied by Heath, who proposes reconstructions of noun class prefixes, verbal pronominal prefixes and verbal suffixes to a “Proto-Prefixing” (Heath 1987, 1990), which would be Proto-non-PamaNyungan. Likewise Blake (1988, 1990b) reconstructed for the “Northern” languages a set of free pronouns that are distinct from but coordinate with a pPN set.⁵³ This division into nPN and PN genetic groups largely matches Capell’s (1956) division into prefixing vs. suffixing typological groupings, although we now recognise some non-prefixing nPN languages (e.g. Kayardild) and at least one prefixing PN language (Yanyuwa). (4) The “Pama-Nyungan offshoot model” is argued to provide the best fit with the facts (Evans 2003a: 9–11). This model is implicit in O’Grady’s (1979) paper, where he suggested that the set of languages called PN in OVV, now relabelled as “Nuclear Pama-Nyungan”, branched off, in a family tree, from a higher node — a revised Proto-Pama-Nyungan — which also included the Gunwinyguan and Tangkic branches. Garrwan is also included as a close sister of PN in Evans and Jones (1997). In the PN offshoot approach, nPN languages assume a greater importance for the reconstruction of pA than does pPN, many of whose features turn out to be innovations and some of which have origins which can be sought in the nPN languages. The research program suggested by this model includes increased attention to reconstruction and more refined classification within the nPN language groups and a willingness to seek explanations of pPN features from material reconstructed from the nPN languages. As an example of the last point, Alpher et al. (2003) and Harvey (2008b) propose that the seeds of the PN conjugational system can be found in the inflectional system of Proto-Gunwinyguan verbs; i.e. consonants that were part of certain inflectional suffixes came to be characteristic markers of whole inflectional classes. Likewise, Sutton and Koch (2008: 487–488) present a scenario that derives formal aspects of the pPN personal pronouns (such as the *-rra in *nhurra 2PL and the *-la in 3DU *pula) from number-markers in the nPN pronouns. One of the most intriguing questions raised by this classification is whether the common original language had pronominal (subject-marking) verbal prefixes and/or noun-class prefixes like the majority of nPN languages: if so, by what mechanisms were these features lost and do any traces remain in non-prefixing languages?

53 The classification of the Garrwan languages, Garrwa and Waanyi, remained indeterminate according to Blake’s pronoun criterion, since they show pronouns from both sets.

Historical relations among the Australian languages 

 65

4.5 Relatedness of nPN language families 4.5.1 Pan-nPN features? A problem that needs to be addressed is whether more genealogical order can be found among the twenty or so nPN language families than was implied by OWH’s treating them as being related only in an Australian “phylum”, which by their definitions implied that there were too few available cognates to do proper comparative reconstruction. Capell’s “Common Australian” included some vocabulary that is shared across most of Australia, including PN languages. Dixon (1980: 426–430) emphasised the importance of a number of monosyllabic verb roots that can be recognised in nPN as well as PN languages and hence support the genetic unity of all (or most) Australian languages — including *ka- ‘carry’, *ma- ‘do’, *na- ‘see’, *pu- ‘hit’, *ru- ‘cry’, *wu‘give’, *ya- ‘go’. Blake (1988, 1990b) compared free personal pronouns across the nPN languages and proposed the reconstructions shown in Table 8. Table 8: Blake’s reconstructed “Northern” pronouns

1 1+2 2 3

Singular

non-Singular

*ngay *nya *nginy *nu, *ngaya (fem)

*nyi-rrV *nga-rrV *nu-rrV, *ku-rrV *pu-rrV

If we are to rely on cognate vocabulary, however, there are still probably too few cognates to give a solid demonstration of genetic relationship. But the morphological subsystems can be exploited to further the case. Harvey (2003b) argues that bound pronominals provide stronger evidence of cognation than Blake’s free pronouns, and provides tentative reconstructions for a set of Minimal and Augmented⁵⁴ bound pronominals that occur as prefixes or proclitics on verbs. Evans (2003a: 19) presents a table which demonstrates comparable verb subject(-object) prefixes across seven representative nPN languages. He also points out that most nPN languages exhibit a presumably inherited verb structure: Subject/Object/Future-ROOT-Reflexive/ReciprocalTense/Aspect/Mood. He illustrates the resemblances by comparing a few forms from Mawng and Wubuy (Nunggubuyu), two languages at opposite ends of Arnhem Land. Table 9 adapts his example (from p. 18), showing the prefixes which occur with cognate inflections of the verb -wu- ‘hit’ and which allow the reconstruction of the elements *ngan- 1SG.OBJ, *ni- 3SG.M.SBJ, *nga- 3SG.F.SBJ, *pa- FUT.

54 Minimal and Augmented are identical to Singular and Plural except in the case of the inclusive 1+2 (speaker plus hearer) person, where Minimal refers to two and Augmented to three or more.

66 

 Harold Koch

Table 9: Mawng and Wubuy (Nunggubuyu) verb prefixes compared

Mawng Wubuy

me.he-(hit.PST)

me.he-(hit.FUT)

me.she-(hit.PST)

me.she-(hit.FUT)

nganingani*ngan-(n)i-

nganpaningampani*ngan-pa-ni-

nganngangangi*ngan-nga-

nganpangangampangi*ngan-pa-nga-

Another morphological subsystem that is worth investigating from a comparative point of view is noun class or gender prefixes. This evidence, as well as that of verb prefixes, is flagged by Evans (2003a: 20–21): “Right across the nonPN area one finds evidence of descent from an ancestral system of at least four, possibly five classes: masculine (I), feminine (II), vegetable (III), certainly one neuter (IV) and possibly another (V).” For most classes there are two forms, which in some languages are distributed according to the case of the noun. In some languages these prefixes are found only as fossilised parts of certain lexemes, and in some languages (notably in the Mirndi family) they have become suffixes through univerbation with erstwhile classmarking demonstratives. Evans (p. 21) presents a sample of matching prefixes in ten nPN languages. Another morphological subsystem that is worth exploring across nPN families is that of verb TAM suffixes. R. Green (2003) compared the inflectional systems of 24 cognate verbs across 13 languages (without counting the core members of the Gunwinyguan family) of Arnhem Land and reconstructed their paradigms for a high-level proto-language she tentatively called Proto-Arnhem.

4.5.2 Discovery of higher-level groupings A number of nPN languages or groups have been proved to belong to higher-level groups. The number of nPN families has thus been reduced from OWH’s 28 to a figure closer to 20. The Mirndi (or Mindi) family encompasses three of OWH’s families: Djamindjungan (Jaminjung, Ngaliwurru, Nungali), Jingiluan, and Wambayan — see Chadwick (1984, 1997), Green (1995), Green and Nordlinger (2004), Harvey et al. (2006), Harvey (2008a). Ian Green (2003) posited a Southern Daly family consisting of Murrinh-Patha plus Ngan’gityemerri with its Ngan’gikurunggurr dialect, based on the evidence of detailed agreements in the verbal auxiliary paradigms. Rebecca Green (2003) established a “Maningrida” family consisting of Gunavidji (Ndjebbana), Nakkara, Burarra and Gurr-goni — on the evidence of shared innovations in verb inflection from a reconstructed Proto-Arnhem. Marie-Elaine van Egmond (2012) has made a solid case for the membership of both Anindhilyakwa and Wubuy (Nunggubuyu) with Ngandi in the large Gunwinyguan family of Arnhem Land — on the evidence of shared basic vocabulary, systematic phonological correspondences, and shared verbal suffixes.

Historical relations among the Australian languages 

 67

4.5.3 Relations between individual nPN families Rebecca Green’s higher-level “Arnhem family” encompasses at least three separate families — Gunwinyguan, Maningrida, and Marran; Evans (1995: 28) adds Iwaidjan and the so-called isolates Umbugarla, Gaagudju, and Mangarrayi. Merlan (2003) provides evidence from nominal prefixes linking Mangarrayi to the Marran family. Evans (1995: 34) notes that the Tangkic languages share with languages of the Roper River area (including Marran and Mangarrayi) the number-marking system of personal pronouns (rr marking dual and l signalling plural) as well as their verbal suffixes -yi- Reflexive and -nytyV- Reciprocal with those of Gunwinyguan and other Arnhem Land languages (p. 37–38). Harvey (2003c) compares the structure of verbs in the Eastern Daly family (Kamu and Matngele) with that reconstructed for Proto-Gunwinyguan. Harvey (2008a: 128–129) points out cognate material between the Mirndi family and Arnhem languages to the north, and Breen (2003) indicates lexical and morphological relations between the Garrwan languages (Garrwa and Waanyi) and those of the Mirndi family.

4.5.4 Prospects for progress in comparative nPN Non-PN families in which considerable phonological reconstruction has been done are: Iwaidjan (Evans 1998, 2000), Gunwinyguan (Harvey 2003a), Worrorran (McGregor and Rumsey 2009), and Tangkic (Round 2009, 2010). Issues that remain to be resolved include: the number of vowels in the various proto-languages (whether three like pPN or perhaps four or more), the origin of the stop contrast involving voicing or length in Arnhem Land languages, and the origin of glottal stop in Arnhem Land languages. Explicit reconstructions of more proto-phonologies are required — to provide the basis for discussing shared innovations — as well as more explicit diachronic descriptions. More comparative work needs to be done on the morphological subsystems of the nPN families – paradigms of verb TAM suffixes, pronominal prefixes on verbs, and nominal gender-class prefixes. It is now worthwhile to pursue the historical relations among nPN languages using three approaches: (a) bottom-up reconstruction of the proto-languages of accepted families, (b) (top-down) comparison across the nPN families, along the lines signalled in Evans’ (2003a: 17–21 “Evidence for deeper-level relatedness”), and (c) diachronic studies of how structural changes take place; e.g. the loss, renewal, and creation of noun class systems, person-number cross-referencing, bipartite verb patterns, and changes involving prefixation and suffixation (e.g. Green 1995).

68 

 Harold Koch

4.6 Higher-level subgroups of PN 4.6.1 Nyungic The OWH classification included a large Southwest or Nyungic group, which encompassed all the PN subgroups of Western Australia plus their Nangga (i.e. Wirangu) and Yura (i.e. Thura-Yura) of South Australia. Although widely cited, this lexicostatistically based grouping has not been demonstrated by the evidence of common innovation, and so has remained a mere hypothesis of genetic relation.⁵⁵ Bowern and Atkinson’s (2012) study, however, gives new life to a version of this hypothesis (see § 4.6.4).

4.6.2 Victorian The languages of Victoria and adjacent cross-border regions have been the subject of considerable consolidation, beginning with Hercus (1969, 1986). Eleven languages are recognised in the classification by Blake and Reid (1998). All except the Gippsland language (Gunnai and Bidawal) have received consolidated accounts — as a monograph or long article. It might now be worth investigating whether a Victorian subgroup can be established. The language situation is somewhat complicated by the fact that the wide-spread Kulin subgroup predominates in the central part of the region. There are, however, apparent cognates that are found on both sides of this Kulin block (Blake and Reid 1998: 15–16). An interesting question is, if a Victorian subgroup can be justified, whether the Lower Murray languages (cf. Horgen 2004) are to be included.⁵⁶

4.6.3 Eastern Pama-Nyungan Dixon and Blake have occasionally mentioned some features that are characteristic of eastern (PN) languages — although an Eastern PN has never been demonstrated as a higher-level subgroup of PN. Their suggestion (Blake 1988: 6, Dixon 2002: 461) that the 3SG.F pronoun *nhan is specifically eastern needs to be amended in the light of cognate forms (albeit without the specifically feminine sense) in northern and western subgroups (Koch 2013). Dixon’s (1980: 359–361) suggestion that the 3SG.M stem *nhu is specifically eastern and corresponds to *ngu- in western languages also needs

55 O’Grady (1966: 103–111) gives 174 tentative Nyungic cognates, but only five mention Yura representatives. 56 Bowern and Atkinson’s (2012) study, described in § 4.6.4, finds support for a Victorian group which includes the Lower Murray languages, as well as a macro-Kulin and an Eastern Victorian group.

Historical relations among the Australian languages 

 69

correction: in Koch (2009) I give evidence that *ngu- is in origin a separate, demonstrative stem and that *nhu- is in fact attested in some western languages alongside *ngu-. A better candidate for eastern PN is the 2SG pronoun stem Dixon (1980: 344) reconstructs as *ngin; the variant stem *nyun, attested in many other PN languages, however, cannot be interpreted as a true cognate (see Alpher 2004: 124–125, who reconstructs *nyun as the pPN (p. 511–515)). This solution leaves ngin as possibly a south-eastern relic (it resembles Blake’s nPN *nginy), a shared innovation, or a set of forms to be explained by independent developments. Dixon (1980: 376) discusses the indefinite-interrogative *minha, which in all of south-eastern Australia (southward from Dyirbal in the Cairns rainforest) means ‘what’ or ‘something’, but in far north Queensland means ‘animal’ or ‘meat’. Dixon plausibly suggests that the direction of semantic shift was from ‘animal’ to ‘something/what’.⁵⁷ This suggests that the south-eastern languages have shared a potentially subgroup-defining semantic innovation.⁵⁸ Further wide-spread eastern forms are: ‘lie’ *yuna- (vs. western *nguna-), ‘give’ *wu- (vs. western *nhu-), ‘eat’ *tha- (vs. western *nga-), ‘beard’ *yarrany (vs. western *ngarnka⁵⁹), ‘leg’ *puyu, ‘thigh’ *tharra (Dixon 2002: 107 notes its eastern distribution), ‘urine’ *kil (vs. western *kumpu). All such distributions deserve to be studied, with a view to determining whether each (a) is indeed attested only in the (south-)east, (b) represents an original form replaced in other areas, or (c) constitutes a common innovation.

4.6.4 Primary branches of Pama-Nyungan Linguists for the most part have avoided the problem of determining the primary splits of a putative pPN, and have been content to leave the classification at some thirty or more subgroups (including isolates). Bowern and Atkinson (2012) address this issue with the help of Bayesian computational phylogenetic methods. Their study was based on lexical cognate data (with cognates said to be established by the comparative method) representing 189 glosses across 194 languages. The results included a number of high-level splits. A western group approximating OWH’s Nyungic (but without the Thura-Yura languages) was clearly indicated, with an important new distinction between a northern set including the Wati (Western Desert), Marrngu and Ngumpin-Yapa subgroups and a southern set which takes in the Ngayarta, Kanyara-

57 A parallel can be seen in some of the Yolngu languages, where a form *yuul ‘person’ (preserved in the extended form yuulngu) has changed to ‘someone/who’, replacing the inherited *wara, which is preserved in other Yolngu languages (Ritharrngu, Djinang). 58 Dixon’s (2002: 334) explanation is that this is likely to be an instance of the “grammaticalisation [of a generic noun] diffusing over a continuous area”. 59 Alpher (2004: 492) reconstructs *ngarnka as pPN.

70 

 Harold Koch

Mantharta, Kartu, and Nyungar/Mirniny languages. This westerly group in turn was strongly linked with a more easterly branch consisting of the Yolngu and Warluwarric subgroups, to collectively constitute a highest-level western branch of PN.⁶⁰ Also strongly supported were a south-eastern group with two subdivisions, a Victorian branch and a branch consisting of most of the languages of NSW and south-eastern Queensland. Two further large groupings are identified: a south-central group (including Arandic, Thura-Yura, Karnic, Yarli, Paakantyi and a few fringe languages), and a north-eastern group with includes Pama-Maric and the Western Torres Strait language (Kala Lagaw Ya), the Mayi subgroup, and Kalkatungu-Yalarnnga. The evidence is not unambiguous about how these last two groups are related to each other or to the large western and south-eastern groups. These results are important in that they appear to show that the PN family is not in principle different from other language families and is not beyond the reach of historical methods. Computational methods further can handle vastly more data than individual linguists can keep in our heads. Australianists should now explore these new hypotheses, to see whether specific innovations of phonology, morphology, or lexicon can be identified to support or possibly refute the suggested groupings. For example, initial apicals, now reconstructed for pPN (Alpher 2004), have been replaced by laminals in most of the languages of the south-western group: could this be a subgroup-defining innovation? If two sets of laminal consonants are reconstructed for pPN, their merger into a single set would be diagnostic of the northern branch of the Western Australia group.⁶¹

5 Language relations and prehistory Answers to questions concerning the historical relations between the languages of Australia — relations of both genealogy and contact — raise further questions concerning the human prehistory of the continent. The geographical distribution of language families presupposes earlier movement of speakers and /or the expansion and contraction of linguistic territory as a consequence of some kind of social dominance (whether demographic or cultural). The presence of loanwords and other borrowed linguistic material is an indication of earlier contact between the speakers of different languages.

60 A relationship between Warluwarric and Ngumpin-Yapa was noted in McConvell and Laughren (2004), but assumed to reflect early contact. O’Grady and Fitzgerald’s classification (1997: 344) includes a “Nyungo-Yuulngic Group”, consisting of OWH’s Nyungic plus the Yolngu languages, but with Yolngu subgrouped most closely with the Kanyara-Mantharta languages (the Warluwarric languages are not mentioned at all). 61 For a critical assessment of Bowern and Atkinson’s (2012) methods see Miceli (in press).

Historical relations among the Australian languages 

 71

5.1 Geographical distribution of language families 5.1.1 The spread of Pama-Nyungan The postulation of the large PN family covering most of the mainland has given rise to questions of when it was spoken, where it spread from and what factors might have motivated its spread. McConvell and Evans (1997b: 14–15) summarise the “PamaNyungan enigma” as follows: Pama-Nyungan languages are spread in relatively homogenous fashion over seven-eighths of the continent, in contrast with the diverse mosaic of their non-Pama-Nyungan relatives, which are confined to the Kimberley, Top End and Barkly regions. This distribution suggests a relatively recent spread of Pama-Nyungan languages at the expense of other languages once spoken there.

5.1.1.1 When was it spoken? The extensive, nearly continent-wide distribution of just one of the language families is usually taken to indicate that its spread took place relatively recently — in comparison to the 50,000 years of human occupation of the continent. There is no secure method of judging the age of pPN; estimates that have been provided are based on a comparison of the degree of linguistic diversity with that of language families whose age is better known, such as Romance or Indo-European.⁶² Evans and McConvell (1998: 189) suggest 4,000–5,000 years ago; O’Grady and Hale (2004: 91) claim “a time depth of 5,000 years or less”; McConvell (1996) suggests a date of 6,000 BP.

5.1.1.2 Where was it spoken? The prevailing view accords with the claim of Hale (1962) that the languages spread from somewhere in the region of northern Australia west of the Gulf of Carpentaria.⁶³ This is consistent with the usual assumption, in accordance with Dyen’s (1956) “migration theory” (Trask 2000: 213, cf. Anttila 1989: 389–390) that the area of greatest diversity represents the oldest location, since it requires the fewest movements. This location places pPN in the northern part of the continent, which is the most diverse linguistically, and in the vicinity of the nPN language families with which PN is as-

62 O’Grady and Fitzgerald (1997: 347) base a pPN date of 3,500–4,500 BP on the fact that PN languages show a slightly lesser degree of diversity than the Finno-Ugric languages, whose ancestor is imputed to 5,000 BP. 63 Hale contrasted the situation in Cape York Peninsula, east of the Gulf, where languages can be shown to be relatively closely related, once allowance is made for recoverable sound changes.

72 

 Harold Koch

sumed to be most closely related — Garrwan, Tangkic and Gunwinyguan — according to the offshoot model (O’Grady 1979; Evans 2003a: 10, 2005: 277). Evans and Jones (1997: 404) specify the homeland more precisely as “the area stretching between the Roper River across the Barkly Tableland into north-western Queensland”.

5.1.1.3 How did it spread? — Phases and direction Evans and McConvell (1998) propose a model of PN expansion that involved at least three phases. First pPN expanded from north-western Queensland throughout eastern Australia c. 5,000–4,000 BP in a social process that included a shift of language on the part of previous inhabitants in coastal and riverine areas. A second phase involved a “move out from riverine bases in Central Queensland about 3,000–2,000 BP to reoccupy the arid interior which had been largely abandoned” (p. 186), followed by later expansion into already occupied peripheral areas. The most recent expansion was that of the Western Desert language during the last millennium “to cover a vast area of the most arid and inhospitable part of Australia” (p. 187). There is a consensus that the direction of this last spread was from west to east.

5.1.1.4 Why did it spread? Unlike in many other parts of the world, language expansion in Australia does not correlate with the spread of agriculture (Bellwood 1997), since the people all had a hunter-gatherer economy. Linguists have looked to archaeology for evidence of technological or cultural change in the mid-Holocene period. Much mention has been made in the literature of correlations between the assumed date of the break-up of pPN and changes noted by archaeologists, including: the spread across the whole continent except for northern areas of new kinds of microliths (small stone tools), the introduction of the dingo (Asian dog),⁶⁴ the exploitation (by complex processing) of new food sources such as cycads and macrozamias, the development of seed-grinding economies, an increase in population density,⁶⁵ growing exploitation of marginal environments, the expansion of trade networks, changes in rock art styles (see Evans and Jones 1997: 406–411, Evans and McConvell 1998: 179–182, Sutton and Koch 2008: 499). Attempts have been made to imagine social mechanisms that could link new lithic technology to language spread. The most elaborate scenario is given in Evans and Jones (1997: 413–417), who propose that Pama-Nyungan might have spread through induction of new recruits into the ceremonial aspects of a new quartzite technology

64 Curiously, the possible role of the introduction of the dingo on population and/or language shift has hardly been discussed. 65 This is now in doubt (Hiscock 2008).

Historical relations among the Australian languages 

 73

by its possessors. In a somewhat similar vein O’Grady and Hale (2004: 92) suggest as a factor in language expansion “the spread of intellectual wealth, including, for example: ever more complex kinship systems; intricate and demanding verse in song cycles; auxiliary languages … together with an impressive battery of ritual law, iconography, itineraries and maps”. Other suggested social explanations include new forms of social organisation that are characterised by links (of marriage, ceremony, ideology, and trade) between different groups (Evans and Jones 1997: 411–413, Evans and McConvell 1998: 183–187). Environmental factors have figured in attempts to explain social (and indirectly linguistic) change. Australia is a dry continent, including regions with low and unpredictable rainfall, a long coast, and a number of fertile river systems. Sutton (1990) suggests a “pulsating heart” model, whereby populations typically exert pressure in the direction from a poorer to a richer environment. This involves several separate directional pressures: from desert plains toward water-bearing hills, from arid undifferentiated drainage areas toward drainage systems, from hinterlands toward rivers, from upriver toward downstream, from inland toward the coast, and from coastal inter-riverine areas toward river embouchures. Also implied are some centrifugal movements in the opposite direction, to exploit hinterland resources and perhaps to relieve social pressures. It is widely assumed that populations spread out into marginal areas when climatic conditions improve, and retreat into more favourable areas (or perish) when weather conditions deteriorate. McConvell (2001) develops a somewhat similar model, applying Johanna Nichols’ (1992) idea of spread and residual zones. He claims that languages can spread as hunter-gatherers move “upstream” into sparsely populated (and less resource-rich) areas, expand through such an area, and eventually reemerge “downstream” into more densely populated and more resource-rich residual zones, perhaps with language shift. A link can be made between the proliferation of small tools about 4,000–5,000 years ago in southern Australia and climate change. According to Hiscock (2008: 156–160), the increased production of multifunctional tools was a strategy to reduce the economic risk: “foraging risk increased with the onset of drier and more variable climatic conditions across southern Australia as the El Niño system intensified” (p. 158). Sutton and Koch (2008: 499–500) explicitly relate such climate changes to the expansion of languages.

5.1.2 Distribution of Pama-Nyungan subgroups Certain inferences concerning population history can be drawn from the extent and relative uniformity of languages and linguistic subgroups. The Western Desert language, consisting of a chain of mutually intelligible dialects spoken over a huge area of mostly desert country in the west of Australia, is understood to result from a relatively recent expansion concomitant with repopulation of the desert from the

74 

 Harold Koch

west during the last two millennia, for which there is support from archaeology (Veth 2000, Smith 2005) as well as human biology (McConvell 1996⁶⁶). Other subgroups that cover considerable area but manifest a low degree of linguistic diversity — and therefore can be assumed to reflect relatively recent expansions — are Maric in Queensland largely west of the great divide, the Central New South Wales subgroup in the western plains west of the Dividing Range (for these two, cf. Sutton and Koch 2008: 500), the Kulin subgroup of central and western Victoria, and the Ngumpin-Yapa subgroup of the northern part of the desert region of Western Australia. The languages of a higher-level group, included in OWH’s “Nyungic” or Bowern and Atkinson’s “Western”, have been considered to display a degree of uniformity that suggests a rather late expansion within the overall time-frame of PN (McConvell 1996).

5.2 Pama-Nyungan proto-culture expressed in language The available pPN lexical items provide little information that would help to locate its speakers in space or time. The main life form term reconstructible is *kuya ‘fish’ (Dixon 2002: 102), which in arid regions of central and western Australia has changed to ‘meat/animal’ (McConvell 1997b). No vegetation terms have been reconstructed beyond the generic *mayi ‘vegetable food’ (Dixon 2002: 102; cf. Nash 1997: 206). The landscape feature *mungka ‘termite mound’ (O’Grady 1990b: 86) is also too ubiquitous to localise pPN. The most reliably reconstructed terms of technology are for wooden implements (cf. Evans and Jones 1997: 397–404: *kana ‘sharpened stick’, *katyi ‘digging stick’ (in some languages ‘spear’), and *kalka ‘spear’⁶⁷ (reconstructed to pPN by O’Grady 1990e: 6 and Alpher 2004: 418)). Although it is virtually impossible to find archaeological confirmation of this social organisation, it is worth commenting on the pPN system of kinship that is emerging. McConvell (2013a, 2013b, cf. McConvell and Keen 2011) reconstructs a Kariera kinship system, with in-principle marriage of first cross-cousins. Other Australian systems, including the widespread centralian “Aranda” system which involves marriage with second cousins and consistently distinguishes four grandparental terms, are interpreted as later developments.

66 McConvell (p. 136–137) assesses the significance of the findings by Joseph Birdsell of an “Aranda scarp”, a clear discontinuity in blood types and gene frequencies for tawny hair between the Western Desert people and their Arrernte neighbours to the east. 67 An apparent cognate in Victorian languages means ‘tree’, ‘wood’ or ‘bone’ (Blake and Reid 1998: 30).

Historical relations among the Australian languages 

 75

5.3 Discontinuities between related languages Discontinuities between related groups point to a physical relocation and/or cultural change of populations. The isolation of the Western Torres Strait language (Kala Lagaw Ya) and the Yolngu subgroup from the remainder of PN⁶⁸ presupposes a migration, certainly over water in the former case and probably also in the latter. In neither case is there a clear indication of the location of a prior homeland nor of their nearest linguistic relatives. Another discontinuity is Yanyuwa, on the south-western coast of the Gulf of Carpentaria, whose closest relatives in the (PN) Warluwarric subgroup are located inland and separated by Garrwa and Waanyi of the Garrwan (nPN) family. A similar discontinuity obtains with the (nPN) Mirndi family, where the western (Yirram) subgroup is separated from the more easterly Jingulu, Ngarnga and Wambaya in the Barkly Tableland by Wardaman and Mudburra. Here it is widely accepted that an earlier contiguity was probably interrupted by the northward expansion of the eastern Ngumpin language Mudburra.

5.4 Evidence of language contact from borrowed elements Contact with seasonal trepang (also called sea-slug, sea-cucumber, and bêche de mer) fishers from the Indonesian island of Sulawesi, the so-called Macassans, during the eighteenth and nineteenth centuries has left traces in loanwords in the languages of the north coast of Australia⁶⁹ — see studies by Walker and Zorc (1981), Evans (1992, 1997). The linguistic evidence of Macassan loanwords yields information concerning the geographic extent and the cultural influence of this interaction, much of which, however, is also known from historical records, archaeology, and oral traditions. The Western Torres Strait language (Kala Lagaw Ya) has long been assumed to display the effects of contact with Meriam Mir, the Papuan language of the eastern Torres Strait islands (OVV 1966: 17, 25; Black 1997). Alpher et al. (2008: 16) claim, however, that “[t]he details of WT’s shared history with Meryam remain to be studied.” A systematic study by Hunter et al. (2011) revealed that there is in fact very little similarity between the two languages, and that the little apparent similarity (especially in phonology) need not be the result of bilateral contact.

68 The Yolngu languages have always been accepted as related to the rest of Pama-Nyungan; the PN affiliation of Kala Lagaw Ya has been supported by linguistic evidence cited in Evans (2005: 254–256) and Alpher et al. (2008). 69 According to Macknight (2011: 134), “the industry in Arnhem Land began around 1780”, after some activity on the Kimberley coast a few decades earlier; the trade was closed down by Australian authorities in 1906.

76 

 Harold Koch

One area where the evidence of movement of populations is discernible primarily from the evidence of language is the riverine district of northern Western Australia. Here McConvell has found plentiful evidence from loanwords and changed grammatical and semantic structures that Eastern Ngumpin languages have spread in relatively recent times from the more inland desert areas into the riverine areas, presumably accompanied by language shift by populations that formerly spoke nPN languages (e.g. McConvell 1997a, 2002, 2009).

6 Conclusions and future prospects Black (2007c: 191) notes Dixon’s (2002: xvii) comment that a motivation for his new approach was his “lack of success in applying the established methodology of historical linguistics to the Australian linguistic situation”, but observes that “as time passes … the ‘established methodology’ has been applied to Australian languages with increasing success”, as demonstrated in Evans (2003b) and Bowern and Koch (2004). Most of the Australian languages that can be documented have by now received at least a basic description; the languages of the continent are well understood from a typological point of view; and a preliminary classification has been applied to all languages. There is by now fairly widespread agreement on the requirements for establishing genetic relations, and it now seems clear that the Australian situation is not different enough from other parts of the world to require special methods of reconstruction. A certain amount of reconstruction has been done, and much has been learned about processes of change. Many issues concerning the role of contact in language change have been canvassed, and doubtless will continue to be debated. Further detailed reconstruction needs to be done on pPN and its subgroups but especially on the nPN families. The etymological basis for all genetic constructs needs to be given, and accessible cognate databases should be made available. Computational methods can be expected to make major contributions, but painstaking etymological studies should also continue. Studies of contact-motivated change need to argue more comprehensively the pros and cons of diffusional vs. genetic explanations. Philological methods should be applied to all the old written sources so the languages known primarily through such documents can be adequately considered in comparative studies. The methods and findings of Australian historical linguistics should continue to be informed by and to inform international developments in the field. The results of linguistic reconstruction should continue to be compared with the findings of other historical disciplines such as archaeology and historical geography with regard to the prehistory of Australia.

Historical relations among the Australian languages 

 77

References Alpher, Barry 1972 On the genetic subgrouping of the languages of southwestern Cape York Peninsula, Australia. Oceania Linguistics 11: 67–87. Alpher, Barry 1976 Some linguistic innovations in Cape York and their sociocultural correlates. In: Peter Sutton (ed.), Languages of Cape York, 84–101. Canberra: Australian Institute of Aboriginal Studies. Alpher, Barry 1990 Some Proto-Pama-Nyungan paradigms: A verb in the hand is worth two in the phylum. In: Geoffrey N. O’Grady and Darrell T. Tryon (eds.), Studies in Comparative Pama-Nyungan (Pacific Linguistics C-111), 155–171. Canberra: Australian National University. Alpher, Barry 2004 Pama-Nyungan: Phonological reconstruction and status as a phylogenetic group. In: Claire Bowern and Harold Koch (eds.), Australian Languages: Classification and the Comparative Method (Current Issues in Linguistic Theory 249), 93–126, 387–574, 681–687. Amsterdam Philadelphia: John Benjamins. Alpher, Barry 2005 Review of R. M. W. Dixon (2002), Australian languages: Their nature and development. Language in Society 34: 795–805. Alpher, Barry, Nicholas Evans and Mark Harvey 2003 Proto Gunwinyguan verb suffixes. In: Nicholas Evans (ed.), The Non-Pama-Nyungan languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region (Pacific Linguistics 552), 305–352. Canberra: Australian National University. Alpher, Barry and David Nash 1999 Lexical replacement and cognate equilibrium in Australia. Australian Journal of Linguistics 19: 5–56. Alpher, Barry, Geoffrey O’Grady and Claire Bowern 2008 Western Torres Strait language classification and development. In: Claire Bowern, Bethwyn Evans and Luisa Miceli (eds.), Morphology and Language History: In Honour of Harold Koch, (Current Issues in Linguistic Theory 298), 15–30. Amsterdam/Philadelphia: John Benjamins. Anttila, Raimo 1989 Historical and Comparative Linguistics. 2nd revised edn. Amsterdam/ Philadelphia: John Benjamins. Austin, Peter 1981a Proto-Kanyara and Proto-Mantharta historical phonology. Lingua 54: 295–333. Austin, Peter 1981b Switch-reference in Australia. Language 57(2): 309–334. Austin, Peter, Robert Ellis and Luise Hercus 1976 ‘Fruit of the eyes’: Semantic diffusion in the lakes languages of South Australia. In: Papers in Australian linguistics No. 10 (Pacific Linguistics A-47), 57–77. Canberra: Australian National University. Austin, Peter, Corinne Williams and Stephen Wurm 1980 The linguistic situation in north central New South Wales. In: Bruce Rigsby and Peter Sutton (eds.), Papers in Australian Linguistics no. 13: Contributions to Australian Linguistics (Pacific Linguistics A-59), 167–180. Canberra: Australian National University. Baldi, Philip (ed.) 1990 Linguistic Change and Reconstruction Methodology. (Trends in Linguistics, Studies and Monographs 45) Berlin/New York: Mouton de Gruyter. Barrett, Bevan 2005 Historical reconstruction of the Maric languages of central Queensland. Canberra: Australian National University Master of Linguistics thesis. Bellwood, Peter 1997 Prehistoric cultural explanations for widespread language expansion In: Patrick McConvell and Nicholas Evans (eds.), Archaeology and Linguistics: Aboriginal Australia in Global Perspective, 123–134. Melbourne: Oxford University Press. Black, Paul 1980 Norman Paman historical phonology. In: Bruce Rigsby and Peter Sutton (eds.), Papers in Australian Linguistics no. 13: Contributions to Australian Linguistics (Pacific Linguistics A-59), 181–239. Canberra: Australian National University.

78 

 Harold Koch

Black, Paul 1997 Lexicostatistics and Australian languages: Problems and prospects. In: Darrell Tryon and Michael Walsh (eds.), Boundary Rider: Essays in Honour of Geoffrey O’Grady (Pacific Linguistics C-136), 51–69. Canberra: Australian National University. Black, Paul 2004 The failure of the evidence of shared innovations in Cape York Peninsula. In: Claire Bowern and Harold Koch (eds.), Australian Languages: Classification and the Comparative Method (Current Issues in Linguistic Theory 249), 241–267. Amsterdam/ Philadelphia: John Benjamins. Black, Paul 2006 Equilibrium Theory applied to Top End Australian languages. In: Keith Allan (ed.), Selected Papers from the 2005 Conference of the Australian Linguistic Society. http://www.als.asn.au. Black, Paul 2007a Comments on Jacque Guy’s ‘Lexicostatistics as chess’. Australian Journal of Linguistics 27(1): 77–79. Black, Paul 2007b Lexicostatistics with massive borrowing: The case of Jingulu and Mudburra. Australian Journal of Linguistics 27(1): 63–71. Black, Paul 2007c [Review of] Nicholas Evans (ed.), The non-Pama-Nyungan languages of northern Australia: Comparative studies of the continent’s most linguistically complex region. Aboriginal History 31: 190–193. Blake, Barry J. 1988 Redefining Pama-Nyungan: Towards the prehistory of Australian languages. In: Nicholas Evans and Steve Johnson (eds.), Aboriginal Linguistics 1, 1–90. Armidale, NSW: Department of Linguistics, University of New England. Blake, Barry J. 1990a Languages of the Queensland/Northern Territory border: Updating the classification. In: Peter Austin, R. M. W. Dixon, Tom Dutton and Isobel White (eds.), Language and History: Essays in Honour of Luise A. Hercus (Pacific Linguistics C-116), 49–66. Canberra: Australian National University. Blake, Barry J. 1990b The significance of pronouns in the history of Australian languages. In: Philip Baldi (ed.), Linguistic Change and Reconstruction Methodology, (Trends in Linguistics, Studies and Monographs 45), 435–450. Berlin: Mouton de Gruyter. Blake, Barry J., Luise Hercus, Stephen Morey and Edward Ryan 2011 The Mathi Group of Languages. (Pacific Linguistics 628) Canberra: Australian National University. Blake, Barry J. and Julie Reid 1998 Classifying Victorian languages. In: Barry J. Blake (ed.), Wathawurrung and the Colac language of Southern Victoria (Pacific Linguistics C-147), 1–58. Canberra: Australian National University. Boretzky, Norbert 1984 The Indo-Europeanist model of sound change and genetic affinity, and change in exotic languages. Diachronica 1: 1–51. Bowern, Claire 2001 Karnic classification revisited. In: Jane Simpson, David Nash, Mary Laughren, Peter Austin and Barry Alpher (eds.), Forty Years On: Ken Hale and Australian Languages (Pacific Linguistics 512), 245–261. Canberra: Australian National University. Bowern, Claire 2004a Bardi verb morphology in historical perspective. Cambridge MA: Harvard University PhD thesis. Bowern, Claire 2004b Diagnostic similarities between Nyulnyulan and neighbouring languages. In: Claire Bowern and Harold Koch (eds.), Australian Languages: Classification and the Comparative Method (Current Issues in Linguistic Theory 249), 269–290. Amsterdam/ Philadelphia: John Benjamins. Bowern, Claire 2005 [Review of] Nicholas Evans 2003, Bininj Gun-wok: A pan-dialectal grammar of Mayali, Kunwinjku and Kune and Nicholas Evans (ed.) 2003, The non-Pama-Nyungan languages of northern Australia: Comparative studies of the continent’s most linguistically complex region. Linguistic Typology 9: 341–350. Bowern, Claire 2006 Another look at Australia as a linguistic area. In: Yaron Matras, April McMahon and Nigel Vincent (eds.), Linguistic Areas: Convergence in Historical and Typological Perspective, 244–265. Basingstoke, Hampshire/New York: Palgrave Macmillan.

Historical relations among the Australian languages 

 79

Bowern, Claire 2009 Reassessing Karnic: A reply to Breen (2007). Australian Journal of Linguistics 29(3): 337–348. Bowern, Claire 2011 Loans in the basic vocabulary of Pama-Nyungan languages. Paper presented at the annual meeting of the Linguistic Society of America, Pittsburg, January 6–9 2011. Bowern, Claire 2012 The riddle of Tasmanian languages. Proceedings. Biological sciences / The Royal Society 279 (1747): 4590–4595. Bowern, Claire and Quentin Atkinson 2012 Computational phylogenetics and the internal structure of Pama-Nyungan. Language 88(4): 817–845. Bowern, Claire, Patience Epps, Russell Gray, Jane Hill, Keith Hunley, Patrick McConvell and Jason Zentz 2011 Does lateral transmission obscure inheritance in hunter-gatherer languages? PLoS ONE 6(9). e25195. doi 10.1371/journal.pone.0025195. Bowern, Claire, Bethwyn Evans and Luisa Miceli (eds.) 2008 Morphology and Language History: In Honour of Harold Koch. (Current Issues in Linguistic Theory 298) Amsterdam/Philadelphia: John Benjamins. Bowern, Claire and Harold Koch (eds.) 2004 Australian Languages: Classification and the Comparative Method. (Current Issues in Linguistic Theory 249) Amsterdam/Philadelphia: John Benjamins. Brammall, Daniel 1991 A comparative grammar of Warluwaric. Canberra: Australian National University BA Honours thesis. Breen, Gavan 1971 Aboriginal languages of Western Queensland. Linguistic Communications (Monash University) 5: 1–88. Breen, Gavan 1990 Salvage Studies of Western Queensland Aboriginal Languages. (Pacific Linguistics B-105) Canberra: Australian National University. Breen, Gavan 2003 Wanyi and Garrwa comparative data. In: Nicholas Evans (ed.), The Non-PamaNyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region (Pacific Linguistics 552), 425–462. Canberra: Australian National University. Breen, Gavan 2011 A new approach to Australian lexicostatistics. Australian Journal of Linguistics 31(2): 233–268. Breen, Gavan and Barry J. Blake 2007 The Grammar of Yalarnnga: A Language of Western Queensland. (Pacific Linguistics 584) Canberra: Australian National University. Campbell, Lyle 2003 Beyond the comparative method? In: Barry J. Blake and Kate Burridge (eds.), Historical Linguistics 2001: Selected Papers from the 15th International Conference on Historical Linguistics, Melbourne, 13–17 August 2001 (Current Issues in Linguistic Theory 237), 33–57. Amsterdam/Philadelphia: John Benjamins. Campbell, Lyle 2004 Historical Linguistics: An Introduction. 2nd edn. Edinburgh: Edinburgh University Press. Campbell, Lyle and William J. Poser 2008 Language Classification: History and Method. Cambridge: Cambridge University Press. Capell, A. 1937 The structure of Australian languages. Oceania 8: 27–61. Capell, A. 1956 A New Approach to Australian Linguistics (Handbook of Australian languages part 1). (Oceania Linguistic Monographs 1) Sydney: University of Sydney. Capell, A. 1962 Some Linguistic Types in Australia (Handbook of Australian languages Part 2). (Oceania Linguistic Monographs 7) Sydney: University of Sydney. Carew, Margaret 1993 Proto-Warluwarric Phonology. Melbourne: Melbourne University BA Honours thesis. Chadwick, Neil 1984 The Relationship of Jingulu and Jaminjungan. Canberra: AIATSIS. Unpublished manuscript.

80 

 Harold Koch

Chadwick, Neil 1997 The Barkly and Jaminjungan languages: A non-contiguous genetic grouping in North Australia. In: Darrell Tryon and Michael Walsh (eds.), Boundary Rider: Essays in Honour of Geoffrey O’Grady (Pacific Linguistics C-136), 95–106. Canberra: Australian National University. Croft, William 2000 Explaining Language Change: An Evolutionary Approach. (Longman Linguistics Library) Essex: Longman. Crowley, Terry 1976 Phonological change in New England. In R. M. W. Dixon (ed.), Grammatical Categories in Australian Languages, 19–50. Canberra: Australian Institute of Aboriginal Studies. Crowley, Terry 1999 Review of R. M. W. Dixon 1997, The rise and fall of languages. Australian Journal of Linguistics 19: 108–115. Crowley, Terry and Claire Bowern 2010 An Introduction to Historical Linguistics. 4th edn. Oxford/ New York: Oxford University Press. Crowley, Terry and R. M. W. Dixon 1981 Tasmanian. In: R. M. W. Dixon and Barry J. Blake (eds.), Handbook of Australian Languages, vol. 2, 394–421. Canberra: Australian National University Press. Curr, Edward M. 1886–1887 The Australian Race: Its Origin, Languages, Customs, Place of Landing in Australia, and the Routes by which it Spread itself over that Continent. 4 vols. Melbourne: J. Ferres, Government Printer. Dench, Alan 1997 Where do complex kin terms come from? In: Darrell Tryon and Michael Walsh (eds.), Boundary Rider: Essays in Honour of Geoffrey O’Grady (Pacific Linguistics C-136), 107– 132. Canberra: Australian National University. Dench, Alan 2001 Descent and diffusion: The complexity of the Pilbara situation. In: Alexandra Y. Aikhenvald and R. M. W. Dixon (eds.), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics, 105–133. Oxford: Oxford University Press. Dixon, R. M. W. 1970a Languages of the Cairns rain forest region. In: Stephen A. Wurm and Donald C. Laycock (eds.), Pacific Linguistic Studies in Honour of Arthur Capell (Pacific Linguistics C-13), 651–687. Canberra: Australian National University. Dixon, R. M. W. 1970b Olgolo syllable structure and what they are doing about it. Linguistic Inquiry 1: 273–276. Dixon, R. M. W. 1970c Proto-Australian laminals. Oceanic Linguistics 9(2): 79–103. Dixon, R. M. W. 1972 The Dyirbal Language of North Queensland. (Cambridge Studies in Linguistics 9) Cambridge: Cambridge University Press. Dixon, R. M. W. 1980 The Languages of Australia. Cambridge: Cambridge University Press. Dixon, R. M. W. 1990 Summary report: Linguistic reconstruction and change in the Australian language family. In: Philip Baldi (ed.), Linguistic Change and Reconstruction Methodology (Trends in Linguistics, Studies and Monographs 45), 393–401. Berlin/New York: Mouton de Gruyter. Dixon, R. M. W. 1991 Mbabaram. In R. M. W. Dixon and Barry J. Blake (eds.), Handbook of Australian Languages Volume 4: The Aboriginal Language of Melbourne and Other Grammatical Sketches, 348–402. Melbourne: Oxford University Press. Dixon, R. M. W. 1997 The Rise and Fall of Languages. Cambridge: Cambridge University Press. Dixon, R. M. W. 2001 The Australian linguistic area. In: Alexandra Y. Aikhenvald and R. M. W. Dixon (eds.), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics, 64–104. Oxford: Oxford University Press. Dixon, R. M. W. 2002 The Australian Languages: Their Nature and Development. Cambridge: Cambridge University Press. Dixon, R. M. W. 2007 Grammatical diffusion of free and bound pronouns. In: Alexandra Y. Aikhenvald and R. M. W. Dixon (eds.), Grammars in Contact: A Cross-linguistic Typology, 67–93. Oxford: Oxford University Press.

Historical relations among the Australian languages 

 81

Dobson, Annette J. and Paul Black 1979 Multidimensional scaling of some lexicostatistical data. The Mathematical Scientist 4: 55–61. Dyen, Isidore 1956 Language distribution and migration theory. Language 32: 611–626. Evans, Nicholas 1988 Arguments for Pama-Nyungan as a genetic subgroup, with particular reference to initial laminalization. In: Nicholas Evans and Steve Johnson (eds.), Aboriginal Linguistics 1, 91–110. Armidale, NSW: Department of Linguistics, University of New England. Evans, Nicholas 1992 Macassan loanwords in Top End languages. Australian Journal of Linguistics 12(1): 45–91. Evans, Nicholas 1995 A Grammar of Kayardild: With Historical-comparative Notes on Tangkic. (Mouton Grammar Library 15) Berlin/New York: Mouton de Gruyter. Evans, Nicholas 1997 Macassan loans and linguistic stratification in western Arnhem Land. In: Patrick McConvell and Nicholas Evans (eds.), Archaeology and Linguistics: Aboriginal Australia in Global Perspective, 237–260. Melbourne: Oxford University Press. Evans, Nicholas 1998 Iwaidja mutation and its origins. In: Anna Siewierska and Jae Jung Song (eds.), Case, Typology and Grammar: In Honor of Barry J. Blake (Typological Studies in Language 38), 115–149. Amsterdam/Philadelphia: John Benjamins. Evans, Nicholas 2000 Iwaidjan, a very un-Australian language family. Linguistic Typology 4: 91–142. Evans, Nicholas 2003a Introduction: Comparative Non-Pama-Nyungan and Australian historical linguistics. In: Nicholas Evans (ed.), The Non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent’s most Linguistically Complex Region (Pacific Linguistics 552), 2–25. Canberra: Australian National University. Evans, Nicholas (ed.) 2003b The Non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region. (Pacific Linguistics 552) Canberra: Australian National University. Evans, Nicholas 2005 Australian languages reconsidered: A review of Dixon (2002). Oceanic Linguistics 44: 242–286. Evans, Nicholas and Rhys Jones 1997 The cradle of the Pama-Nyungans: Archaeological and linguistic speculations. In: Patrick McConvell and Nicholas Evans (eds), Archaeology and Linguistics: Aboriginal Australia in Global Perspective, 385–417. Melbourne: Oxford University Press. Evans, Nicholas and Patrick McConvell 1998 The enigma of Pama-Nyungan expansion in Australia. In: Roger Blench and Matthew Spriggs (eds.), Archaeology and Language, Volume 2: Correlating Archaeological and Linguistic Hypotheses (One World Archaeology 29), 174–192. London: Routledge. François, Alexandre (in press) Trees, waves and linkages: Models of language diversification. In Claire Bowern and Bethwyn Evans (eds.), The Routledge Handbook of Historical Linguistics. New York: Routledge. Green, Ian 1995 The death of ‘prefixing’: Contact-induced typological change in northern Australia. Berkeley Linguistic Society 21: 414–425. Green, Ian 2003 The genetic status of Murrinh-patha. In: Nicholas Evans (ed.), The Non-PamaNyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region (Pacific Linguistics 552), 125–158. Canberra: Australian National University. Green, Ian and Rachel Nordlinger 2004 Revisiting Proto-Mirndi. In: Claire Bowern and Harold Koch (eds.), Australian Languages: Classification and the Comparative Method (Current Issues in Linguistic Theory 249), 291–311. Amsterdam / Philadelphia: John Benjamins. Green, Rebecca 2003 Proto Maningrida within Proto Arnhem: Evidence from verbal inflectional suffixes. In: Nicholas Evans (ed.), The Non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region (Pacific Linguistics 552), 369–421. Canberra: Australian National University.

82 

 Harold Koch

Greenberg, Joseph H. 1971 The Indo-Pacific hypothesis. In: Thomas A. Sebeok (ed.), Current Trends in Linguistics Volume 8: Linguistics in Oceania, 807–871. The Hague: Mouton. Greenhill, Simon J., Thomas E. Currie and Russell D. Gray 2009 Does horizontal transmission invalidate cultural phylogenies? Proceedings of the Royal Society. B, Biological sciences 276(1665): 2299–2306. Hale, Kenneth 1961 Toward a genetic classification of Australian languages. Paper presented at the 36th Annual Meeting of the Linguistic Society of America. [Unpublished ms. pMS 745, Canberra: AIATSIS]. Hale, Kenneth 1962 Linguistic evidence for routes of entry in the Aboriginal population of Australia. Paper presented at the 61st Meeting of the American Anthropological Association. [Unpublished ms. pMS 643, Canberra: AIATSIS]. Hale, Kenneth 1976 On ergative and locative suffixial alternations in Australian languages. In: R. M. W. Dixon (ed.), Grammatical Categories in Australian Languages, 414–417. Canberra: Australian Institute of Aboriginal Studies. Hale, Kenneth 1982 Review of J.G. Breen, The Mayi languages of the Queensland Gulf country. Anthropological Linguistics 24: 372–376. Hamilton, Philip 1996 Phonetic constraints and markedness in the phonotactics of Australian Aboriginal languages. Toronto: University of Toronto PhD thesis. Harvey, Mark 1997 The temporal interpretation of linguistic diversity in the Top End. In: Patrick McConvell and Nicholas Evans (eds.), Archaeology and Linguistics: Aboriginal Australia in Global Perspective, 179–185. Melbourne: Oxford University Press. Harvey, Mark 2003a An initial reconstruction of Proto Gunwinyguan phonology. In: Nicholas Evans (ed.), The Non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region (Pacific Linguistics 552), 205–268. Canberra: Australian National University. Harvey, Mark 2003b Reconstruction of pronominals among the non-Pama-Nyungan languages. In: Nicholas Evans (ed.), The Non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region (Pacific Linguistics 552), 475–513. Canberra: Australian National University. Harvey, Mark 2003c The evolution of verb systems in the Eastern Daly language family. In: Nicholas Evans (ed.), The Non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex region (Pacific Linguistics 552), 159–184. Canberra: Australian National University. Harvey, Mark 2008a Proto-Mirndi: A Discontinuous Language Family in Northern Australia. (Pacific Linguistics 593) Canberra: Australian National University. Harvey, Mark 2008b The origin of conjugation markers in Australian languages. In: Claire Bowern, Bethwyn Evans and Luisa Miceli (eds.), Morphology and Language History: In Honour of Harold Koch (Current Issues in Linguistic Theory 298), 123–137. Amsterdam/Philadelphia: John Benjamins. Harvey, Mark 2009 The genetic status of Garrwan. Australian Journal of Linguistics 29(2): 195–244. Harvey, Mark 2011 Lexical change in pre-colonial Australia. Diachronica 28(3): 345–381. Harvey, Mark 2012 Warndarrang and Marra: A diffusional or genetic relationship? Australian Journal of Linguistics 32(3): 327–360. Harvey, Mark, Ian Green and Rachel Nordlinger 2006 From prefixes to suffixes: Typological change in northern Australia. Diachronica 23: 289–311. Haspelmath, Martin 2004 How hopeless is genealogical linguistics, and how advanced is areal linguistics? Studies in Language 28(1): 209–223. Haspelmath, Martin and Uri Tadmor (eds.) 2009 Loanwords in the World’s Languages: A Comparative Handbook. Berlin: de Gruyter Mouton.

Historical relations among the Australian languages 

 83

Haviland, John B. 1974 A last look at Cook’s Guugu Yimidhirr word list. Oceania 44: 216–232. Heath, Jeffrey 1978 Linguistic Diffusion in Arnhem Land. Canberra: Australian Institute of Aboriginal Studies. Heath, Jeffery 1979 Diffusional linguistics in Australia: Problems and prospects. In Stephen A. Wurm (ed.), Australian Linguistic Studies (Pacific Linguistics C-54), 395–418. Canberra: Australian National University. Heath, Jeffrey 1981 A case of intensive lexical diffusion: Arnhem Land, Australia. Language 57: 335–367. Heath, Jeffery 1982 Review of R. M. W. Dixon 1980, The languages of Australia. Journal of Linguistics 18: 190–194. Heath, Jeffery 1987 Story of *-n-: *CV- vs. *CV-n- noun-class prefixes in Australian languages. In: Donald C. Laycock and Werner Winter (eds.), A World of Languages: Papers Presented to S. A. Wurm on his 65th birthday (Pacific Linguistics C-100), 233–243. Canberra: Australian National University. Heath, Jeffrey 1990 Verbal inflection and macro-subgroupings of Australian languages: The search for conjugation markers in non-Pama-Nyungan. In: Philip Baldi (ed.), Linguistic change and reconstuction methodology (Trends in Linguistics Studies and Monographs 45), 403–417. Berlin: Mouton de Gruyter. Hendrie, Timothy R. 1990 Initial apicals in Nuclear Pama-Nyungan. In: Geoffrey N. O’Grady and Darrell T. Tryon (eds.), Studies in Comparative Pama-Nyungan (Pacific Linguistics C-111), 15–77. Canberra: Australian National University. Hercus, Luise A. 1969 The Languages of Victoria: A Late Survey. 2 vols. Canberra: Australian Institute of Aboriginal Studies. Hercus, Luise A. 1972 The prestopped nasal and lateral consonants of Arabana-Wanganguru. Anthropological Linguistics 14: 293–305. Hercus, L uise A. 1979 In the margins of an Arabana-Wanganguru dictionary: The loss of initial consonants. In Stephen A. Wurm (ed.), Australian Linguistic Studies (Pacific Linguistics C-54), 621–651. Canberra: Australian National University. Hercus, Luise A. 1986 Victorian Languages: A Late Survey. (Pacific Linguistics B-77) Canberra: Australian National University. Hercus, Luise A. and Peter Austin 2004 The Yarli languages. In: Claire Bowern and Harold Koch (eds.), Australian Languages: Classification and the Comparative Method (Current Issues in Linguistic Theory 249), 207–222. Amsterdam / Philadelphia: John Benjamins. Hercus, Luise A. and Isobel M. White 1973 Perception of kinship structure reflected in the Adnjamathanha pronouns. In: Bernhard Schebeck, Luise A. Hercus and Isobel M. White, Papers in Australian Linguistics No. 6 (Pacific Linguistics A-36), 47–75. Canberra: Australian National University. Hiscock, Peter 2008 Archaeology of Ancient Australia. London/New York: Routledge. Hock, Hans Henrich and Brian D. Joseph 1996 Language History, Language Change, and Language Relationship: An Introduction to Historical and Comparative Linguistics. (Trends in Linguistics Studies and Monographs 93) Berlin/New York: Mouton de Gruyter. Hoenigswald, Henry M. 1990 Is the comparative method general or family-specifi c? In: Philip Baldi (ed.), Linguistic Change and Reconstruction Methodology (Trends in Linguistics Studies and Monographs 45), 377–383. Berlin: Mouton de Gruyter. Horgen, Michael 2004 The languages of the Lower-Murray. Melbourne: La Trobe University MA thesis. Hunter, Jessica, Claire Bowern and Erich Round 2011 Reappraising the effects of language contact in the Torres Strait. Journal of Language Contact 4: 106–140. Johanson, Lars and Martine Robbeets (eds.) 2012 Copies Versus Cognates in Bound Morphology. (Brill’s Studies in Language, Cognition and Culture 2) Leiden/Boston: Brill.

84 

 Harold Koch

Jones, Caroline 2008 Darkinyung Grammar and Dictionary: Revitalising a Language from Historical Sources. Nambucca Heads NSW: Muurrbay Aboriginal Language and Cultural Cooperative. Jones, Caroline and Shawn Laffan 2008 Lexical similarity and endemism in historical wordlists of Australian Aboriginal languages of the greater Sydney region. Transactions of the Philological Society 106(3): 456–486. Koch, Harold 1982 Kinship categories in Kaytej pronouns. In: Jeffrey Heath, Francesca Merlan and Alan Rumsey (eds.), Languages of Kinship in Aboriginal Australia (Oceania Linguistic Monographs 24), 64–71. Sydney: University of Sydney. Koch, Harold 1995 The creation of morphological zeroes. In: Geert Booij and Jaap van Marle (eds.), Yearbook of Morphology 1994, 31–71. Dordrecht: Kluwer Academic. Koch, Harold 2003 The case for Pama-Nyungan: Evidence from inflectional morphology. In: E. Hajičová, A. Kotěšovcová and J. Mírovský (eds.), Proceedings of XVII International Congress of Linguists, Prague, Czech Republic, July 24–29 [CD-ROM]. Prague, Czech Republic: Matfyzpress, Univerzity Karlovi. Koch, Harold 2004a A methodological history of Australian linguistic classification. In: Claire Bowern and Harold Koch (eds.), Australian Languages: Classification and the Comparative Method (Current Issues in Linguistic Theory 249), 127–150, 575–580. Amsterdam/Philadelphia: John Benjamins. Koch, Harold 2004b The Arandic subgroup of Australian languages. In: Claire Bowern and Harold Koch (eds.), Australian Languages: Classification and the Comparative Method (Current Issues in Linguistic Theory 249), 127–150, 575–580. Amsterdam / Philadelphia: John Benjamins. Koch, Harold 2009 On reconstructing pronominal proto-paradigms: Methodological considerations from the Pama-Nyungan language family of Australia. In: Bethwyn Evans (ed.), Discovering History Through Language: Papers in Honour of Malcolm Ross (Pacific Linguistics 605), 317–344. Canberra: Australian National University. Koch, Harold 2011 Aboriginal languages and social groups in the Canberra region: Interpreting the historical documentation. In: Brett Baker, Ilana Mushin, Mark Harvey and Rod Gardner (eds.), Indigenous Language and Social Identity: Papers in Honour of Michael Walsh (Pacific Linguistics 626), 123–144. Canberra: Australian National University. Koch, Harold 2013 The etymology of a paradigm: The Pama-Nyungan 3SgFem reconsidered. In: Robert Mailhammer (ed.), Lexical and Structural Etymology: Beyond Word Histories (Studies in Language Change 11), 53–82. Berlin: de Gruyter Mouton. Koch, Harold 2014 The reconstruction of inflectional classes in morphology: History, method and Pama-Nyungan (Australian) verbs. In: Rob Pensalfini, Myfany Turpin and Diana Guilleman (eds.), Language Description Informed by Theory, 153–189. Amsterdam/ Philadelphia: John Benjamins. Koch, Harold in press Loanwords between the Arandic languages and their western neighbours: principles of identification and phonological adaptation. In: Lauren Gawne and Jill Vaughan (eds.), The 43rd Annual Australian Linguistic Society Conference Proceedings – 2013. Kroeber, A. L. 1923 Relationship of the Australian languages. Journal of the Royal Society of New South Wales 57: 101–117. Kuteva, Tania 1999 Language and societies: The punctuated equilibrium model of language development. Language and Communication 19: 213–228. Laffan, Kate 2003 Reconstruction of the Wakka-Kabi languages of south-eastern Queensland. Canberra: Australian National University BA Honours thesis. Lewis, M. Paul (ed.) 2009 Ethnologue: Languages of the World. 16th edn. Dallas, Tex.: SIL International. http://www.ethnologue.com/ Laycock, Don 1982 Review of R. M. W. Dixon 1980, The languages of Australia. Language 58(3): 701–704.

Historical relations among the Australian languages 

 85

Mailhammer, Robert 2013 Towards a framework of contact etymology. In: Robert Mailhammer (ed.), Lexical and Structural Etymology: Beyond Word Histories (Studies in Language Change 11), 9–31. Berlin: de Gruyter Mouton. Macknight, Campbell 2011 The view from Marege’: Australian knowledge of Makassar and the impact of the trepang industry across two centuries. Aboriginal History 35: 121–143. Mathew, John 1899 Eaglehawk and Crow: A Study of the Australian Aborigines Including an Inquiry into their Origin and a Survey of Australian Languages. London: David Nutt. McConvell, Patrick 1985 The origin of subsections in northern Australia. Oceania 56: 1–33. McConvell, Patrick 1996 Backtracking to Babel: The chronology of Pama-Nyungan expansion in Australia. Archaeology in Oceania 31: 125–144. McConvell, Patrick 1997a Long-lost relations: Pama-Nyungan and Northern kinship. In: Patrick McConvell and Nicholas Evans (eds.), Archaeology and Linguistics: Aboriginal Australia in Global Perspective, 207–235. Melbourne: Oxford University Press. McConvell, Patrick 1997b Semantic shifts between fish and meat and the prehistory of PamaNyungan. In: Darrell Tryon and Michael Walsh (eds.), Boundary Rider: Essays in Honour of Geoffrey O’Grady (Pacific Linguistics C-136), 303–325. Canberra: Australian National University. McConvell, Patrick 2001 Language shift and language spread among hunter-gatherers. In: Catherine Panter-Brick, Robert H. Layton and Peter Rowley-Conwy (eds.), Hunter-gatherers: An Interdisciplinary Perspective, 143–169. Cambridge: Cambridge University Press. McConvell, Patrick 2002 Linguistic stratigraphy and native title: The case of ethnonyms. In: John Henderson and David Nash (eds.), Linguistics and Native Title (Native Title Research Series), 261–292. Canberra: Aboriginal Studies Press. McConvell, Patrick 2008 Grand-daddy morphs: The importance of suffixes in reconstructing PamaNyungan kinship. In: Claire Bowern, Bethwyn Evans and Luisa Miceli (eds.), Morphology and Language History: In Honour of Harold Koch (Current Issues in Linguistic Theory 298), 313–327. Amsterdam/Philadelphia: John Benjamins. McConvell, Patrick 2009 Loanwords in Gurindji, a Pama-Nyungan language of Australia. In: Martin Haspelmath and Uri Tadmor (eds.), Loanwords in the World’s Languages: A Comparative Handbook, 790–822. Berlin: Mouton de Gruyter. McConvell, Patrick 2010 Contact and Indigenous languages in Australia. In: Raymond Hickey (ed.), The Handbook of Language Contact, 770–794. Oxford: Blackwell. McConvell, Patrick 2013a Granny got cross: Semantic change of kami ‘mother’s mother’ to ‘father’s mother’ in Pama-Nyungan. In: Robert Mailhammer (ed.), Lexical and Structural Etymology: Beyond Word Histories (Studies in Language Change 11), 147–184. Berlin: de Gruyter Mouton. McConvell, Patrick 2013b Proto-Pama-Nyungan kinship and the AUSTKIN project: Reconstructing terms for proto-Mother’s Father and their transformations. In: Patrick McConvell, Ian Keen and Rachel Hendery (eds.), Kinship Systems: Change and Reconstruction, 190–214. Salt Lake City: University of Utah Press. McConvell, Patrick and Nicholas Evans (eds.) 1997a Archaeology and Linguistics: Aboriginal Australia in Global Perspective. Melbourne: Oxford University Press. McConvell, Patrick and Nicholas Evans 1997b Clues to Australia’s human past: Pulling together the strands. In: Patrick McConvell and Nicholas Evans (eds.), Archaeology and Linguistics: Aboriginal Australia in Global Perspective, 1–16. Melbourne: Oxford University Press. McConvell, Patrick and Mary Laughren 2004 The Ngumpin-Yapa subgroup. In: Claire Bowern and Harold Koch (eds.), Australian Languages: Classification and the Comparative Method (Current Issues in Linguistic Theory 249), 151–177. Amsterdam/Philadelphia: John Benjamins.

86 

 Harold Koch

McConvell, Patrick and Ian Keen 2011 The transition from Kariera to an asymmetrical system: Cape York Peninsula to North-East Arnhem Land. In: Doug Jones and Bojka Milic (eds.), Kinship, Language, and Prehistory: Per Hage and the Renaissance in Kinship Studies, 99–132. Salt Lake City: The University of Utah Press. McGregor, William B. 2008 The origin of noun classes in Worrorran languages. In: Claire Bowern, Bethwyn Evans and Luisa Miceli (eds.), Morphology and Language History: In Honour of Harold Koch (Current Issues in Linguistic Theory 298), 185–200. Amsterdam/Philadelphia: John Benjamins. McGregor, William B. and Alan Rumsey 2009 Worrorran Revisited: The Case for Genetic Relations among Languages of the Northern Kimberley Region of Western Australia. (Studies in Language Change/Pacific Linguistics 600) Canberra: Australian National University. McMahon, April and Robert McMahon 2006 Why linguists don’t do dates: Evidence from IndoEuropean and Australian languages. In: Peter Forster and Colin Renfrew (eds.), Phylogenetic Methods and the Prehistory of Languages, 153–160. Cambridge, UK: McDonald Institute for Archaeological Research. Meillet, Antoine 1925 La méthode comparative en linguistique historique (Instituttet for sammenlignende kulturforskning, Oslo. Ser. A: Forelesninger, 2). Paris: H. Champion. Meillet, Antoine 1967 The Comparative Method in Historical Linguistics. Trans. Gordon B. Ford, Jr. Paris: Champion. Merlan, Francesca 2003 The genetic position of Mangarrayi: Evidence from nominal prefixation. In: Nicholas Evans (ed.), The Non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region (Pacific Linguistics 552), 353– 367. Canberra: Australian National University. Miceli, Luisa 2004 Pama-Nyungan as a genetic entity. In: Claire Bowern and Harold Koch (eds.), Australian Languages: Classification and the Comparative Method (Current Issues in Linguistic Theory 249), 61–68. Amsterdam/Philadelphia: John Benjamins. Miceli, Luisa 2008 Morphological reconstruction and Australian languages. In: Claire Bowern, Bethwyn Evans and Luisa Miceli (eds), Morphology and Language History: In Honour of Harold Koch (Current Issues in Linguistic Theory 298), 211–219. Amsterdam/Philadelphia: John Benjamins. Miceli, Luisa (in press) Pama-Nyungan. In Claire Bowern and Bethwyn Evans (eds), The Routledge Handbook of Historical Linguistics. New York: Routledge. Milroy, James and Leslie Milroy 1985 Linguistic change, social network and speaker innovation. Journal of Linguistics 21: 339–384. Mushin, Ilana and Jane Simpson 2008 Free to bound to free?: Interactions between pragmatics and syntax in the development of Australian pronominal systems. Language 84(3): 566–596. Nash, David 1997 Comparative flora terminology of the central Northern Territory. In: Patrick McConvell and Nicholas Evans (eds), Archaeology and Linguistics: Aboriginal Australia in Global Perspective, 187–206. Melbourne: Oxford University Press. Nash, David 2002 Historical linguistic geography of south-east Western Australia. In: John Henderson and David Nash (eds.), Language in Native Title (Native Title Research Series), 205–230. Canberra: Aboriginal Studies Press. Nash, David 2008 Warlpiri verb roots in comparative perspective. In: Claire Bowern, Bethwyn Evans and Luisa Miceli (eds.), Morphology and Language History: In Honour of Harold Koch (Current Issues in Linguistic Theory 298), 221–234. Amsterdam/Philadelphia: John Benjamins. Nash, David and Jane Simpson 1981 No-name in central Australia. In: Carrie S. Masek, Roberta A. Hendrick and Mary Frances Miller (eds.), Papers from the Parassession on Language and Behavior, Chicago Linguistic Society, 165–177. Chicago: Chicago Linguistic Society.

Historical relations among the Australian languages 

 87

Nichols, Johanna 1992 Linguistic Diversity in Space and Time. Chicago/London: University of Chicago Press. Nichols, Johanna 1996 The comparative method as heuristic. In: Mark Durie and Malcolm Ross (eds.), The Comparative Method Reviewed: Regularity and Irregularity in Language Change, 39–71. New York: Oxford University Press. Nichols, Johanna 1997 Sprung from two common sources: Sahul as a linguistic area. In: Patrick McConvell and Nicholas Evans (eds.), Archaeology and Linguistics: Aboriginal Australia in Global Perspective, 135–168. Melbourne: Oxford University Press. O’Grady, Geoffrey N. 1960 More on lexicostatistics: Comments. Current Anthropology 1: 338–339. O’Grady, Geoffrey N. 1966 Proto-Ngayarda phonology. Oceanic Linguistics 5: 71–130. O’Grady, Geoffrey N. 1979 Preliminaries to a Proto Nuclear Pama-Nyungan stem list. In: Stephen A. Wurm (ed.), Australian Linguistic Studies (Pacific Linguistics C-54), 107–139. Canberra: Australian National University. O’Grady, Geoffrey N. 1981 Review of R. M. W. Dixon 1980, The languages of Australia. AUMLA 56: 273–276. O’Grady, Geoffrey N. 1990a Introduction. In: Geoffrey N. O’Grady and Darrell T. Tryon (eds.), Studies in Comparative Pama-Nyungan (Pacific Linguistics C-111), xiii–xxii. Canberra: Australian National University. O’Grady, Geoffrey N. 1990b Pama-Nyungan *m-, *j- and *k-. In Geoffrey N. O’Grady and Darrell T. Tryon (eds.), Studies in Comparative Pama-Nyungan (Pacific Linguistics C-111), 79–103. Canberra: Australian National University. O’Grady, Geoffrey N. 1990c Pama-Nyungan: the tip of the lexical iceberg. In: Geoffrey N. O’Grady and Darrell T. Tryon (eds.), Studies in Comparative Pama-Nyungan (Pacific Linguistics C-111), 209–259. Canberra: Australian National University. O’Grady, Geoff N. 1990d Prenasalisation in Pama-Nyungan. In: Philip Baldi (ed.), Linguistic Change and Reconstruction Methodology (Trends in Linguistics Studies and Monographs 45), 451–476. Berlin: Mouton de Gruyter. O’Grady, Geoffrey N. 1990e Wadjuk and Umpila: A long-shot approach to Pama-Nyungan. In: Geoffrey N. O’Grady and Darrell T. Tryon (eds.), Studies in Comparative Pama-Nyungan (Pacific Linguistics C-111), 1–10. Canberra: Australian National University. O’Grady, Geoff and Susan Fitzgerald 1997 Cognate search in the Pama-Nyungan language family. In: Patrick McConvell and Nicholas Evans (eds.), Archaeology and Linguistics: Aboriginal Australia in Global Perspective, 341–355. Melbourne: Oxford University Press. O’Grady, Geoff and Kenneth Hale 2004 The coherence and distinctiveness of the Pama-Nyungan language family within the Australian linguistic phylum. In: Claire Bowern and Harold Koch (eds.), Australian Languages: Classification and the Comparative Method (Current Issues in Linguistic Theory 249), 69–92. Amsterdam / Philadelphia: John Benjamins. O’Grady, Geoffrey N. and Terry Klokeid 1969 Australian linguistic classification: A plea for coordination of effort. Oceania 39: 298–311. O’Grady, Geoffrey N. and Darrell T. Tryon (eds.) 1990b Studies in Comparative Pama-Nyungan. (Pacific Linguistics C-111) Canberra: Australian National University. O’Grady, Geoffrey N., C. F. Voegelin and F. M. Voegelin 1966 Languages of the world: Indo-Pacific fascicle 6. Anthropological Linguistics 8(2): 1–199. O’Grady, Geoffrey N., Stephen A. Wurm and Kenneth L. Hale 1966 Aboriginal languages of Australia (a preliminary classification) [map]. Victoria, B.C.: Department of Linguistics, University of Victoria. Pawley, Andrew 2009 Greenberg’s Indo-Pacific hypothesis: An assessment. In: Bethwyn Evans (ed.), Discovering History through Language: Papers in Honour of Malcolm Ross (Pacific Linguistics 605), 153–180. Canberra: Australian National University.

88 

 Harold Koch

Pensalfini, Rob 2001 On the typological and genetic affiliation of Jingulu. In: Jane Simpson, David Nash, Mary Laughren, Peter Austin and Barry Alpher (eds), Forty Years On: Ken Hale and Australian Languages (Pacific Linguistics 512), 385–399. Canberra: Australian National University. Plomley, N. J. B. 1976 A Word-list of the Tasmanian Aboriginal Languages. Launceston: The author in association with the Government of Tasmania. Reesink, Ger, Ruth Singer and Michael Dunn 2009 Explaining the linguistic diversity of Sahul using population models. PLoS Biol 7(11): e1000241. doi:10.1371/ journal.pbio.100024. Round, Erich R. 2009 Kayardild morphology, phonology and morphosyntax. New Haven CT: Yale University PhD thesis. Round, Erich R. 2010 Syntactic reconstruction by phonology: Edge aligned reconstruction and its application to Tangkic truncation. In: Rachel Hendery and Jennifer Hendriks (eds.), Grammatical Change: Theory and Description (Studies in Language Change 6 / Pacific Linguistics 609), 65–81. Canberra: Australian National University. Sands, Kristina 1995 Nominal classification in Australia. Anthropological linguistics 37: 247–346. Sands, Kristina 1996 The Ergative in Proto-Australian. (Studies in Australian Languages 1) München: Lincom Europa. Schmidt, Wilhelm 1919a Die Gliederung der australischen Sprachen. Vienna: MecharistenBuchdrückerei. Schmidt, Wilhelm 1919b Die Personalpronomina in den australischen Sprachen. (Akademie der Wissenschaften in Wien, Philosophische-historische Klasse, Denkschriften 64. Band, 1. Abhandlung) Vienna: Alfred Hölder. Simpson, Jane 1988 Case and complementiser suffixes in Warlpiri. In: Peter Austin (ed.), Complex Sentence Constructions in Australian Languages (Typological Studies in Language 15), 205–218. Amsterdam/Philadelphia: John Benjamins. Simpson, Jane, David Nash, Mary Laughren, Peter Austin and Barry Alpher (eds.) 2001 Forty Years On: Ken Hale and Australian Languages. (Pacific Linguistics 512) Canberra: Australian National University. Smith, Mike 2005 Desert archaeology, linguistic stratigraphy, and the spread of the Western Desert language. In: Peter Veth, Mike Smith and Peter Hiscock (eds.), Desert Peoples: Archaeological Perspectives, 222–242. Oxford: Blackwell. Sutton, Peter (ed.) 1976 Languages of Cape York. Canberra: Australian Institute of Aboriginal Studies. Sutton, Peter 1990 The pulsating heart: Large-scale cultural and demographic processes in Aboriginal Australia. In: Betty Meehan and Neville White (eds.), Hunter-Gatherer Demography: Past and Present, 71–80. Sydney: University of Sydney. Sutton, Peter and Harold Koch 2008 Australian languages: A singular vision [Review article on R. M. W. Dixon 2002, Australian languages: Their nature and development]. Journal of Linguistics 44: 471–504. Tadmor, Uri 2009 Loanwords in the world’s languages: Findings and results. In: Martin Haspelmath and Uri Tadmor (eds.), Loanwords in the World’s Languages: A Comparative Handbook, 55–75. Berlin: Mouton de Gruyter. Taplin, George 1879 The Folklore, Manner, Customs, and Languages of the South Australian Aborigines. Adelaide: Government Printer. Thomason, Sarah 2010 Contact explanations in linguistics. In: Raymond Hickey (ed.), The Handbook of Language Contact, 31–46. Oxford: Wiley-Blackwell. Toulmin, Matthew 2009 From Linguistic to Sociolinguistic Reconstruction: The Kamta Historical Subgroup of Indo-Aryan. (Studies in Language Change/Pacific Linguistics 604) Canberra: Australian National University.

Historical relations among the Australian languages 

 89

Trask, R. L. 2000 The Dictionary of Historical and Comparative Linguistics. Edinburgh: Edinburgh University Press. Tryon, Darrell and Michael Walsh (eds.) 1997 Boundary Rider: Essays in Honour of Geoffrey O’Grady. (Pacific Linguistics C-136) Canberra: Australian National University. van Egmond, Marie-Elaine 2012 Enindhilyakwa phonology, morphosyntax and genetic position. Sydney: University of Sydney PhD thesis. Veth, Peter 2000 Origins of the Western Desert language: Convergence in linguistic and archaeological space and time models. Archaeology in Oceania 35(1): 11–19. Voegelin, C. F., F. M. Voegelin, Stephen Wurm, Geoffrey O’Grady and Tokuichiro Matsuda 1963 Obtaining an index of phonological differentiation from the construction of non-existent minimax systems. International Journal of American Linguistics 29(1): 4–29. Walker, Alan and R. David Zorc 1981 Austronesian loanwords in Yolngu-Matha of northeast Arnhem Land. Aboriginal History 5: 109–134. Walsh, Michael J. and Stephen A. Wurm 1981 Maps of Australia and Tasmania. In: Stephen A. Wurm and Shiro Hattori (eds.), Language Atlas of the Pacific Area. Part 1: New Guinea Area, Oceania, Australia, Maps 20–23. Canberra: Australian Academy of the Humanities. Waters, Bruce E. 1989 Djinang and Djinpa: A Grammatical and Historical Perspective. (Pacific Linguistics C-114) Canberra: Australian National University. Watkins, Calvert 2001 An Indo-European linguistic area and its characteristics: Ancient Anatolia: Areal diffusion as a challenge to the comparative method? In: Alexandra Y. Aikhenvald and R. M. W. Dixon (eds.), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics, 44–63. Oxford: Oxford University Press. Weber, Natalie 2009 Comparative reconstruction of Marrngu. Houston TX: Rice University BA thesis. Wurm, Stephen A. 1971 Classifications of Australian languages, including Tasmanian. In: Thomas A. Sebeok (ed.), Current Trends in Linguistics Volume 8: Linguistics in Oceania, 721–778. The Hague: Mouton Wurm, Stephen A. 1972 Languages of Australia and Tasmania. The Hague: Mouton.

Janet Fletcher and Andrew Butcher

3 Sound patterns of Australian Languages 1 Introduction At the beginning of his seminal review of the phonological characteristics of Australian languages, Evans (1995a: 723) refers to the “classical period” in the description of Australian language sound systems from the late sixties to the early eighties (e.g. Capell 1967, Dixon 1980, Blake 1981, Yallop 1982).¹ A definitive summary of Australian sound systems is also included in Dixon (2002: Ch 12). From this body of work has emerged a general picture of some of the hallmark traits of Australian language sound patterns. Some of the key features of Australian languages are that they have complex consonant inventories with multiple place of articulation contrasts in the stop and sonorant series, a notable lack of contrastive fricatives or a stop voicing contrast, coupled with small vowel inventories (e.g. see also Capell 1967; Yallop 1982; Blake 1981). In the last twenty years, research largely undertaken within the segmental descriptive tradition, has been augmented with a growing body of experimental phonetic analysis that has looked at a range of phonological and phonetic phenomena ranging from the acoustic features of vowel inventories to intonation. This chapter aims to introduce the reader to some of the salient phonetic and phonological features of Australian languages, beginning with an overview of segmental properties, and finishing with a survey of connected speech processes, stress and prosodic prominence, and intonational phenomena. While most phonetic and phonological interest has focused on the characteristics of the consonant inventories and metrical stress patterns, a range of newer analyses have been conducted on hitherto neglected features of Australian languages, namely the vowel inventories, consonant coarticulation, and aspects of post-lexical phonetics and phonology, including intonation, rhythm and other temporal patterns such as pausing and tempo variation. The results of these studies are summarised in the sections that follow.

2 Vowels Australian languages fall into the small percentage of the world’s languages (generally thought to be around 5%) that have small vowel inventories consisting of three to five contrastive vowels (e.g. Crothers 1978; Schwartz et al. 1997; Becker-Kristal 2010;

1 The authors would like to thank John Ingram, Brett Baker, and the editors for helpful comments on an earlier version of this manuscript. This chapter is dedicated to the memory of Colin Yallop.

92 

 Janet Fletcher and Andrew Butcher

Maddieson 2011). Earlier surveys of Australian languages show that around 54% have “symmetrical” triangular vowel spaces that are typically transcribed as /i/, /a/ and /u/, with an additional length contrast in many cases (e.g. Busby 1980). Table 1 presents a summary of vowel systems for some of the Australian languages included in UPSID (UCLA Phonetic Segment Inventory Database — Maddieson 2004) and in Busby’s survey. The table also shows alternative phonemic representations for a range of languages. Around 9% of the 134 languages surveyed by Busby (generally those spoken in the north of Australia, including the Kimberleys and Arnhem Land), have symmetrical inventories consisting of five vowels /i e a o u/. A further 12% of Australian languages have four or six vowels. In languages such as Dalabon, the sixth vowel has been variously described as close central /ɨ/ (e.g. Alpher 1982; Fletcher and Evans 2002) or /ə/ (Capell 1967). Most languages with more than three contrastive vowel qualities are non-Pama-Nyungan languages (after Dixon 1980). Busby (1980) also describes some of the vowel systems of Australian languages as either semisymmetrical or “ungrouped”, indicating an atypical vowel inventory. In the case of semi-symmetrical four-vowel inventories, the additional vowel is usually transcribed as /o/ or /oː/. The clearest example of an “ungrouped” system from Busby’s survey is Kaytetye, which has been analysed as having only two contrastive central vowels /ə, a/. Breen (2001) makes the important point that for many of the Arandic languages (including Kaytetye), the vowels /i/, /u/ can be better analysed as a phonetic consequence of palatal or labio-velar allophony. Indeed, he further suggests that the term “triangular” is inappropriate and the vowel systems can be better described as vertical, with two major phoneme categories /ə ɐ/. When one compares the vowel inventories of this group of Australian languages to other non-Australian languages in classic phoneme databases like UPSID, the basic 3-vowel inventories appear similar to those of other small-vowel languages from around the world. However, there are competing analyses for many of the languages listed in Table 1. For example, the most recent analysis of Central Arrernte posits the contrastive vowels as /i ə ɐ/ (Tabain and Breen 2011 and references therein). Earlier analyses of other Arrernte varieties have suggested this can be further reduced to two contrastive vowels /ə a/ (e.g. Dixon 1980), although this analysis has now been abandoned (see Breen and Pensalfini 1999: 23), but is still maintained for Western Anmatyerre, for example (Breen 2001). The absence of a contrastive back rounded vowel (or close front unrounded vowel, depending on the analysis) is an unusual typological feature of this language. However, Henderson (1998) suggests that /u/ is a fourth possible phoneme in the language of younger speakers, and this vowel is included in the vowel inventory that Breen and Dobson (2005) present for Central Arrernte. According to Busby (1980) almost 55% of languages included in his survey also have a phonemic length contrast although, once again, there are competing analyses for certain languages (e.g. Iwaidja, Ngalakgan). Long vowels tend to be less frequent for the most part, with relatively low text frequency compared to plain vowels. Some

Sound patterns of Australian Languages 

 93

Table 1: Various representations of vowel inventories of Australian languages in UPSID (U), supplemented with some of the languages included in Busby’s (1980) survey (B). The languages with an * have also been analysed as 2-vowel languages. Alternative transcription conventions in parentheses are also shown for some of the languages. Vowels iau ɪɐʊ ɪ ə ɐ (ʊ) / i a u i a u i: a: u:

i a u a: o: ɪeɐʊ ieau iɛaʊ ɪaɔʊ i a ɔ u i: a: u: iɛaɔu ɪɛɐɔu ieaou ieaoʊ i e a o u i: e: a: o: u: i e a u i: e: a: u: i ɛ a̟ ɔ u iɛɐɔʊ ieaouə ieaouɨ ieauə

Languages Diyari, Dyirbal, Kuku Yalanji, Iwaidja, Kaytetye*, Nyulnyul, Nyangumarta, Wambaya, Western Desert, Yankunytjatjara, Yanyuwa, Yidiny Garrwa, Kalkatungu Arrernte* Alyawarr, Gupapuyngu, Guugu Yimidhirr, Jingulu, Kayardild, Nhanda, Ngiyampaa, Nunggubuyu (Wubuy), Pitjantjatjara, Walmajarri, Warlpiri, Warrwa, Warumungu, Yindjibarndi, Yolngu, Gooniyandi Alawa (B:ieau) Bundjalung, Jaminjung, Marrithiyel (B:ieauə) Murrinh-Patha (ieau) Tiwi (U:iɐɔu) Bardi Burarra, Mawng (B:ieaou), Wagiman (ɪɛɐɔʊ) Bininj Gun-wok (ieaou) Ngalakgan Wardaman (B) Kuuk Thaayorre, Wik Mungkan (U:iɛɐɔʊ) Lardil Warray Ungarinyin (B:ieaoʊ) Kala Lagaw Ya, Rembarrnga, Yir Yoront Dalabon PunguPungu (B)

Australian languages also have non-contrastive phonetic diphthongs. For example, in the Kunwinjku variety of Bininj Gun-wok, there are eight phonetic diphthongs [iw ew aw ow ej aj oj uj], which Evans (2003) analyses as phonological sequences of two separate vowel segments. Most phonological descriptions of Australian languages conclude there is wide variation in the actual quality of vowels transcribed as the three classic point vowels /i a u/. As a consequence there is also a great deal of variation with regard to the choice of symbols that are used to represent the vowels. This is clear from Table 1 and it is typical of most surveys of phoneme-based segmental inventories like UPSID. This point is also made in other surveys of vowel systems across the world’s languages (e.g. Becker-Kristal 2010). The quality of vowels transcribed as close /i/ and /u/ in some three-vowel Australian languages like Warlpiri, tends to be somewhere between close and close-mid, rather than close, and the open vowel is generally best described as open central /ɐ/ (e.g. Butcher 1994, 2006a). In fact, the kinds of variation in open

94 

 Janet Fletcher and Andrew Butcher

vowel representation that Becker-Kristal reports for small vowel languages is also reflected in certain descriptions of Australian languages, with the typical open vowel being described as open front (i.e. closer to true Cardinal 4 for some of the Arandic languages) or open back [ɑ] in others (e.g. Ngan’gityemerri, Reid 2011; Bardi, Bowern, McDonough and Kelliher 2012). The less-close symbols /ɪ ʊ/ are frequently used to represent the close vowels in many languages. All analysts of Australian languages also agree there is an enormous degree of allophonic variation within the major vowel categories, which is a typical feature of most small-vowel inventory languages. This is largely due to the strong coarticulatory influences of surrounding consonants (e.g. Dixon 2002; Breen 2001). However, a notably atypical pattern is the absence of nasalised vowels in the vicinity of nasals, which is a very unusual feature of Australian languages (Butcher 1999). Newer acoustic phonetic techniques have been applied effectively to address a number of features of Australian vowel systems and have helped refine the characteristic features of typical vowel spaces in these languages (e.g. Butcher 1994; Fletcher and Butcher 2002, 2003; Fletcher et al. 2007; Tabain and Breen 2011). The major findings of these studies are summarised in the following sections.

2.1 Acoustic properties of vowels in Australian languages For the most part, acoustic phonetic analyses of vowels in Australian languages show that they adhere to principles of inventory organisation that are predicted by implementations of Dispersion Theory (e.g. Lindblom 1986; Becker-Kristal 2010), where maintenance of perceptual contrast is the key organisational principle. Classic dispersion theory predicts that vowel contrasts are essentially systemic and relational, not absolute or local. Articulatory economy counterbalances the perceptual demands for a contrast, which is usually cited as one important reason why three vowel systems tend to consist of the point vowels /i/, /a/, /u/, and five vowel systems consist of /i/, /e/, /a/, /o/ and /u/. However, in a preliminary acoustic phonetic study of a typical 3-vowel language (Warlpiri) and 5-vowel language (Burarra), Butcher (1994) suggests that the acoustic vowel spaces of Australian languages tend to be “compact” and show minimal or sufficient dispersion, with less peripheral vowels compared to languages with large vowel inventories like English or Swedish, for example. This is shown clearly in Figures 1–3, which show vowel formant plots of approximately 800 to 1200 vowels from the three-vowel languages Warlpiri (after Butcher and Harrington 2003), Iwaidja and Kayardild, and a typical five-vowel northern Australian language Bininj Gun-wok. While the mean F1 and F2 values (first and second formants, respectively) shown as the centroid value in Hz in each ellipse illustrate the symmetrical nature of the three and five vowel spaces, the vowel spaces of Warlpiri, Iwaidja (male speaker) and Kayardild can reliably be described as compressed and compact compared to those of relatively large vowel inventory languages like English or German.

Sound patterns of Australian Languages 

 95

The five-vowel language shown here has a more peripheral vowel space than the three-vowel languages (with the exception of the Iwaidja female speaker). The mean F1 centroid value for /i/ in Bininj Gun-wok is 339 Hz and 369 Hz for male and female speakers respectively, whereas Butcher (2006a) has noted much higher F1 values that approach 450–500 Hz in Warlpiri, suggesting a more [e̝]-like vowel. These differences are evident in Figure 1 and Figure 2, even taking into account speaker-sex differences (where F1 is generally higher in females compared to males). The two speakers of Kayardild also have somewhat lowered close vowels, with short /i/ having a mean F1 value between 464–488 Hz, although the long vowel /iː/ has F1 values approaching those noted for General American English, particularly for the male speaker (e.g. Hillenbrand et al. 1995), suggesting a more peripheral vowel. In the F2 dimension, some interesting differences emerge, with speakers from the fivevowel language Bininj Gun-wok having high F2 centroid values across the board (e.g. 2151 Hz for males, 2367 Hz for females) compared to Iwaidja (male speaker) and the short vowels of Kayardild, according well with predictions of Dispersion Theory. In particular, the F2 centroid value of /i/ for Iwaidja (male speaker) at 1843 Hz is somewhat low compared to the averaged F2 frequencies for close vowels produced by men in corpus studies of American English, which range from 2322 Hz for /i/ to 2089 Hz for /e/ (e.g. Hillenbrand et al. 1995). The short back vowel /u/ in Kayardild has F1 centroid values from 470–490 Hz and F2 values within the 1042–1148 Hz range indicative

Figure 1: Warlpiri vowels: F1-F2 vowel target ellipses (Hz) for three female speakers of Warlpiri for the three vowels /ɪ ɐ ʊ / occurring in initial, medial and final word position.

96 

 Janet Fletcher and Andrew Butcher

Figure 2: Iwaidja and Kayardild vowels plotted in the F1-F2 dimension (Hz) for two speakers (female and male) of each language. Ellipses include 95% of the data points in the Iwaidja plots. The centroid frequencies of Kayardild long vowels are also shown.

of a less close [ʊ]. The long vowels in Kayardild are also more peripheral, according well with observations from other languages that have vowel quantity contrasts (e.g. House 1961; Disner 1983; Tsukada 2009). In the 5-vowel language, Bininj Gun-wok, centroid values of F1 and F2 also suggest a less close vowel [ʊ]. Likewise the central open position of the /a/ for the four languages shown here is closer to IPA [ɐ], supporting the use of this symbol for open vowels in many of the inventories shown in Table 1. Butcher (2006a) (after Dixon 1980) suggests that the lack of lip rounding in “back” vowels is also possibly why F2 extends beyond 1000Hz in many /u/ or /ʊ/ vowels. The most striking feature of the vowel formant plots in Figures  1 and 2 is the high degree of overlap between vowel ellipses, particularly in word initial and me-

Sound patterns of Australian Languages 

 97

Figure 3: Bininj Gun-wok vowels: F1-F2 vowel target ellipses (Hz) for 2 (female) speakers of Bininj Gun-wok for vowels in accentually prominent syllables and unaccented medial syllables. Ellipses include 95% of the data points.

dial positions in Warlpiri, in both accented and unaccented contexts in Bininj Gunwok, and in Kayardild and Iwaidja (male speaker). This degree of vowel variation has also been observed acoustically in the central vowels of Arrernte (Tabain and Breen 2011), Burarra (Butcher 1994; Graetzer 2012) and Gupapuyngu (Graetzer et al. 2012). Interestingly, in studies where factors like position in word or presence or absence of accentual prominence have been taken into account (and as is clear in the plots shown here for Bininj Gun-wok), there have been equivocal results regarding their influence on vowel formant frequencies (e.g. Fletcher and Evans 2002; Fletcher and Butcher 2002, 2003; Fletcher et al. 2007; Tabain and Breen 2011; Graetzer et al. 2012; Tabain and Fletcher 2012; Tabain, Fletcher, and Butcher 2014). It is well known that in languages like English and German, accentual prominence can correlate with sharp-

98 

 Janet Fletcher and Andrew Butcher

er, more peripheral vowel quality and an overall expansion of the vowel space in accented syllables (Harrington et al. 2000; Mooshammer and Geng 2008). In languages like Arrernte, Pitjantjatjara, Warlpiri, Dalabon and Bininj Gun-wok, only /ɐ/ tends to become more open and longer in prosodically prominent contexts, and this is not always consistent across speakers. Other phonetic correlates of metrical stress and prosodic prominence are further explored in § 5.1. It is also of note that the degree of overlap is minimised in final position in Warlpiri (Figure 1). One reason for this is the extra vowel lengthening that is often observed in word-final position, which in turn can result in more extreme F1/F2 values in Warlpiri and other Australian languages (e.g. Pentland 2004; Fletcher and Evans 2002). In a related vein, less overlap in final position may also be due to the lack of a final consonant, so there is less coarticulation than in non-final contexts (Ingram, personal communication). The critical factor that governs vowel allophony is contextual influence of place of articulation of neighbouring consonants. The pattern of acoustic variation observed in the languages illustrated here is typical of many small-vowel languages which have a large range of place of articulation contrasts (e.g. Marshallese, Choi 1992; Tashlhiyt Berber, Coleman 1999). It is well known that consonant place of articulation primarily influences F2 of neighbouring vowels (e.g. Stevens and House 1961), with more palatalised consonant constrictions resulting in higher F2 values and velar constrictions resulting in lower values. The close vowels in Warlpiri, Bininj Gun-wok and Iwaidja (Figures 1–3) show a great deal of variation in the F2 dimension which is highly suggestive of consonant-vowel coarticulation. A pattern of greater F2 variation (for the most part) is also observed in Arrernte central vowels, due to consonantal context, although stronger effects of preceding consonant on vowel F2 are observed (Tabain and Breen 2011). Laminal alveopalatal consonants exert both leftwards and rightwards coarticulatory pressure on the realisation of central vowels, with fronting and occasional raising in Arrernte (Tabain and Breen 2011), although both raising and fronting is consistently observed in Burarra and Gupapuyngu (Graetzer 2012). Labials tend to lower both F1 and F2 in all the languages for which we have acoustic phonetic analyses, and retroflex consonants appear to exert most effect on F3 values, as has also been found in many other languages of the world that have retroflex consonants. Moreover velar consonants in Australian languages tend for the most part to have a more uvular constriction resulting in a lower F2 locus than English (for example Butcher and Tabain 2004; Graetzer 2012). The high degree of phonetic variation and vowel allophony in languages like Arrernte has also been interpreted in certain quarters as evidence that consonants carry more functional load in terms of phonological contrast (e.g. see Breen 2001; Butcher and Tabain 2004; Tabain and Breen 2011).

Sound patterns of Australian Languages 

 99

2.2 Australian creoles and Aboriginal English The phonetics and phonology of Australian creoles and Aboriginal English are influenced to varying degrees by the phonetics and phonology of Australian Aboriginal languages. This influence ranges from creoles whose phonology is heavily influenced by the local Indigenous language, through to the lightest varieties of Aboriginal English, which may be distinguished from Standard Australian English (SAE), for example, only by the use of a clear (non-velarised) /l/ in post-vocalic position. Varieties of Aboriginal Australian English (AAE), which are heavily influenced by the Indigenous substrate, may have a very restricted set of vowels compared to SAE (e.g. Butcher and Anderson 2008; Butcher 2008; Jones et al. 2011). A complicating factor is that, while the vowel systems are small, allophonic variation is wide and the rules of allophony may vary from those for the standard accent. Amongst the most striking examples of this is the considerable fronting and raising of vowels in the presence of palatal consonants. Thus cat and dam are [kæt] and [dæm], but catch and yam are [kɛtɕ] and [jɛm] respectively. Likewise, food is [fʊd] or [pʊd], but shoot is [ʃʉt] or [sʉt]. There is no phonetic diphthongisation of the close monophthongs (/i/ → [əi], /ʉ/ → [əʉ]) as in broad SAE. Less striking, but nevertheless very consistent, is the lack of anticipatory assimilation of nasality in vowels. Thus dance and sing are pronounced [dæns] and [sɪŋ], rather than [dæ̃ns] and [sɪ̃ŋ], as in SAE. This speech habit appears to be a persistent feature of AAE phonetics, even for AAE speakers with no knowledge of an Indigenous language. The diphthongs /ɑe, æɔ, æɪ/ and /əʉ/ are much shortened in many creoles; /æɔ/ and /ɑe/ and generally both become [ɐ], as in [ɐs] for house and [mɐt] for might ([d̪ɛt ˈɐs mɐt bɪ ˈɛmti]). In Gurindji Kriol, for example, rising and falling diphthongs have shorter trajectories than compariable diphthongs in Katherine English (Jones et al. 2011). The vowel in words like boat also tends to be somewhat monophthongal. In general, however, diphthongs receive a more standard (broad) pronunciation in all but the heaviest varieties of AAE. In the more acrolectal varieties the monophthongs are also very similar to those of broad SAE. Figure 4 shows the mean formant frequency values for the standard /hVd/ vowel set as spoken by four adult female speakers of AAE. All the speakers live (mainly) in Alice Springs and speak an Indigenous language as their first language (2 Eastern Arrernte, 1 Warlpiri and 1 Western Desert Language). The values are compared with those of a group of 92 female speakers of SAE from South Australia (Butcher 2006b). Since the data are not normalised, some differences between the two groups may be attributed to the fact that the Aboriginal speakers are somewhat older than the SAE speakers and may have somewhat longer vocal tracts. Thus all first formant values (except those of /oː/) are lower in the AAE group. Of the second formant values, however, only those of the high front and central vowels (/iː, ɪ, e, ʉː, ɜː/) are lower; all others are somewhat higher. This suggests that the AAE speakers are using a somewhat smaller overall vowel space. The comparison of the two superimposed vowel spaces

100 

 Janet Fletcher and Andrew Butcher

shows that the lower boundaries of the AAE and Indigenous language spaces are very similar, i.e. that the /ɐː, ɐ/ pair in AAE is very close to the /ɐ/ of the Indigenous languages and the /e/ and /oː/ of AAE are close to the Indigenous /ɪ/ and /ʊ/ respectively. Thus, whereas the SAE vowel space represents an expansion in all directions compared with the Indigenous space, the AAE space represents an expansion in an “upward” (lower F1) direction only. Within their respective spaces, the relative positions of the vowels are quite similar across the two varieties. In AAE the /æ/ is rather closer and the /oː/ rather more open relative to their neighbours. Both /ʉː/ and /ɜː/ are slightly further back. All of these differences can be viewed as more conservative features, as the movements in question — i.e. lowering of /æ/, raising of /oː/, fronting of /ʉː/ and /ɜː/ –have occurred comparatively recently in SAE. The diphthongs of AAE fall largely within the space defined by the monophthongs. In terms of relative movement within the vowel space, the main differences from SAE concern /ɑe/ and /æɔ/, both of which have somewhat shorter trajectories than the standard accent. Phonetically these vowels are more like [ɑə] and [æɞ] in AAE.

Figure 4: Mean formant values of monophthongs in 4 female Australian Aboriginal English speakers, compared with Standard Australian English of South Australia (from Butcher and Anderson 2008).

Sound patterns of Australian Languages 

 101

3 Consonants 3.1 General overview While the vowel systems of Australian languages are somewhat typical of many smallvowel languages (with the notable exception of the Arandic languages and the general lack of phonetically nasalised vowels), the consonant systems show a number of unique typological features. A prototypical consonant inventory for Australian languages is shown in Table  2 (after Capell 1967: 86; Yallop 1982: 59, 65–66; Dixon 2002: 549; Butcher 2006a). Butcher (2006a) describes the classic consonant inventory of an Australian language as “long and thin”. This is due to the lack of a voicing or voice-onset-time contrast in the stop series and a lack of contrastive fricatives for most languages. This is offset by multiple place of articulation contrasts across stops and sonorant classes. A notable exception to this pattern are the Daly languages (spoken around the Daly river in the Northern Territory) like Ngan’gityemerri that have contrastive fricatives as well as a voiced/voiceless stop series (Reid 2011). As a general rule, however, the proportion of sonorants to obstruents in Australian languages is approximately 70% to 30%, whereas in the majority of the world’s languages the proportion is the other way round. Another important general feature of Australian languages is that the full set of phonological contrasts is only found word-medially and neutralisation is typical in word-initial and word-final contexts. The number of place of articulation contrasts ranges from a minimum of four to a maximum of seven. About half of the languages surveyed by Busby (1980) have five places of articulation; a further one third have six. A few languages such as Yanyuwa have been analysed as having a contrast between front and back velars (Butcher and Tabain 2004), while Arandic languages like Kaytetye have contrastive pre-palatalised apico-alveolars (e.g. Turpin 2000; Breen 2001; Koch 2006).² Articulatory and acoustic studies of Central Arrernte also show (phonetic) pre-palatalisation of retroflex stops in utterance-initial and word-initial əC contexts in older speakers (e.g. Tabain 2009; Graetzer 2012). Many of the languages spoken in Arnhem land (e.g. Dalabon, Bininj Gun-wok and most varieties of Yolngu Matha) and Cape York (e.g. most Wik languages) also have a contrastive glottal stop (Harvey 1991; Evans 1995a). Almost all Australian languages have two distinct rhotic sounds and some have three. Palatal and labialvelar glides also occur in every Australian consonant system. Contrastive fricatives are only found in a handful of languages, including Marrithiyel, Ngan’gityemerri and Anguthimri, for example. A group of languages spoken largely in the northern part

2  Alternative analyses have also been proposed, in which these sounds are treated as clusters – see e.g. Harvey (2011).

102 

 Janet Fletcher and Andrew Butcher

Table 2: Prototypical consonant inventory of Australian languages showing six places of articulation, and the four major classes, stops, nasals, laterals and rhotics (after Butcher 2006a, Yallop 1982, Dixon 2002). consonants

labial

dental

alveolar

retroflex

alveopalatal

velar

stops nasals laterals rhotics glides

p m

t ̪ n̪ l ̪

t n l r

ʈ ɳ ɭ ɹ

c ɲ ʎ

k ŋ

w

j

Table 3: Consonant inventory of a Double apical/ Double laminal Australian language: Central and Eastern Arrernte. Consonants are grouped according to natural classes. consonants

peripheral labial velar

apical alveolar

retro

laminal dental

alveopal

stop nasal

p m

k ŋ

t n

ʈ ɳ

t  ̪ n

c ɲ

pre-stopped nasal

p

k

t n l r

ʈ ɳ ɭ ɹ

t ̪ n̪ l ̪

c

w

ɰ

lateral rhotic glide

m

ŋ

ɲ ʎ

j

of Australia also have a more complex stop series which is described as a fortis-lenis contrast or a short-long contrast (e.g. Bininj Gun-wok, Dalabon, Evans 2003). Of particular note in Australian languages is the rich set of coronal place of articulation contrasts, with many “double-apical” and/or “double-laminal” languages contrasting apico-alveolar, retroflex, laminal alveopalatal, and laminal dental stops and nasals. A good example of a language that exhibits this system is Arrernte, as shown in Table 3. The peripheral natural class includes labial and velar places of articulation. Of additional note here also is the incidence of pre-stopped nasals that contrast with their plain counterparts. This is a typical feature of the Arandic languages (e.g. see Breen 2001). Just over 20% of the languages in Busby’s (1980) survey have an apical contrast within the coronal set, with no laminal contrast. This type of system is found in many languages of Western Australia and the Northern Territory. For example Warlpiri, unlike its neighbour Arrernte, does not have a contrastive laminal dental but has a lamino-alveopalatal stop, nasal, lateral and approximant. There is experimental phonetic evidence that there is a degree of phonetic variation among the individual coronal categories (e.g. see Butcher 1995). Electropalatographic analyses show that the lamino-alveopalatal nasal, in Warlpiri for example, is a more fronted articulation compared to its counterpart in Central Arrernte (e.g. Tabain, Fletcher, and Butcher 2011). However acoustic characteristics of the burst frequencies of apical stops in Warlpiri

Sound patterns of Australian Languages 

 103

and Central Arrernte also show a degree of similarity (Tabain and Beare 2011). On the other hand, about a quarter of Busby’s languages have a laminal distinction (dental versus alveopalatal) in the stop and nasal classes but lack an apical distinction (alveolar versus retroflex). Most of these are found in the Cape York Peninsula and eastern coastal regions of Queensland and New South Wales. With regard to laterals, languages in the eastern third of Australia, as well as Garrwa and Waanyi have a single apical lateral (after Dixon 2002: 565), whereas many other languages have a lateral corresponding to each of the four coronal categories. However, some languages may have only /l/, /ɭ/ and /ʎ/ or even only /l/ and /ɭ/. As a general rule, the laminal contrast amongst laterals is found only in the presence of an apical contrast in that series. The Iwaidjan languages also have a set of pre-lateralised stops that contrast with lateral approximants (see Evans 1995a: 735). The majority of Australian languages neutralise alveolar and retroflex contrasts in word-initial position, regardless of their status as single/double apical or single/ double laminal languages. There are some notable exceptions like Nunggubuyu, now known as Wubuy, which has a four-way contrast between /t ̪ t ʈ c/ in word-initial position (Heath 1984). Recent experimental analysis (acoustic) shows that coronal contrasts are clearly maintained in word-initial position, regardless of position in utterance (Bundgaard-Nielsen et al. 2012). According to Evans’ (1995a) and Dixon’s (2002) survey, laminal alveopalatals and laminal dentals can also be neutralised in word-final position in some languages. Other typical patterns of allophony across the board include voicing and/or leniting of stops word-medially, whereas voiceless unaspirated allophones, with the notable exception of laminal alveopalatals, which are generally spirantised, tend to be realised word-initially or in word-final position. Thus, whilst the overwhelming majority of the world’s languages (81% of UPSID entries) have only one coronal place of articulation, the most frequently occurring pattern in Australian languages extends to four coronal categories (only 3.5% of UPSID languages have three or four coronal categories; of these 16 languages, 14 are Australian). The question arises as to how (and how consistently) speakers distinguish between these categories articulatorily. Two important measures of tongue articulation are (1) anteriority (how far forward is the contact in the mouth?) and (2) contact depth (how spread out is the contact from back to front?). Table 4 and Figure 5, adapted from Butcher (2012) show the results of measuring these distances from palatograms.³ Both measures are expressed as a percentage of the distance from the back

3 Note that all these languages (with the arguable exception of Guugu Yimidhirr) have an apical contrast. The “non-contrastive apicals” are taken from word-initial position in all languages. Only E. Arrernte, Guugu Yimidhirr, Murrinh-Patha, Yanyuwa and Yindjibarndi have a laminal contrast. The “non-contrastive laminals” are taken from intervocalic position in the single-laminal languages Bininj Gun-wok, Burarra, Nyangumarta, Warlpiri and the Western Desert Language, and therefore may include both “more dental” and “more alveopalatal” diaphones of this category.

104 

 Janet Fletcher and Andrew Butcher

of the hard palate (border with the velum) and the edges of the upper front incisors, along the centre line. In general terms, there appears to be a clear difference in contact depth between the two major categories of apical and laminal. Furthermore within these categories there is a clear gradation in anteriority between the three members of each. T-tests show that there is no significant difference (p2min.o

nya-nya intake-pst

jaru-ngga language-loc

jarragab-gula talk-loc ‘We saw you when we (but not you) were talking language.’ (76) ah jubu=rnanyjurra ah just=1min.sbj>2aug.o

gurrugurru=warla hear.rdp=foc

ya-na-rni, go-prs-hither

jarragab-jirri talk-all ‘I just came up listening to you mob talking.’

8 Future directions The last 40–50 years has seen an enormous amount of development in our understanding of the grammatical structure of Australian languages, and particularly the unique contributions they make to cross-linguistic typology in the areas of constituency relations, case relations and configurationality. There remain, however, many issues to be explored and enriched by future research. Major areas in need of substantial research include the role that discourse and information structure play in the organisation of syntactic structure (e.g. Mushin and Baker 2008), and the interactions between syntactic structure, discourse and prosody (see Fletcher and Butcher (this volume) for further discussion, also Croft (2007); Evans, Fletcher, and Ross (2008); Ross (2011) and Schultze-Berndt and Simard (2012)). Another important direction for future research lies in the detailed syntactic analysis of individual languages. A large majority of the work discussed above has been based on detailed analysis of only a small number of Australian languages (most

Constituency and Grammatical Relations in Australian languages  

 253

notably, Warlpiri, Wambaya, Jiwarli, Diyari, Kayardild, Dyirbal and a few others). We cannot, however, assume that surface typological similarities across Australian languages necessarily reflect identical syntactic organisation at all levels. Hopefully, now that many more Australian languages have been the subject of grammatical description and documentation, we can move towards more nuanced research into their syntactic structures and reveal many more interesting properties to expand our understanding of the nature of language.

References Anderson, Stephen 1976 On the notion of subject in ergative languages. In: Charles N. Li (ed.), Subject and Topic, 1–23. New York: Academic Press. Andrews, Avery 1996 Semantic case-stacking and inside-out unification. Australian Journal of Linguistics 16(1): 1–55. Andrews, Avery 2007 Relative Clauses. In: Timothy Shopen (ed.), Language Typology and Syntactic Description, 2nd edn., vol. 2, 206–236. Cambridge: Cambridge University Press. Austin, Peter 1981a A Grammar of Diyari, South Australia. Cambridge: Cambridge University Press. Austin, Peter 1981b Switch-reference in Australia. Language 57(2): 309–334. Austin, Peter 1981c Case marking in southern Pilbara languages. Australian Journal of Linguistics 1: 211–226. Austin, Peter (ed.) 1988 Complex Sentence Constructions in Australian Languages. (Typological Studies in Language 15) Amsterdam: John Benjamins. Austin, Peter 2001 Word order in a free word order language: the case of Jiwarli. In: Jane, David Nash, Mary Laughren, Peter Austin and Barry Alpher (eds.), Forty Years On: Ken Hale and Australian Languages (Pacific Linguistics 512), 305–323. Canberra: Australian National University. Austin, Peter and Joan Bresnan 1986 Non-configurationality in Australian Aboriginal languages. Natural Language and Linguistic Theory 14(2): 215–268. Baker, Brett 2002 How referential is agreement? The interpretation of polysynthetic dis-agreement morphology in Ngalakgan. In: Nicholas Evans and Hans-Jürgen Sasse (eds.), Problems of Polysynthesis (Studia Typologica 4), 51–85. Berlin: Akademie Verlag. Baker, Brett 2008 The interpretation of complex nominal expressions in Southeast Arnhem Land languages. In: Ilana Mushin and Brett Baker (eds.), Discourse and Grammar in Australian Languages (Studies in Language Companion Series 104), 135–166. Amsterdam: John Benjamins. Baker, Brett and Rachel Nordlinger 2008 Noun-adjective compounds in Gunwinyguan languages. In: Miriam Butt and Tracy Holloway King (eds.). Proceedings of the LFG08 Conference, 109–128. Stanford: CSLI Publications. Baker, Brett, Kate Horrack, Rachel Nordlinger and Louisa Sadler 2010 Putting it all together: agreement, incorporation, coordination and external possession in Wubuy (Australia). In: Miriam Butt and Tracy Holloway King (eds.), Proceedings of the LFG10 Conference, 64–84. Stanford: CSLI Publications. Baker, Mark 1988 Incorporation: A Theory of Grammatical Gunction Changing. Chicago: University of Chicago Press. Baker, Mark 1996 The Polysynthesis Parameter. Oxford: Oxford University Press. Baker, Mark 2001 The natures of nonconfigurationality. In: Mark Baltin and Chris Collins (eds.) Handbook of Contemporary Syntactic Theory, 407–438. Oxford: Blackwell.

254 

 Rachel Nordlinger

Bicevskis, Katie 2012 Incorporated modifier scope in Gunwinyguan languages. Melbourne: University of Melbourne BA Honours thesis. Bittner, Maria and Kenneth Hale 1995 Remarks on definiteness in Warlpiri. In: Emmon Bach, Eloise Jelinek, Angelika Kratzer and Barbara Partee (eds.), Quantification in Natural Languages (Studies in Linguistics and Philosophy 54), 81–107. Dordrecht: Kluwer Academic Publishers. Bittner, Maria and Kenneth Hale 1996a The structural determination of case and agreement. Linguistic Inquiry 27(1): 1–68. Bittner, Maria and Kenneth Hale 1996b Ergativity: towards a theory of a heterogeneous class. Linguistic Inquiry 27(4): 531–604. Blake, Barry 1983 Structure and word order in Kalkatungu: the anatomy of a flat language. Australian Journal of Linguistics 3(2): 143–175. Blake, Barry 1977 Case Marking in Australian Languages. (Linguistics Series 23) Canberra: Australian Institute of Aboriginal Studies. Blake, Barry 1979 Pitta Pitta. In: R. M. W. Dixon and Barry Blake (eds.), Handbook of Australian languages, vol. 1, 183–242. Canberra: Australian National University Press. Blake, Barry 1987 Australian Aboriginal Grammar. London: Croom Helm. Blake, Barry 1993 Verb affixes from case markers: some Australian examples. La Trobe Working Papers in Linguistics 6: 33–58. Blake, Barry 1999 Nominal marking on verbs: some Australian cases. Word 50(3): 299–317. Blake, Barry 2001 The noun phrase in Australian languages. In: Jane Simpson, David Nash, Mary Laughren, Peter Austin and Barry Alpher (eds.), Forty Years On: Ken Hale and Australian Languages (Pacific Linguistics 512), 415–425. Canberra: Australian National University. Blythe, Joseph 2009 Doing referring in Murrinh-Patha conversation. Sydney: University of Sydney PhD thesis. Bowe, Heather 1990 Categories, Constituents, and Constituent Order in Pitjantjatjara: An Aboriginal Language of Australia. London: Routledge. Bowern, Claire 2006 Correlates of nonconfigurationality. In: NELS 36. University of Massachusetts, Amherst. http://scholarship.rice.edu/bitstream/handle/1911/8323/Bowern_nels.pdf (accessed February 2013) Bowern, Claire 2008 Bardi arguments: referentiality, agreement, and omission in Bardi discourse. In: Ilana Mushin and Brett Baker (eds), Discourse and Grammar in Australian Languages (Studies in Language Companion Series 104), 59–85. Amsterdam: John Benjamins. Bresnan, Joan 2001 Lexical-functional Syntax. Oxford: Blackwell. Bresnan, Joan and Sam Mchombo 1987 Topic, pronoun and agreement in Chichewa. Language 63(4): 741–782. Capell, Arthur 1956 A New Approach to Australian Linguistics (Handbook of Australian languages, part 1). (Oceania Linguistic Monographs 1) Sydney: University of Sydney. Capell, Arthur 1962 Some Linguistic Types in Australia. (Oceania Linguistic Monographs 7) Sydney: University of Sydney. Chomsky, Noam 1981 Lectures on Government and Binding. (Studies in Generative Grammar 9) Dordrecht: Foris Publications. Chomsky, Noam 1995 The Minimalist Program. (Current Studies in Linguistics 28) Cambridge, MIT: MIT Press. Comrie, Bernard 1981a Ergativity and grammatical relations in Kalaw Lagaw Ya (Saibai dialect). Australian Journal of Linguistics 1(1): 1–42. Comrie, Bernard 1981b Language Universals and Linguistic Typology. Oxford: Blackwell. Comrie, Bernard 1989 Language Universals and Linguistic Typology. 2nd edn. Chicago: University of Chicago.

Constituency and Grammatical Relations in Australian languages  

 255

Cristofaro, Sonia 2003 Subordination. Oxford: Oxford University Press. Croft, William 2007 Intonation units and grammatical structure in Wardaman and in cross-linguistic perspective. Australian Journal of Linguistics 27(1): 1–39. Delancey, Scott 1981 An interpretation of split ergativity and related patterns. Language 57(3): 626–657. Dench, Alan 1982 The development of an accusative marking pattern in the Ngayarda languages of Western Australia. Australian Journal of Linguistics 2(1): 43–60. Dench, Alan 1988 Complex sentences in Martuthunira. In: Peter Austin (ed.), Complex Sentence Constructions in Australian Languages (Typological Studies in Language 15), 97–140. Amsterdam: John Benjamins. Dench, Alan 1994 Panyjima. In: R. M. W. Dixon and Barry Blake (eds.), Handbook of Australian Languages, vol. 4, 125–243. Melbourne: Oxford University Press. Dench, Alan 1995 Martuthunira: A Language of the Pilbara Region of Western Australia. (Pacific Linguistics C-125) Canberra: Australian National University. Dench, Alan 2006 Case marking strategies in subordinate clauses in Pilbara languages – some diachronic speculations. Australian Journal of Linguistics 26(1): 81–105. Dench, Alan and Nicholas Evans 1988 Multiple case-marking in Australian languages. Australian Journal of Linguistics 8(1): 1–47. Diessel, Helge 2001 The ordering distribution of main and adverbial clauses: A typological study. Language 77(3): 433–455. Dixon, R. M. W. 1972 The Dyirbal Language of North Queensland. Cambridge: Cambridge University Press. Dixon, R. M. W. 1977 A Grammar of Yidiny. Cambridge: Cambridge University Press. Dixon, R. M. W. 1979 Ergativity. Language 55(1): 59–138. Dixon, R. M. W. 1980 The Languages of Australia. Cambridge: Cambridge University Press. Dixon, R. M. W. 1994 Ergativity. Cambridge: Cambridge University Press. Dixon, R. M. W. 2002 Australian Languages: Their Nature and Development. Cambridge: Cambridge University Press. Dixon, R. M. W. 2006 Adjective classes in typological perspective. In: R. M. W. Dixon and Alexandra Y. Aikhenvald (eds.), Adjective Classes: A Cross-Linguistic Typology, 1–49. Oxford: Oxford University Press. Donaldson, Tamsin 1980 Ngiyambaa: the language of the Wangaybuwan. Cambridge: Cambridge University Press. Dowty, David 1991 Thematic Proto-Roles and Argument Selection. Language 67(3): 547–619. Du Bois, John W. 1987 The discourse basis of ergativity. Language 63(4): 805–855. Evans, Nicholas 1995 A Grammar of Kayardild: With Historical-comparative Notes on Tangkic. (Mouton Grammar Library 15) Berlin: Mouton de Gruyter. Evans, Nicholas 1996 The syntax and semantics of body part incorporation in Mayali. In: Hilary Chappell and William McGregor (eds.), The Grammar of Inalienability: A Typological Perspective on Body Part Terms and the Part-Whole Relation (Empirical Approaches to Language Typology 14), 65–109. Berlin: Mouton de Gruyter. Evans, Nicholas 1997 Role or cast? Complex predicates and noun incorporation in Mayali. In: Alex Alsina, Joan Bresnan and Peter Sells (eds.), Complex Predicates (CSLI Lecture Notes 64), 397–430. Stanford: CSLI Publications. Evans, Nicholas 2002 The true status of grammatical object affixes: evidence from Bininj Gunwok. In: Nicholas Evans and Hans-Jürgen Sasse (eds.), Problems of Polysynthesis (Studia Typologica 4), 15–50. Berlin: Akademie Verlag. Evans, Nicholas 2003 Bininj Gun-Wok: A Pan-dialectal Grammar of Mayali, Kunwinjku and Kune, vols 1 & 2. (Pacific Linguistics 541) Canberra: Australian National University.

256 

 Rachel Nordlinger

Evans, Nicholas 2006 Who said polysynthetic languages avoid subordination? Multiple subordination strategies in Dalabon. Australian Journal of Linguistics 26(1): 31–58. Evans, Nicholas, Janet Fletcher and Belinda Ross 2008 Big words, small phrases: mismatches between pause units and the polysynthetic word in Dalabon. Linguistics 46(1): 89–129. Evans, Nicholas and Rachel Nordlinger 2004 Extreme Morphological Shift: Verbal case in Kayardild. Paper presented at LFG04 Conference, Christchurch N.Z., July 2004. Evans, Nicholas and Hans-Jürgen Sasse 2002 Introduction: problems of polysynthesis. In: Nicholas Evans and Hans-Jürgen Sasse (eds.), Problems of Polysynthesis (Studia Typologica 4), 1–13. Berlin: Akademie Verlag. Forshaw, William 2011 A continuum of incorporation: Noun incorporation in Murrinh-Patha. Melbourne: University of Melbourne BA Honours thesis. Green, Ian 1989 Marrithiyel: A language of the Daly River region of Australia’s Northern Territory. Canberra: Australian National University PhD thesis. Gaby, Alice R. 2006 A grammar of Kuuk Thaayorre. Melbourne: University of Melbourne PhD thesis. Gaby, Alice 2008 Pragmatically case-marked: non-syntactic functions of the Kuuk Thaayorre ergative suffix. In: Ilana Mushin and Brett Baker (eds.), Discourse and Grammar in Australian Languages (Studies in Language Companion Series 104), 111–134. Amsterdam: John Benjamins. Gaby, Alice 2010 From discourse to syntax and back: the lifecycle of Kuuk Thaayorre ergative morphology. Lingua 120(7): 1677–1692. Garrett, Andrew 1990 The origin of NP split ergativity. Language 66(2): 261–296. Genetti, Carol 1991 From postposition to subordinator in Newari. In: Elizabeth C. Traugott amd Bernd Heine (eds.), Approaches to Grammaticaliszation, vol. 2 (Typological Studies in Language 19), 227–255. Amsterdam: John Benjamins, Goddard, Cliff 1982 Case systems and case marking in Australian languages: a new interpretation. Australian Journal of Linguistics 2(1): 167–196. Hale, Kenneth 1967 Some Productive Rules in Lardil (Mornington Island) Syntax. In Carl Georg von Brandenstein, Arthur Capell and Kenneth Hale (eds.), Papers in Australian Linguistics, no. 2, 63–73. (Pacific Linguistics A-11). Canberra: Australian National University. Hale, Kenneth 1976 The adjoined relative clause in Australia. In: R. M. W. Dixon (ed.), Grammatical Categories in Australian Languages (Linguistics Series 22), 78–105. Canberra: Australian Institute of Aboriginal Studies. Hale, Kenneth 1981 On the position of Walbiri in a typology of the base. [Distributed by] Bloomington: Indiana University Linguistics Club. Hale, Kenneth 1982 Some essential features of Warlpiri verbal clauses. In: Stephen Swartz (ed.), Papers in Warlpiri Grammar: Papers in Memory of Lothar Jagst (SIL-AAB A6), 217–315. Darwin: Summer Institute of Linguistics. Hale, Kenneth 1983 Warlpiri and the grammar of nonconfigurational languages. Natural Language and Linguistic Theory 1: 5–47. Hale, Kenneth 1989. On nonconfigurational structures. In: László Marácz and Pieter Muysken (eds.), Configurationality: The Typology of Asymmetries (Studies in Generative Grammar 34), 293–300. Dordrecht: Foris Publications. Hale, Kenneth 1992 Basic word order in two ‘free word order’ languages. In: Doris Payne (ed.) Pragmatics of Word Order Flexibility (Typological Studies in Language 22), 63–82. Amsterdam: John Benjamins. Hale, Kenneth 1994 Core structures and adjunctions in Warlpiri syntax. In: Norbert Corver and Henk C. van Reimsdijk (eds.), Studies on Scrambling: Movement and Non-movement Approaches to Free Word-order Phenomena (Studies in Generative Grammar 41), 185–219. Berlin: Mouton de Gruyter,

Constituency and Grammatical Relations in Australian languages  

 257

Harris, Alice and Lyle Campbell 1995 Historical Syntax in Cross-linguistic Perspective. Cambridge: Cambridge University Press. Harvey, Mark 1992 The noun phrase in Australian languages: a comment. Australian Journal of Linguistics 12(2): 307–315. Harvey, Mark 1996 Body parts in Warray. In: Hilary Chappell and William McGregor (eds.), The Grammar of Inalienability: A Typological Perspective on Body Part Terms and the Part-Whole Relation, 111–154. Berlin: Mouton de Gruyter. Heath, Jeffrey 1978 Ngandi Grammar, Texts and Dictionary. Canberra: Australian Institute of Aboriginal Studies. Heath, Jeffrey 1979 Is Dyirbal ergative? Linguistics 17(5–6): 401–464. Heath, Jeffrey 1984 Functional Grammar of Nunggubuyu. (AIAS New Series 53) Canberra: Australian Institute of Aboriginal Studies. Heath, Jeffrey 1986 Syntactic and lexical aspects of nonconfigurationality in Nunggubuyu (Australia). Natural Language and Linguistic Theory 4: 375-408. Himmelmann, Nikolaus P. and Eva Schultze-Berndt 2005 Issues in the syntax and semantics of participant-oriented adjuncts: an introduction. In: Nikolaus P. Himmelmann and Eva SchultzeBerndt (eds.), Secondary Predication and Adverbial Modification: The Typology of Depictives, 1-68. Oxford: Oxford University Press. Horrack, Kate 2010 Expressing possession in Wubuy: A Lexical-Functional approach. Armidale, N. S. W.: University of New England BA Honours thesis. Huang, C-T James 1984 On the distribution and reference of empty pronouns. Linguistic Inquiry 5: 531–574. Jelinek, Eloise 1984 Empty categories, case and configurationality. Natural Language and Linguistic Theory 2: 39–76. Karlsson, Fred 2007 Constraints on multiple center-embedding of clauses. Journal of Linguistics 43: 365–392. Keen, Sandra 1983 Yukulta. In: R. M. W Dixon and Barry Blake (eds.), Handbook of Australian Languages, vol. 3, 190–304. Canberra: Australian National University Press. Keenan Edward L. 1985 Relative Clauses. In: Timothy Shopen (ed.), Language Typology and Syntactic Description, vol. 2, 141–170. Cambridge: Cambridge University Press. Kroeger, Paul, R. 2004 Analyzing Syntax: A Lexical-functional Approach. Cambridge: Cambridge University Press. Laughren, Mary 1989 The configurationality parameter and Warlpiri. In: László Marácz and Pieter Muysken (eds.), Configurationality: The Typology of Asymmetries (Studies in Generative Grammar 34), 319–353. Dordrecht: Foris Publications. Laughren, Mary 2000 Constraints on the pre-auxiliary position in Warlpiri and the nature of the auxiliary. In: John Henderson (ed.), Proceedings of the 1999 conference of the Australian Linguistic Society. http://www.als.asn.au/proceedings/als1999/proceedings.html. Laughren, Mary 2002 Syntactic constraints in a ‘free word order’ language. In: Mengistu Amberber and Peter Collins (eds.), Language Universals and Variation (Perspectives on Cognitive Science), 83–130. Westport, CT: Praeger. Laughren, Mary, Robert Pensalfini and Tom Mylne 2005 Accounting for verb initial order in an Australian language. In: Andrew Carnie, ‎Heidi Harley and Sheila Ann Dooley (eds.), Verb First: On the Syntax of Verb-initial Languages (Linguistik Aktuell/Linguistics Today 73), 367–401. Amsterdam: John Benjamins. Legate, Julie 2001 The configurational structure of a nonconfigurational language. Linguistic Variation Yearbook 1: 63–99. Legate, Julie 2003 Reconstructing nonconfigurationality. In: Anne-Marie Di Sciullo (ed.), Asymmetry in Grammar, vol. 1: Syntax and Semantics (Linguistik Aktuell/Linguistics Today 57), 99–116. Amsterdam: John Benjamins.

258 

 Rachel Nordlinger

Legate, Julie 2005 Split absolutive. In: Alana Johns, Diane Massam and Juvenal Ndayiragije (eds.), Ergativity: Emerging Issues (Studies in Natural Language and Linguistic Theory 65), 143–171. Dordrecht: Kluwer Academic Publishers. Legate, Julie 2008 Dyirbal ergativity. Paper presented at Linguistic Society of America annual meeting, Chicago IL, 3–6 January. Legate, Julie 2009 Clausal recursion and embedding in Warlpiri. MS. http://www.ling.upenn. edu/~jlegate/warlpiri_recursion.pdf (accessed February 2013). Lehmann, Christian 1988 Towards a typology of clause linkage. In: John Haiman and Sandra A. Thompson (eds.), Clause Combining in Grammar and Discourse (Typological Studies in Language 18), 181–225. Amsterdam: John Benjamins. Levin, Beth 1983 On the Nature of Ergativity. Cambridge, MA: Massachusetts Institute of Technology PhD thesis. Malouf, Robert 2000. A head-driven account of long-distance case assignment. In: Ronnie Cann Claire Grover and Philip Miller (eds.), Grammatical Interfaces in HPSG (Studies in ConstraintBased Lexicalism), 201–214. Stanford: CSLI Publications,. Manning, Christopher 1996 Ergativity: Argument Structure and Grammatical Relations. (Dissertations in Linguistics) Stanford: CSLI Publications. Marácz, László and Pieter Muysken 1989 Introduction. In: László Marácz and Pieter Muysken (eds.), Configurationality: The Typology of Asymmetries (Studies in Generative Grammar 34), 1–38. Dordrecht: Foris Publications. Marácz László and Pieter Muysken (eds.) 1989 Configurationality: The Typology of Asymmetries. (Studies in Generative Grammar 34) Dordrecht: Foris Publications. Marantz, Alec 1984 On the Nature of Grammatical Relations. Cambridge, MA: MIT Press. McConvell, Patrick 2006 Grammaticalization of demonstratives as subordinate complementizers in Ngumpin-Yapa. Australian Journal of Linguistics 26(1): 107–137. McGregor, William 1988 Mood and subordination in Kuniyanti. In: Peter Austin (ed.), Complex Sentence Constructions in Australian Languages (Typological Studies in Language 15), 37–68. Amsterdam: John Benjamins. McGregor, William 1990 A Functional Grammar of Gooniyandi. (Studies in Language Companion Series 22) Amsterdam: John Benjamins. McGregor, William 1992a The semantics of ergative case marking in Gooniyandi. Linguistics 30(2): 275–318. McGregor, William 1992b The noun phrase as a grammatical category in (some) Australian languages: a reply to Mark Harvey. Australian Journal of Linguistics 12(2): 315–319. McGregor, William 1994 Complex sentence constructions in Nyulnyul, Western Australia. Functions of Language 1: 25–66. McGregor, William 2004 The Languages of the Kimberley, Western Australia. London: Routledge Curzon. McGregor, William 2005 Quantifying depictive secondary predicates in Australian languages. In: Nikolaus P. Himmelmann and Eva Schultze-Berndt (eds.), Secondary Predication and Adverbial Modification: The Typology of Depictives, 173–200. Oxford: Oxford University Press. McGregor, William 2006 Focal and optional ergative marking in Warrwa (Kimberley, Western Australia). Lingua 116(4): 393–423. McGregor, William 2009 Typology of ergativity. Language and Linguistics Compass 3(1): 480–508. McGregor, William 2010 Optional ergative case marking systems in a typological-semiotic perspective. Lingua 120(7): 1610–1636. McGregor, William and Jean-Christophe Verstraete 2010 Optional ergative marking and its implications for linguistic theory. Lingua 120(7): 1607–1609.

Constituency and Grammatical Relations in Australian languages  

 259

McKay, Graeme 2007 Noun incorporation in Rembarrnga discourse. In: Jeff Siegel, John Lynch and Diana Eades (eds.), Language Description, History and Development: Linguistic Indulgence in Memory of Terry Crowley (Creole Language Library 30), 41–52. Amsterdam: John Benjamins. Meakins, Felicity and Rachel Nordlinger 2014 A Grammar of Bilinarra. (Pacific Linguistics 640) Berlin: Mouton de Gruyter. Meakins, Felicity and Carmel O’Shannessy 2010 Ordering arguments about: word order and discourse motivations in the development and use of the ergative marker in two Australian mixed languages. Lingua 120(7): 1693–1713. Merlan, Francesca 1981 Some functional relations among subordination, mood, aspect and focus in Australian languages. Australian Journal of Linguistics 1(2): 175–210. Mithun, Marianne 1984a How to avoid subordination. In: Proceedings of the Tenth Annual Meeting of the Berkeley Linguistics Society (1984): 493–509. Mithun, Marianne 1984b The evolution of noun incorporation. Language 60(4): 847–894. Mithun, Marianne 1986 On the nature of noun incorporation. Language 62(1): 32–37. Mushin, Ilana 2006 Motivations for second position: evidence from North-Central Australia. Linguistic Typology 10(3): 287–326. Mushin, Ilana 2008 Diverging paths: variation in Garrwa tense/aspect clitic placement. In: Ilana Mushin and Brett Baker (eds), Discourse and Grammar in Australian Languages (Studies in Language Companion Series 104), 87–109. Amsterdam: John Benjamins. Mushin, Ilana and Brett Baker (eds.) 2008 Discourse and Grammar in Australian Languages. (Studies in Language Companion Series 104) Amsterdam: John Benjamins. Nash, David 1980 Topics in Warlpiri grammar. Cambridge MA: Massachusetts Institute of Technology PhD thesis. Nichols, Johanna 1986 Head-marking grammar and dependent-marking grammar. Language 62(1): 56–119. Nordlinger, Rachel 1998a A Grammar of Wambaya. (Pacific Linguistics C-140) Canberra: Australian National University. Nordlinger, Rachel 1998b Constructive Case: Evidence from Australian Languages. (Dissertations in Linguistics) Stanford: CSLI Publications. Nordlinger, Rachel 2000 Australian case systems: towards a constructive solution. In: Miriam Butt and Tracy Holloway King (eds.), Argument Realization (Studies in Constraint-Based Lexicalism), 41–71. Stanford: CSLI Publications. Nordlinger, Rachel 2006 Spearing the emu drinking: subordination and the adjoined relative clause in Wambaya. Australian Journal of Linguistics 26(1): 5–29. Nordlinger, Rachel 2010 Verbal morphology in Murrinh-Patha: evidence for templates. Morphology 20(2): 321–341. Nordlinger, Rachel 2011 From body parts to applicatives. Paper presented at the Association for Linguistic Typology Conference, Hong Kong, July 2011. Nordlinger, Rachel and Joan Bresnan 1996 Nonconfigurational tense in Wambaya. In: Proceedings of the LFG Conference, Grenoble, August 1996. Stanford: CSLI Publications. http:// cslipublications.stanford.edu/LFG/1/lfg1.html. Nordlinger, Rachel and Louisa Sadler 2004a Nominal Tense in Cross-Linguistic Perspective. Language 80(4): 776–806. Nordlinger, Rachel and Louisa Sadler 2004b Tense beyond the verb: encoding clausal tense/ aspect/mood on nominal dependents. Natural Language and Linguistic Theory 22(3): 597–641. Nordlinger, Rachel and Louisa Sadler 2008 From juxtaposition to incorporation: an approach to generic–specific constructions. In: Miriam Butt and Tracy Holloway King, T. H. (eds.) Proceedings of the LFG08 Conference, 394–412. Stanford: CSLI Publications.

260 

 Rachel Nordlinger

Nordlinger, Rachel and Adam Saulwick 2002 Infinitives in polysynthesis: the case of Rembarrnga. In: Nicholas Evans and Hans-Jürgen Sasse (eds.), Problems of Polysynthesis (Studia Typologica 4), 185–201. Berlin: Akademie Verlag. Pensalfini, Robert 1992 Degrees of freedom: Word order in Pama-Nyungan languages. Perth: University of Western Australia BA Honours thesis. Pensalfini, Robert 1999 The rise of case suffixes as discourse markers in Jingulu – a case study of innovation in an obsolescent language. Australian Journal of Linguistics 19(2): 225–240. Pensalfini, Robert 2004 Towards a typology of nonconfigurationality. Natural Language and Linguistic Theory 22(2): 359–408. Pollard, Carl and Ivan Sag 1994 Head-driven Phrase Structure Grammar. Stanford: CSLI Publications. Plank, Frans (ed.) 1995 Double Case: Agreement by Suffixaufnahme. Oxford: Oxford University Press. Reid, Nicholas 1990 Ngan’gityemerri: A language of the Daly River region, Northern Territory of Australia. Canberra: Australian National University PhD thesis. Richards, Norvin 2013 Lardil “case stacking” and the timing of case assignment. Syntax 16(1): 42–76. Rosen, Sara T. 1989 Two types of noun incorporation: A lexical analysis. Language 65(2): 294–317. Ross, Belinda 2011 Prosody and grammar in Dalabon and Kayardild. Melbourne: University of Melbourne PhD thesis. Round, Erich 2012 Kayardild Morphology and Syntax. Oxford: Oxford University Press. Sadler, Louisa and Rachel Nordlinger 2004 Relating Morphology to Syntax. In: Louisa Sadler and Andrew Spencer (eds.), Projecting Morphology (Stanford Studies in Morphology and the Lexicon), 159–185. Stanford: CSLI Publications. Sadler, Louisa and Rachel Nordlinger 2006. Case stacking in Realizational Morphology. Linguistics 44(3): 459–487. Sadler, Louisa and Rachel Nordlinger 2010 Nominal juxtaposition in Australian languages: an LFG analysis. Journal of Linguistics 46(2): 415–452. Schmidt, Annette 1985 The fate of ergativity in dying Dyirbal. Language 61(2): 378–396. Schultze-Berndt, Eva 2006 Secondary Predicates in Australian Languages. In: Martin Everaert and Henk van Riemsdijk (eds.), The Blackwell Companion to Syntax, vol. IV, 180–208. Oxford: Blackwell. Schultze–Berndt, Eva and Nikolaus P. Himmelmann 2004 Depictive secondary predicates in crosslinguistic perspective. Linguistic Typology 8(1): 59–131. Schultze–Berndt, Eva and Candide Simard 2012 Constraints on noun phrase discontinuity in an Australian language: The role of prosody and information structure. Linguistics 50(5): 1015–1058. Silverstein, Michael 1976 Hierarchy of features and ergativity. In: R. M. W. Dixon (ed.), Grammatical Categories in Australian Languages (Linguistics Series 22), 112–171. Canberra: Australian Institute of Aboriginal Studies. Silverstein, Michael 1981 Case marking and the nature of language. Australian Journal of Linguistics 1(2): 227–244. Simpson, Jane 1983 Aspects of Warlpiri Morphology and Syntax. Cambridge MA: Massachusetts Institute of Technology PhD thesis. Simpson, Jane 1988 Case and complementiser suffixes in Warlpiri. In: Peter Austin (ed.), Complex Sentence Constructions in Australian Languages (Typological Studies in Language 15), 205–218. Amsterdam: John Benjamins. Simpson, Jane. 1991 Warlpiri Morpho-syntax: A Lexicalist Approach. (Studies in Natural Language and Linguistic Theory 23) Dordrecht: Kluwer Academic Publishers.

Constituency and Grammatical Relations in Australian languages  

 261

Simpson, Jane 2005 Depictives in English and Warlpiri. In: Nikolaus P. Himmelmann and Eva Schultze-Berndt (eds.), Secondary Predication and Adverbial Modification: The Typology of Depictives, 69–106. Oxford: Oxford University Press. Simpson, Jane 2007 Expressing pragmatic constraints on word order. In: Annie Zaenen, Jane Simpson, Tracy Holloway King, Jane Grimshaw, Joan Maling and Chris Manning (eds.), Architectures, Rules and Preferences: Variations on Themes by Joan W. Bresnan (CSLI Lecture Notes), 405–427. Stanford: CSLI Publications. Simpson, Jane and Joan Bresnan 1983 Control and obviation in Warlpiri. Natural Language and Linguistic Theory 1: 49–64. Simpson, Jane and Ilana Mushin 2008 Clause initial position in four Australian languages. In: Ilana Mushin and Brett Baker, Discourse and Grammar in Australian Languages (Studies in Language Companion Series 104), 25–57. Amsterdam: John Benjamins. Singer, Ruth 2001 The inclusory construction in Australian languages. Melbourne: University of Melbourne BA Honours thesis. Speas, Margaret 1990 Phrase structure in natural language. (Studies in Natural Language and Linguistic Theory 21) Dordrecht: Kluwer Academic Publishers. Stirling, Lesley 2008 “Double reference” in Kala Lagaw Ya narratives. In: Ilana Mushin and Brett Baker, Discourse and Grammar in Australian Languages (Studies in Language Companion Series 104), 167–202. Amsterdam: John Benjamins. Swartz, Stephen 1988 Pragmatic structure and word order in Warlpiri. In: Papers in Australian Linguistics no. 17 (Pacific Linguistics A-71), 151–166. Canberra: Australian National University. Swartz, Stephen 1991 Constraints on Zero Anaphora and word order in Warlpiri narrative text. (SILAAIB Occasional Papers 1) Darwin: Summer Institute of Linguistics. Thomsen, Ole Nedergaard 1994 Dyirbal ergativity and embedding. Studies in Language 18(2): 411–488. Tsunoda, Tasaku 2011 A Grammar of Warrongo. (Mouton Grammar Library 53) Berlin: de Gruyter Mouton. Van Egmond, Marie–Elaine 2012 Enindhilyakwa phonology, morphosyntax and genetic position. Sydney: University of Sydney PhD thesis. van Valin, Robert 1981 Grammatical relations in ergative languages. Studies in Language 5: 361–94. Verstraete, Jean-Christophe 2010 Animacy and information structure in the system of ergative marking in Umpithamu. Lingua 120(7): 1637–1651. Walsh, Michael. 1996. Body parts in Murrinh–Patha: incorporation, grammar and metaphor. In: Hilary Chappell and William McGregor (eds.), The Grammar of Inalienability: A Typological Perspective on Body Part Terms and the Part-Whole Relation, 327–380. Berlin: Mouton de Gruyter. Walsh, Michael 2005 Classifying the world in an Aboriginal language. In: Michael Walsh and Colin Yallop (eds.), Language and Culture in Aboriginal Australia, 107–122. Canberra: Aboriginal Studies Press. Wierzbicka, Anna 1981 Case marking and human nature. Australian Journal of Linguistics 1(1): 43–80. Yallop, Colin 1982 Australian Aboriginal languages. (Language Library) London: André Deutsch.

Claire Bowern

6 Complex Predicates in Australian languages 1 Preliminaries Complex predication is a good area to illustrate diversity in Australian languages, and where it has been overlooked in previous descriptions.¹ Although complex predicates have been recognised as a construction in many northern Australian languages for many years, it is only recently that they have been equated with complex predicates in Turkic and other languages elsewhere in the world (Bowern 2008a; Amberber, Baker, and Harvey 2010), and more recently still that researchers have begun to explore the cross-linguistic syntactic diversity of the construction within Australia. Work on such constructions has tended to focus on light verb constructions. Australian languages exhibit several other types of complex predicate, however; serial verb constructions, for example, are found in several parts of the country. Thirdly, there are a number of structures which are both monoclausal and “complex” in the sense that more than one element contributes information to the predicate which can be associated with a verbal head; most common are “associated motion” constructions (Koch 1984). However, work on this area has suffered, in part because work on these topics in other areas has been primarily driven by theoretical questions, while Australian language documentations have tended towards functional orientations which have underplayed the role of both comparative syntax and questions of lexicality.² I here provide an overview of complex verb constructions (CVCs) in Australian languages. I focus on complex predicates with light verbs, since they have been better researched in these languages than serial verb constructions and other complex predicate types. I begin with an overview of definitions, and discussion of previous language-based and broad theoretical treatments of the constructions. I then move in § 2 to a summary of the types of constructions found in Australian languages. In § 3 I give more details about the syntax of light verb complex predicates, serialisation, and associated motion constructions; §  4 provides some discussion of diachronic issues.

1 Work for this chapter was funded in part by NSF grant BCS-0902114 ‘Dynamics of Hunter-Gatherer Language Change’. Thanks to Harold Koch and Rachel Nordlinger for comments on this work. All errors are my own. 2 See Bowern (2005) for further discussion.

264 

 Claire Bowern

1.1 Definitions As is now well known from the literature (Butt 1995; Butt and Geuder 2001; Seiss 2009; Bowern 2008a), complex predicates are those in which the information normally associated with the head of a verbal predicate is spread over several parts of the predicate. That is, multiple phonological words jointly determine the argument structure and lexical-conceptual structure of the predicate. For example, in light verb structures, two words — the inflecting light verb and the uninflecting coverb³ – jointly provide information about the number of arguments in the predicate and their grammatical roles. In the Bardi example in (1), the semantic roles of the event participants are determined by the coverb wajim ‘wash’, while tense, aspect and participant person and number information are provided by the inflecting light verb -ma- ‘put’. The light verb in this example can also be used as an independent verb, as in (2), where it means ‘put’.⁴ (1)

Bardi wajim i-n-ma-n=irr wash 3-tr-‘put’-cont=3pl.obj coverb light verb ‘He/she washes them’.

(2)

i-n-ma-n=irr 3-tr-‘put’-cont=3pl.obj ‘He/she put them [somewhere]’.

The light verb -ma- ‘put’ is just one of 13 light verbs commonly used in Bardi to form complex predicates. There are an additional 26 inflecting verbs which may also appear in complex predicate constructions, but which do so only with one or two preverbs. The most common light verbs in Bardi are -joo- ~ -di- ‘do/say’, -ni- ‘sit’, -jiidi‘go’, -inya- ‘catch’, and -boo- ‘hit’. These light verbs are also very common cross-linguistically (both within Australia and elsewhere in the world). Example (1) shows a coverb which has been borrowed from English. Complex predicate constructions are employed when loan verbs are used in Bardi; this also appears to be typical in other languages. The word class of coverbs, however, is diverse, containing not only

3 Terms used for this item in Australia are most commonly “preverb” (e.g. Austin and Bresnan 1996; Schultze-Berndt 2003; Nash 1982), “uninflecting verb” (e.g. McGregor 2002) and “coverb” (Wilson 1999; Schultze-Berndt 2003). 4 The Leipzig glossing rules are followed. Additional abbreviations are asp, aspect marker; cont, continuous; intens, intensifier; perl, perlative; precon, pre-contemporaneous; ser, serialisation; ss, same subject; usit, usitative; verb, verbaliser.

Complex Predicates in Australian languages 

 265

items which function solely as coverbs, but also adjectives, adverbs, and some nouns. In Bardi, coverbs never appear without a light verb, though this is fairly unusual in Australia; in other languages, coverbs can be used without light verbs in contexts such as imperatives and non-finite clauses. Complex predicates in Bardi are a single syntactic word; that is, the coverb and light verb share a single argument structure, with information contributed from both components of the predicate. The light verb typically specifies the number of arguments, and may constrain semantic roles, while the coverb typically provides most of the semantic content of the predicate, such as information about the event. Temporal structure may be spread across both parts of the predicate; the light verb also provides information about the event type.⁵ An illustration of these principles is given in (3a), using the coverb maanka ‘black’ (which is also an adjective) and the light verb -ni- ‘sit’, which productively combines with adjectives to denote states. (3)

a. maanka i-ni-n. black 3-sit-cont ‘It’s black.’ b. -nilight- [State BEident (X)], maanka [Property BLACK]

a-structure:

(wstate (xj))

c. maanka -ni- [State BEIdent ([Wj], [ATIdent ([Property BLACK])])] Parts b and c of example (3) provide a schematic of how argument and event structure information is unified from coverb and light verb, using tools of conceptual semantics (Jackendoff 1992) and following Bowern (2004a), who adapted arguments from Wilson (1999). (3b) gives the lexical entries for the light verb and coverb. The light verb introduces a stative event; the argument structure for the verb contains a single argument entry. The coverb introduces a property concept (here ‘black’), but no further information. Coverbs may, however, have more or less complex entries depending on how much information they contribute to the predicate (here, it is assumed that maanka ‘black’ contributes only property information). Example (3c) provides the unified entry of the coverb and light verb, with the property information from maanka, and information about the state and its argument from the light verb. Complex predicates of this type are found in a swathe of Australian languages across the north of the country, as well as less commonly in other areas. McGregor

5 The details of preverb-light verb unification in Bardi are very complex; see Bowern (2004a) for discussion and illustration of the major patterns. The material presented here is a simplification. Australian languages appear to vary very considerably in this regard, however.

266 

 Claire Bowern

(2002) is a recent summary and more information on distribution is provided below. Although most of the literature on complex verb constructions in Australian languages has focused on complex predication with light verb constructions, there are, in fact, several predicate types in these languages. This is not surprising given the number of families, and the definition of complex predication is broad enough to admit a wide variety of predicate types, of which light verb constructions are simply the most common. An overview of types is given in § 2. Not all multiword verbal predicates fall under the definition of complex verb constructions, however. Auxiliary verb constructions, for example, are not complex predicates because the auxiliary plays no role in the determination of argument structure (Butt 1995; Seiss 2009). True auxiliaries are rare in Australia, though they are found, for example in Ngumpin-Yapa languages (Simpson 1991), Wambaya (Nordlinger 1998), Jingulu (Pensalfini 2003), and Diyari (Austin 1981: 88–91).⁶ In Ngumpin-Yapa, the auxiliary is a second position clitic which carries tense information and is a host for agreement clitics. The main verb may also carry limited tense inflection. In Diyari, auxiliary constructions are optional, and provide further distinctions about tense and mood functions. The auxiliary is homophonous with a stance verb, and the non-finite verb receives participial inflection. Some languages have switch-reference systems. Switch-reference systems are also probably not complex verb constructions because they are taken to be biclausal (Austin 1988). The distinction is perhaps arbitrary, since there are cases of switch-reference marked by verb serialisation, where several verbs appear under a single tense value and are marked for whether their subjects are the same or different. Indeed, in Austin (1988), papers are divided between complex sentences constructions which are clearly multi-clausal and those which are fairly clearly monoclausal but which nonetheless show complexity in predicate formation. Finally, under some theories of syntax, all verbal predicates contain multiple heads, and can therefore be analysed as “complex” (see further Hale and Keyser 2002; Folli, Harley, and Karimi 2005). Under this view, the difference between simple and complex predicates lies in whether a given predicate is lexicalised as a single word or as two words, rather than in underlying structure. I concentrate here on a more restrictive definition of complex predication, where there is a lexicon-syntax

6 As will be seen below, however, discussion of Australian complex predicates is complicated by the fact that many authors have not made a systematic distinction between auxiliaries and light verbs. Clendon (2000a), for example, calls light verbs “auxiliaries” and coverbs “infinitives”. While this nomenclature captures the fact that the structures are bipartite, and that one is inflecting and carries agreement and tense/mood information, it misses several crucial distinctions in argument structure and in the relationship between simple and complex predicates. For example, infinitives are nonfinite forms of main verbs, which in other constructions do receive inflection; this is not the case for coverbs.

Complex Predicates in Australian languages 

 267

mismatch,⁷ since those constructions are most likely to be of interest to a wider audience, and are most testable with the data at hand.

1.2 Data Data for this article are taken from published descriptions such as reference grammars, theoretical analyses, and my field notes for Bardi (see also Bowern 2012). The languages which feature in the survey in particular are given in Figure 1 below. A list of languages and sources is given in the Appendix to this paper. Some other languages are either known to have had complex verb constructions or are likely to have had them, but syntactic information was not sufficient to allow their inclusion. For example, Noongar (Nyungar) (Douglas 1976: 42) appears to have had at least some classificatory light verb constructions, but too little is known of the syntax of the language to permit full inclusion. These languages do not appear on this map. In addition to the general syntactic questions discussed in § 3, six languages were coded for etymology of 250 verbal meanings (whether simple or complex).⁸ The verbal meanings are based on the verbs in the Intercontinental Dictionary Series (also used in Haspelmath and Tadmor 2009) and were coded for whether the meaning was expressed by a complex or simple predicate, if complex, what the light verb was, and the etymology of the predicate (whether inherited, a loan, an inheritance which had undergone semantic shift, or a unique item in the language without cognates elsewhere).

7 As Rachel Nordlinger (pers. comm) points out, a definition of complex predicates which relies on a lexicon-syntax mismatch (or on differences in phonological wordhood and syntactic wordhood) is problematic for languages such as Alawa and those of the Daly River region. These languages have bipartite verbs which show structures which have many things in common with multi-word complex predicates; one part of the verb carries most of the semantic information of the predicate, while another part classifies event structure and otherwise behaves like a light verb. In some cases, there are clear etymological connections between the bipartite complex verbs and light verb constructions. In one case (Reid 2003) we have evidence of the reanalysis from a multi-word complex predicate to a single phonological word within a few generations. Thus while I mostly refer to multi-word complex verb constructions in this article, this should not imply that questions such as light verb choice or coverb-light verb argument structure unification are not also relevant to languages with bipartite stem structures. 8 Thanks to Patrick McConvell for coding the Gurindji and Mudburra data; this dataset was also used in McConvell and Bowern (2011).

268 

 Claire Bowern

Figure 1: Map of languages included in survey of light verb constructions

2 Previous work on complex predicates 2.1 Types of complex predicate constructions in Australian languages Within the descriptive/documentary tradition of work on Australian languages, the approach to complex predicates has been piecemeal, and the term “complex predicate” has not been universally accepted. Nash (1982) makes a connection between constructions described as complex predicates outside Australia and Warlpiri bipartite verb constructions, as does Silverstein (1986) with respect to Worrorra, but it is only with Wilson’s (1999) work on Wagiman that formal similarities are discussed in detail and a shared approach in analysis is employed. McGregor (2002: 259–61), in

Complex Predicates in Australian languages 

 269

contrast, draws a distinction between complex predication and classifier constructions where the coverb and inflecting verb each contribute information. Australian languages exhibit several types of construction which fall under the rubric of “complex predication”. Most notable for theoretical work has been the complex verb constructions (see McGregor 2002) involving an uninflecting coverb and inflecting light verb. Such constructions are found in a swathe of languages across the north of the country. There are, however, several other constructions too. Serial verb constructions are found in a few languages, such as in the Yolngu (Yolŋu) languages of north-east Arnhem Land. There are also other structures in which the verbal predicate is not the only item which makes a contribution to the semantics of the predicate. Yolngu languages from Arnhem Land have body-part “compounds” (where “compound” here is a pretheoretical term). In such examples, the prefixed body part can provide aspectual information about the predicate as a whole. In other cases, as given in (4) below, the compounds are semantically opaque. They are thus different from the complex predicate constructions, where both coverb and light verb provide information about argument structure. (Wilkinson 1991 provides further information about this construction.) (4)

(Yan-nhaŋu) (Dhalwaŋu)

mel-gäma mel-djulŋithin

‘forget’ (lit. ‘eye-carry’) ‘be proud of’

There are also incorporation structures, where a noun is combined with a verb stem which results in a change to the argument structure, valency, or semantics of the predicate. The Arnhem Land language Bininj Gun-wok (Evans 2003) provides examples. (5)

A-bikbik-bakme-ng. 1-rib-crack(intr)-pst.ptcp ‘I cracked my ribs.’ (Evans 2003: 454)

Restructuring predicates (Alsina 1997) are also treated by some as complex predicates. In several Romance languages, for example, causative predicates with verbs meaning ‘do’ or ‘make’ are biclausal under some syntactic tests, but monoclausal in others (see Cardinaletti and Shlonsky 2004 for discussion). These constructions have not been reported in Australian languages to my knowledge. Another interesting type of predicate which is fairly common in Australia (across multiple families and Pama-Nyungan subgroups) but rare outside the country is the “associated motion” construction. The term is due to Koch (1984) and the construction was first described in the Arandic (Pama-Nyungan) language Kaytetye, but has subsequently been recognised in other areas as well (cf. Wambaya in Nordlinger 2001). Such constructions do not easily fall under either serialisation or light verb complex predication, though they may be expressed by both; they are more semantically specialised than verb serialisation tends to be, for example. Furthermore, as will

270 

 Claire Bowern

be seen, motion verbs in light verb constructions tend to lose their motion meaning, while in associated motion constructions, the motion verb provides only motion information. They form verb compounds in some languages, while in others the motion marking is fairly clearly adverbial.⁹ Examples are given in (6) from Ngan’gityemerri, from Adnyamathanha in (7), and Kaytetye in (8). (6)

nga-ganim-fifi 1sg.subj-go.prs-smoke ‘I’m going along smoking.’ (Baker and Harvey 2010: 40)

(7)

artu-nga veldha marli-wandha-anggu woman-erg clothes wash-and.leave-prf ‘The woman washed the clothes and cleared off.’ (Tunbridge 1988: 273)

(8)

arntwe nte eyle-yene-ne water 2sg.erg get-go.and-imp ‘You go and get water.’ (Koch 1984: 27)

Australian languages are thus an important testing ground for theoretical approaches to complex predication. They are also a good area to test diversity in complex predicate behaviour, since there are languages from unrelated families which show the relevant constructions. Finally, complex predicates are also a good opportunity to consider potential unique or exceptional features that occur in Australian languages, but not elsewhere. Comparatively little work has been done on the other types of complex predicates in Australia, and they are probably less numerous (or at least less salient to grammar writers).

2.2 CVCs and their identification Many grammars of Australian languages identify a construction where the inflecting verb in the clause does not make the sole contribution of identifying arguments. A typical example comes from Capell and Hinch’s (1970) grammar of Mawng, where the uniqueness of the construction is recognised, but the word classes are subsumed under the labels of “root” (= coverb) and “auxiliary” (light verb).

9 The line between verb compounding and complex predication is not entirely straightforward either; on the one hand, verb compounds do not provide problems of lexicality, since compounds are single phonological words formed in the lexicon, and are therefore assumed to have a single Lexical Conceptual Structure. On the other hand, if the compounding is productive and compositional, there may be some evidence that the motion root adds argument structure to the predicate.

Complex Predicates in Australian languages 

(9)

 271

juɹan alja he.go.pst forget auxiliary root ‘He forgot.’ (Capell and Hinch 1970: 69)

Other terms which have been used to describe complex verbs are “adverbs” (e.g. Wardaman: Merlan 1994), “participles” (Cook 1988) and “verbs”.

2.3 Theoretical analyses There is research on CVCs in Australian languages using government and binding theory, Lexical-Functional Grammar (LFG), and Construction Grammar, as well as functional grammar work which makes theoretically explicit and testable statements about language structure. Unfortunately, with the exception of Warlpiri, there is little overlap in the languages (or even language families) in which the theoretical work applies. Minimalist work, for example, is confined to the Ngumpin-Yapa language Warlpiri (e.g. Laughren 2010); LFG work is the most widespread, in languages such as Wagiman (S. Wilson 1999; A. Wilson 2009), Wambaya (Nordlinger 2010) and Bardi (Bowern 2004a). Work in Construction Grammar is mostly confined to Jaminjung (e.g. Schultze-Berndt 2000). A substantial difficulty with previous theoretical work on complex verb constructions is the lack of consistent treatment of theoretical frameworks across languages; that is, there is work on different languages in different theories, and while authors may criticise the work of others on theoretical grounds, it is not at all clear that the basis for the difference is simply that the languages have different syntax. We lack much work which involves systematic cross-linguistic comparisons within the same theory, or cross-theoretical analyses of a single language.

2.3.1 Event Structure No matter what the theoretical orientation of the work, five issues are particularly of concern for complex predicate analysis. The first is “event structure” and the division of labour between the coverb and the light verb. That is, how is meaning constructed when there is more than one predicational element? The most explicit statement of this, to my knowledge, occurs in the LFG work of Wilson (1999). Baker and Harvey (2010) adopt some aspects of Wilson’s analysis, particularly the unification of conceptual structure. Building on theoretical apparatus from conceptual semantics (Jackendoff 1992), they propose specific contributions from the light verb and coverb, and specific mechanisms for their combination, from which follow the types of light verb meaning which tend to be grammaticalised, and the types of verbs which are com-

272 

 Claire Bowern

monly used as light verbs. They argue that in one type of predicate, compatible information from the two parts of the predicate merges into a single, complex, predicate.¹⁰ In the other structure, no merger takes place, and information is instead coindexed. Note that Baker and Harvey do not discuss argument transfer, which was the type of complex predicate argument licensing first identified in Grimshaw and Mester (1988). Such an analysis is probably too restrictive to cover all types of light verb construction across Australia, however. Bowern (2004a: Ch 13) shows for Bardi, for example, that there are at least four types of complex predicates in the language, depending on the relationship between the coverb and the light verb; similar comments appear to apply to Yawuru (Hosokawa 1987, 1991). Nash (1982) shows for Warlpiri that relationships between coverb and light verb are not straightforward and that a number of syntactic relations obtain between coverb and light verb. He identifies several types of boundary between coverb and light verb, several argument structure relations, and categories of coverbs that interact with each other. Baker and Harvey’s theory, as currently stated, allows for only a single type of coverb-light verb interaction (where the two items fuse arguments).

2.3.2 Classification This brings us to the second area of debate, the role of classification in complex predicates. To what extent does the light verb “classify” the coverb, or provide further information about its event type? McGregor (2002) suggests that complex predication and verb classification are to be treated separately; see above for comments on this approach. Under the LFG view (of e.g. Wilson 1999), classification “comes for free”, in that the addition of a predicate type can be interpreted as classifying the event into particular types (MOTION events, inchoative events, etc). However, the classification structures vary considerably from language to language, and do not only involve classification by event variables. Nyulnyul’s are given in McGregor (2002) and are reproduced in (10) below. (10) n- ‘sit’ vs. j- ‘do’: static vs. dynamic events jid- ‘go’ vs. kal- ‘wander’: directed vs. undirected motion m- ‘put’ vs. w- ‘give’: action extending out vs. action drawing in k- ‘carry’ vs. m- ‘put’: source of energy moves with trajectory, versus not Languages as close as Nyulnyul and Bardi, which contain a great deal of common vocabulary and are probably about as close as Dutch and German, nonetheless have

10 Note that Baker and Harvey are by no means the first to suggest that complex predicate formation involves merger of conceptual structure; see, for example, Wilson’s (1999) analysis of Wagiman.

Complex Predicates in Australian languages 

 273

rather different light verb classificatory structures. The classification system given in (10) for Nyulnyul does not apply to Bardi. A few light verbs occur in pairs; transitive complex predicates with the light verb -inya- ‘catch’, for example, usually form anti-causatives with -jiidi- ‘go’. But such patterns are in the minority. Although light verb inventories tend to be quite similar across languages (not only within Australia but elsewhere), the organisational principles for light verb classification appear to be rather unstable.

2.3.3 Light verb inventory and stability Related to this is the light verb inventory, its stability from language to language, and the functions of particular light verbs in particular languages. As has long been noted, complex predicates with light verbs tend to use the same (or similar) set of verbs across languages from all over the world, typically ‘do’, ‘say’, ‘make’, ‘bring’, ‘take’, ‘go’, ‘spear’, ‘share’, and others.¹¹ No one has tackled why it is that these verbs tend to be grammaticalised as light verbs, though the answer is presumably the obvious one – that these verbs tend to have general semantics, to be frequently subject to metaphorical extension, and have high token frequency, so are thus both suitable for reanalysis and easily extensible to novel meanings. The suitability for reanalysis and the general semantics of the light verbs no doubt provides an explanation for why such systems may have rather different principles of organisation, even in the face of quite similar inventories. The argument structure relations of coverb and light verb are also of interest. In most theories of syntax, a single element is responsible for specifying the number of arguments in a clause and their grammatical roles. This has been a principle of all formal theories since the 1960s. Complex predicates in some languages clearly violate that assumption, however. In some languages, the coverb provides grammatical relations and semantic roles, while the light verb gives the number of arguments which are licensed in the clause. In others, argument structure information from the light verb appears to be “overwritten” by the coverb; see Bowern (2008b) for some discussion of the phenomenon in Bardi, and Schultze-Berndt (2000) for examples from Jaminjung. Finally, there are cases where the coverb and light verb jointly determine both argument structure and grammatical relations.

11 As noted below, however, light verb inventories vary substantially across Australia, from 1 to 45 verbs.

274 

 Claire Bowern

2.3.4 Nexus The fourth area of research is the issue of “nexus” between coverb and light verb. This covers not only wordhood concerns, such as whether the coverb and light verb form distinct phonological or syntactic words, but also constituency: what type of constituent is the complex predicate, how inflection is distributed across the predicate, and how constituency functions in comparison to other constituents in the language. It is striking, for example, how many languages with otherwise “free” word order at other levels of constituency have more restrictions on complex predicate syntax. While free constituent order at the clause level is not universal in Australia, it is very common. Coverb ordering restrictions, however, are much more frequent. While for the majority of the languages under consideration coverb and light verb are independent phonological words (for example, with distinct primary stresses), in a few languages, both preverb and light verb form a single phonological word. Alawa (Sharpe 1972) shows this behaviour, for example, as do several of the languages of the Daly River region. Authors have generally seen languages such as Alawa as being at one end of a continuum in which phonological wordhood may correspond more or less precisely with syntactic wordhood.

2.3.5 Coverb etymology The final area to be considered is the word status of coverbs and their etymology. Some grammars treat coverbs purely as a special type of adverb (e.g. Merlan 1994), while others treat them as a type of verb (Clendon 2000a); a third treatment classes coverbs as their own word class (Schultze-Berndt 2000; Bowern 2004a; Wilson 1999 and others). In some languages, coverbs are clearly a distinct part of speech, with no overlap with other word classes; in other languages, however, coverbs may be derived, either from nouns, adverbs, adjectives, or verbs.

3 Typology Here I provide an overview of the syntactic typology of complex predicates in Australian languages. For reasons of space and because they have been better studied than other types of complex predicate, I focus on light verb constructions. However some comments are made on serialisation constructions in § 3.2.

Complex Predicates in Australian languages 

 275

3.1 Light Verb Constructions Light verbs are predominantly found in north Australia (see the map in Figure  1 above), particularly in the Ngumpin-Yapa subgroup of Pama-Nyungan, and adjacent non-Pama-Nyungan languages. They are, however, also found sporadically in other areas. In Muruwari (Pama-Nyungan; New South Wales), Oates (1988) describes a construction where a number of different nouns may combine with the verb roots pa- ‘act, do, perform’ or yi- ‘be, have’, as in (11). (11) ngathu kunturl pa-rri-yu thuu 1sg.nom hunger do-refl-1sg much ‘I am very hungry.’ (Oates 1988: 193, ex 5.454) Muruwari’s geographical neighbour, Ngiyampaa (Donaldson 1980), has a similar construction, though it is much more productive and involves a wider array of verb roots. The other Pama-Nyungan subgroup outside of the west where CVC-like structures light verbs have been reported is Paman. Patz (2002: 100–101), for example, describes lexical N+V compounds in Kuku Yalanji, in which one of a restricted set of verbs (which often appear as light verbs in other languages) combine with a noun or adjective, for example walu-dungay ‘faint’ (lit. ‘face-go’).¹² Clearer examples come from Yir Yoront. Alpher (1991) also describes constructions in Yir Yoront where English verbs are combined with the Yir Yoront verb tha ‘do’, as in (12) below: (12) ngoyo five-o’clock knock-off=tharrarr 1sg 5PM knock-off=do.go:cont.npst ‘I knock off at 5pm.’ (Alpher 1991: 52) Yir Yoront also has a set of ideophones which may combine with a small set of verbs (the same verbs which are used as light verbs across the sample). The ideophones are usually closely bound to the following verb. While ideophone constructions resemble CVCs in some fashion, the two are not identical. For example, ideophone + verb constructions are overwhelmingly (though not exclusively) associated with verbs in past tense (Alpher 2001), and cannot occur in subordinate clauses. CVCs, conversely, always have the same distribution as simple verbs. Thus while Yir Yoront ideophonic constructions have some of the properties of complex verb constructions, they lack the generality of true complex predicates.

12 It is not clear from Patz’s materials whether these constructions satisfy all definitions of CVCs; they may be more similar to the N+V body part compounds described for Yolngu languages in example (4) above.

276 

 Claire Bowern

3.1.1 Number of light verbs Languages in the region differ extensively as to how many light verbs participate in the complex predicate system. The numbers are summarised in Figure 2. The smallest is Young People’s Tiwi, with a single light verb ‘do’; Muruwari has just two. The largest system is Wagiman’s, with around 45 light verbs.

Figure 2: Number of light verbs

It is not known how this distribution of light verbs compares to languages in other parts of the world.¹³ As seen here, however, there is considerable variation in light verb counts across Australia. As might be expected, the difference in number of light verbs in complex predicates also translates into a difference in the basis of the classification systems. Some languages have no classificatory light verbs at all (e.g. Tiwi and Ngiyampaa); others have only classificatory light verbs (Gooniyandi), while a third set of languages have both a classificatory system and other complex predicates which do not fit under the rubric of “classification”. Most of the classificatory systems contain between eight and 14 light verbs. This is discussed further below.

3.1.2 Functional load of light verbs Most Australian languages with complex predicates also show simple predicates. The number of simple predicates in the system varies. Numbers of simple predicates vary

13 Bowern (2004b) shows that Turkic languages vary in terms of the number of light verbs in the system; however, the numbers in that family tend to be smaller than the upper limits found here.

Complex Predicates in Australian languages 

 277

from 0 (all verbal predicates in the language are complex; see Gooniyandi (McGregor 1990), for example) to more than 500+, where simple (non-CVC) predicates form an open word class. The number of light verbs in a language is not correlated with the number of simple predicates in a language (r = -0.12)

Figure 3: Number of simple predicates

Most of the languages in the survey have restricted numbers of simple predicates. That is, while there tend to be more simple predicates than light verbs, complex predicate constructions play an important functional role in the verb system. Warndarrang (Heath 1980) and Muruwari (Oates 1988) have open classes of simple verbs in addition to complex predicates, while languages of the northern and far western Kimberley (from the Worrorran and Nyulnyulan families) tend to have large but closed classes of simple predicates in addition to their complex predicate constructions (Clendon 2000a; Bowern 2008c; Rumsey 1984). An exception to the relative numbers of simple and complex predicates comes from the Daly River, where languages such as Murrinh-Patha allow only a subset of light verbs to be used in simplex constructions (R. Nordlinger, pers. comm.). To get a better idea of the extent to which complex predicates bear a significant functional load in the verbal system, six languages were coded for the percentage of meanings on a 248-item verb list which were expressed by simple predicates, versus those expressed by complex predicates. Those figures are given in Figure  4. (The y axis represents percentages; percentages for all languages do not add to 1 because missing data were excluded.)

278 

 Claire Bowern

Figure 4: Proportion of meanings expressed by simple and complex predicates

As seen here, the proportion of the verbal lexicon expressed by complex predicates varies considerably over these languages, from just about half in Bardi to nearly 90% in Gurindji.

3.1.3 Word status of complex predicates The clear majority of languages in the sample had complex predicates which were phrasal rather than a single phonological word; that is, the coverb and the light verb

Figure 5: Phrasehood of complex verb constructions

Complex Predicates in Australian languages 

 279

had their own primary stresses, and satisfied other tests for wordhood. In five languages, the coverb and light verb comprised a single phonological word, where coverb and light verb were adjacent (“continuous” is the terminology used by SchultzeBerndt (2000), whose categories are used here). In another five, the coverb and light verb form a single phonological word, but agreement or tense marking intervenes. Finally, a few languages have “variable” marking, where the coverb can appear both as a single word with the light verb (and other morphology), or separated from it. The wordhood status of complex predicates is thus just one facet of coverb-light verb “nexus”.

3.1.4 Inflection of coverbs Typical definitions of coverbs imply that they do not typically inflect (in contrast to light verbs, which carry person, number, tense, and other inflection). However, about half the languages in the sample allow some kind of inflection on coverbs.¹⁴ Figure 6 provides a summary of the categories of inflection found in the dataset.

Figure 6: Inflection on coverbs

14 The inflection discussed here does not include items where the coverb acts as a host for clitics. For example, both Bardi and (to some extent) Warlpiri allow coverbs to act as hosts for Wackernagel (second position) clitics when the coverb is the first word in a clause. This does not count as specifically coverbal inflection, however, because a word from any word class can be a host for such inflection. This section discusses morphology specific to coverbs. A reviewer raised the question of whether the morphological marking discussed here is inflectional or derivational. In the absence of information to the contrary, I assume that tense, aspect, and number marking is inflectional in these languages.

280 

 Claire Bowern

As can be seen from Figure 6, the most common type of coverb inflection is aspect marking. 16 languages allow marking of aspectual categories on the coverb. An example is given from Yawuru (Hosokawa 1991). (13) kardu-garriny wubardu wangkurr-gadya + i-nga-rn (??)¹⁵-yet little(abs)i cry-intens+3i-aux(be)-ipfv ‘The child is still crying.’ (Hosokawa 1991: § 8.2.1.2, ex. 32) Other inflectional categories are considerably less common. Tense, for example, is marked in a single language (Ngiyampaa). Two languages have person marking: Worrorra and Wagiman. The strategies for marking are different in each language, however. In the case of Worrorran, coverbs are often referential nouns which may take possessive prefixes referring to the subject of the verb, as in example (14) below. The light verb, however, is still present and takes its usual array of agreement and tense/ aspect marking. (14) Doreen

arrerndu kajaanya arri-kwarndu D 1dl.excl mother’s.mother ‘Doreen’s and my grandmother died.’

nyiman nyimnyanu nyiN=man nyiN-mnya=nu 3f=dead 3f-prox=be (Clendon 2000a: 359, ex. 9.2a)

In Wagiman, the subject person markers may appear attached to a coverb if no light verb is present, as shown in example (15). This appears to be productive in Wagiman. The coverb is given in bold. (15) a. ga-yu guk-ga gahan 3sg.be.prs sleep-asp that ‘That baby is sleeping.’

labingan baby

b. warren gahan ga-guk-ga-n child that 3sg-sleep-verb-prs ‘That kid is sleeping.’ (Wilson 1999: 82) The behaviour of Wagiman coverb inflection is intriguing, since it argues against an analysis of complex predicates in which the light verb is required to license argument morphology which is unavailable for hosting on the coverb (Grimshaw and Mester 1988). The fact that simplex, coverb-only predicates apparently alternate with CVCs in this language requires further investigation. Wilson (1999: 82–83) notes that many of

Except where noted, the marking is productive and interacts with clearly inflectional categories on the light verb. However, the point is not crucial here and “inflection” could be taken here to mean simply “morphological marking”. 15 The gloss is as in Hosokawa (1991); gard in Bardi means ‘still’.

Complex Predicates in Australian languages 

 281

the inflecting verbs in Wardaman (Merlan 1994) are historically derived from coverbs, so it appears that this type of alternation was not confined to Wagiman. While person marking is very rare on coverbs, 10 languages in the survey do mark participant number. This is mostly achieved through the reduplication of coverbs, though some languages (e.g. Miriwoong, see Kofod (1976; 1978)) have suppletive coverbs which vary according to the number of the subject, and Eastern Nyulnyulan languages (Yawuru and Nyikina) and Wagiman have subject number affixes. In Wagiman, the dual nominal marker –giwu may follow coverb aspect marking, though this is rare (Wilson 1999: 57). The descriptions are unclear regarding whether reduplication should be properly regarded as grammatical number marking, or whether it should rather be taken as a type of pluractional aspect marking (Wood 2007), which by the nature of pluractionality tends to have a plural number of participants. Finally, coverbs can also be inflected for case and nominalisation derivational marking. A few languages allow case to be added directly to coverbs,¹⁶ while most of the other languages require the coverb to be first affixed with nominal derivational morphology. In summary, coverb affixation provides further evidence for the diversity of syntactic constructions in complex predicates across northern Australia.

3.1.5 Light verb omission In some languages, light verbs are obligatory in all contexts: Bardi (Bowern 2004a, 2012), Ungarinyin (Ngarinyin) (Rumsey 1984), Marrithiyel (Green 1989), Muruwari (Oates 1988). In Bardi and Ungarinyin, speakers categorically rejected example sentences where the light verb was not present; contexts which in other languages allow light verb omission (in imperatives, for example), always showed light verbs, and (at least in Bardi) speakers were never observed to drop light verbs in casual speech.¹⁷ In Bardi, non-finite contexts (another area where other languages can show light verb omission) are expressed with coverbs and light verbs with gerundial marking, as in (16) below. Omission of the light verb here is ungrammatical. (16) arra=gid oo-la-laba-n baanigarr joornkoo-ngan neg=thus 3-irr-hold-cont now run-purp ‘He stopped her from running away.’ (NI: WIR.012)

(*)ma-nya-n inf-‘catch’-cont

16 A few languages also allow phrasal coverbs which are derived from nouns marked for case. Bardi, for example, has some coverbs which contain locative markers; cf. wiliwilon -kal- ‘go fishing’ (wiliwili ‘fishing line’; -on vowel-final morpheme of locative’; -kal- ‘wander’). 17 No information is available for Ungarinyin on this point.

282 

 Claire Bowern

Bardi appears to be unusual in the Nyulnyulan family in exhibiting this behaviour, since other languages in the family do appear to allow light verb omission in certain contexts.¹⁸ Likewise, Ungarinyin appears to stand out within Worrorran, and other Worrorran languages do not seem to have the same prohibition on light verb omission. Worrorra, for example, allows light verb omission in narratives (though not in imperatives). Note, however, that many Worrorran languages are poorly recorded. It is possible that the Bunuban language Bunuba (Rumsey 2000) may also have obligatory light verbs in everyday speech, though there is some doubt. It also seems that the mother-in-law respect register behaves differently with respect to complex predicates in this language.¹⁹ Twenty-one of the languages in the sample allow light verb omission in at least some contexts, such as imperatives, narrative (verb chaining) or elsewhere. Imperative constructions and narrative chaining were the two most common contexts in which light verb omission has been reported. Other reported contexts involved subordination and in the use of nominalised coverbs as arguments. Twelve of the languages allow multiple contexts, while four allow imperatives but not narrative chaining, and one (Jaru: Tsunoda 1981) allows light verb omission in narrative chaining but not in imperatives. Thus in summary, there is substantial evidence for complex predicate diversity across the country, which should be further investigated.

3.2 Verb serialisation Languages with serialisation constructions are less frequently described in the literature on Australian languages than complex verb constructions. This may not be because serialisation is actually poorly attested, however. A tendency to view Australian languages as having few clausal dependencies, along with problems in data gathering, has likely led to undersampling in this area. There are also differences in how researchers have viewed similar constructions. For example, I report Bardi as having narrative serialisation on the basis of evidence from intonation and argument sharing (see Bowern 2012: 663), but such constructions are not reported for next-door Nyulnyul, nor in previous work on Bardi, which appears to assume clause chaining.

18 Information is clouded, however, by the fact that some coverbs in these languages also appear as adverbs, and detailed syntactic tests are needed to disambiguate the behaviour of the two categories. It is thus possible that many of the environments which have been identified as light verb omission in other Nyulnyulan languages are in fact simply adverbial uses of coverbs. 19 Another language where light verbs may be obligatory is Kuku Yalanji, but difficulties in defining the construction make this hard to state with certainty. Muruwari is coded as having obligatory light verbs because the complex predicates are formed from Noun + Verb pairs; without the light verb, the sentence would presumably simply have the noun reading.

Complex Predicates in Australian languages 

 283

This section provides an overview of common constructions, but should not be taken to represent a systematic survey. Serial constructions in the literature fall into four main types. The first is where a motion or stance verb provides aspectual information about the clause. These have been described in Mantharta languages (Austin 1998) and in the Daly region, and also appear in the Yolngu languages of Northeast Arnhem Land. Examples are given for Jiwarli in (17) and Yan-nhaŋu in (18). In Yan-nhaŋu the combination of ‘stand’ and ‘go’ is used to describe items which are spread out along a line. Morphy (1983) reports that stance verbs in Djapu (another Yolngu variety) are used in serialisation constructions, but suggests that they mark aspect rather than spatial extent, as is found in Yan-nhaŋu.²⁰ (17) ngatha kumpa-artu tharla-rnu papa-jaka 1sg.nom sit-usit feed-ipfv:ss water-com ‘I used to feed (him) with water.’ (Austin 1998: 24) (18) bambitj mana bamparra garama dhukarr-murru tree cont stand go road-perl ‘Trees line the road.’ (Bowern, fieldnotes) Grammaticalised stance verbs in serial constructions are also found in certain Western Desert (Wati) varieties (Goddard 1988). In Yankunytjatjara, stance verbs are only one of several serialisation constructions, in which one verb appears with a (non-finite) ‘serial’ suffix and the other is fully inflected. As seen in (17), the Jiwarli construction involves the verb kumpa ‘sit’, which takes tense marking, and a second verb, which is marked for switch-reference categories (same-subject or different-subject). Diyari (Austin 1981) also has stance verbs marking tense, but the construction appears to involve auxiliation rather than serialisation. The Yan-nhaŋu construction, however, has two finite verbs, and contrasts semantically with a sentence in which one verb is morphologically dependent on the other. Serialisation is also found in languages which use a second inflecting verb to provide information about the direction or manner of motion. Languages with serialisation of this type include Dyirbal (Dixon 1972), Ngiyampaa (Donaldson 1980), and Wambaya (Nordlinger 2014), from which example (19) is taken.

20 As seen in the following examples, “serialisation” constructions vary extensively across PamaNyungan. While I have followed the authors of grammars in what have been labeled serial verb constructions, some authors do not provide explicit grammatical criteria for serialisation (as opposed to embedded or biclausal constructions).

284 

 Claire Bowern

(19) warima ng-a marrajana yardugami hold 1sg.a-pst pillow.iv.acc make.strong ‘I held the pillow tightly.’ (Nordlinger 2014: 269) In Dyirbal, the verbs in a serialisation construction must exhibit the same tense morphology and must agree in transitivity. This is not true of some other languages with this construction; for example, the Ngiyampaa sentence in (20) has verbs which agree in tense but not transitivity. (20) winaru mingga gunumi-yi bagi-yi woman-erg burrow do.with.energy dig ‘The woman dug up the burrow with energy.’ (Donaldson 1980: 204) A few languages use serial verb constructions to mark causation. Burarra (Green 1987), for example, has a construction where the first (transitive) verb has an agent which causes the patient of that verb to enact the action of the second verb, as shown in (21) below. Burarra also uses serialisation to mark aspect (in a manner similar to the Yolngu languages to its east, as illustrated in (18) above). (21) ngu-jerrmu-nga gochilawa a-bo-na 1sg>3sg-send-precon beach 3Man-go-precon. ‘I made him go to the beach.’ (Green 1987: 82) The fourth type of serialisation in Australia is narrative serialisation. This occurs in languages such as Mparntwe Arrernte (Wilkins 1988), Wardaman (Croft 2007) and Bardi (Bowern 2012). It probably occurs more widely than has been reported in the literature, but this cannot be further investigated because of the difficulty in determining the difference between monoclausal serialisation constructions and juxtaposed clauses without explicit syntactic tests. Narrative serialisation involves a single actor, who performs a number of events in a series. The actor occurs only once in the clause. Example (22) is from Mparntwe Arrernte and is illustrative; the actor, Kwementyaye, occurs once; all verbs except the last receive a “seriation” affix -mele. (22) Kwementyaye-le ure ine-mele, ure nthile-mele, tea ite-ke. K-erg firewood set-ser fire light-ser tea cook-pc ‘Kwementyaye got the firewood, lit a fire and made the tea.’ (Wilkins 1988: 162) An equivalent construction is found in Bardi, except in Bardi, all verbs are finite. Bardi shares other features of serialisation with Mparntwe Arrernte, however. There is a single actor, who is mentioned only once and is subsequently tracked by agreement marking (see (23) below). All verbs appear under a single intonation contour;

Complex Predicates in Australian languages 

 285

finally, arguments from a lower verb may be fronted to initial position; fronting is only possible in Bardi within a clause. (23) Ginyinggon then

roowil walk

i-n-nya-na Ngarrigoonbooroo 3-tr-catch-rem.pst Ng.

darr i-n-ar-na=jirri niimana aamba come 3-tr-spear-rem.pst=3pl.obl manymen and

baali-ngan shade-all

agal ambooriny people

Ngoolbirnd-i. Ng.-loc ‘Then Ngarrigoonbooroo walked to her camp and came across many people at Ngoolbirndi.’ (Laves 129, ln 19) There is considerably more to say about serial verb constructions in these languages. While most of the constructions described here involve two (or more) verbs with finite inflection, there appear to be functionally equivalent constructions which contain a single tense/aspect-marked verb, while the other verb is marked as ‘serialised’ or otherwise unspecified for tense. In Wambaya (Nordlinger 2014), either marking choice is acceptable. Wambaya also raises the question of the range of function of serialisation constructions available in a language; Wambaya shows the complete range of serial verb functions reported in Foley and Olson (1985); this appears to be rather unusual for Australian languages.

4 Historical Considerations In this final section, I make some observations on the place of complex predicates in historical linguistics.

4.1 Reconstructibility of CVCs CVCs are difficult to reconstruct as full predicates, even among close-knit languages. In the Nyulnyulan family, for example, the CVC domain is not very conservative, even though the languages share many cognate words and pieces of CVCs are readily identifiable across the family.²¹ All eight well attested languages in the family have complex predicates. Comparative work shows that almost all the verbs used as light verbs in the languages are reconstructible to Proto-Nyulnyulan. The only verb that isn’t is

21 This section reports results from Bowern (2008c).

286 

 Claire Bowern

the verb ‘go’, which is different in Eastern and Western Nyulnyulan and reconstructible to those intermediate proto-languages only. Many preverbs are also reconstructible to Proto-Nyulnyulan, though it is here that problems arise. In Nyulnyulan languages, the category “preverb” is both a well defined word class and a structural position in the verb phrase; that is, while there are many words which exist only as preverbs in the languages, words from other word classes may also be used as preverbs. The syntax of such items varies from language to language, though in all Nyulnyulan languages (to my knowledge) a reliable syntactic test for preverb status is whether the item may appear with a negated verb. If the preverb appears between the negation marker and the light verb, and if the verb takes irrealis marking,²² that is good evidence that the item in question is structurally a preverb and not, for example, an adverb. Languages in the family differ, however, as to whether any given item is a derived preverb or an underived one. For example, Bardi has preverbs which exist only as preverbs in the language, but which nonetheless have cognates in one or more other Nyulnyulan languages as nouns or adverbs. In modern Bardi, the derivation of coverbs is only productive with adjectives; other word classes show only lexically specified coverb derivation. In the absence of evidence as to directionality of coverb derivation, if a preverb is only found in a few languages, and is a preverb in one language but a member of another part of speech in another, we cannot reconstruct the word class of the item in Proto-Nyulnyulan. Secondly, even when an item can be reconstructed as a preverb, the preverb-light verb pair often cannot be reconstructed with any certainty. Even when the preverblight verb combinations are attested in languages in both sides of the family,²³ the same preverb frequently takes a different light verb. Preverbs can occur with different numbers of light verbs in each language; where the alternations are systematic, they might not appear in dictionaries. In a few cases, there are some systematic differences in the languages that allow us to infer a change. A case in point are the preverbs which appear with the light verb -k- ‘carry’ in Nyulnyul and -ma- ‘put’ in Bardi. To carry someone/something on one’s shoulders, for example, is kurndu -k- in Nyulnyul, but goondoo -ma- in Bardi. In this case, although Bardi also has preverbs with the cognate light verb -ga-, they are much less numerous than the preverbs which take -ma-. These facts, taken together, allow us to infer that there was a change in Bardi whereby a class of coverbs shifted light verb, and the functions of -ga- came to be taken over by -ma-.

22 Clausal negation in Nyulnyulan languages forces the verb into the irrealis mood. See further McGregor and Wagner (2006). 23 Part of the problem is that the information was not systematically recorded in the dictionaries for Yawuru and Nyikina, and published lexical information for Warrwa is incomplete; thus unless the preverb appears in the grammatical descriptions of those languages in an example sentence, we do not have information about which light verb(s) the preverb can combine with. This limits the data pool.

Complex Predicates in Australian languages 

 287

In some other cases, however, complex predicate reconstruction is more hopeful. For example, it appears that complex predicates – at least with the verbs ‘do’ and ‘make’ – can be reconstructed to Proto-Western Pama-Nyungan.²⁴ Cognate verbs are found in Yolngu languages as derivational stem formatives; they also appear in fossilised and lexically determined meanings. For example, the verbaliser -thu combines with kinship terms to mean ‘call someone by that kinship term’ (e.g. yapa’yun ‘call someone “elder sister”’). The construction is found in Yolngu, and in at least four other distinct subgroups of Western Pama-Nyungan (Marrngu, Ngumpin-Yapa, Wati, and Nyungar), and traces are found in Kartu (Nhanda: see Blevins 2001:115).²⁵

4.2 CVCs in language contact CVCs have been implicated elsewhere in the world in language contact processes. CVCs have been claimed both to facilitate verb borrowing in languages which already have them, and to be more likely to arise in language contact. It has been noted, for example, that the rise of complex predicates with the light verb suru ‘do’ in Japanese is contemporaneous with the increase in borrowing of lexical items from Chinese in the language (Lanz 2009).²⁶ There is some evidence that this is the case in Australia too. Firstly, most of the languages with the most productive complex predicate constructions are found around the Pama-Nyungan ~ non-Pama-Nyungan border areas. While I have presented evidence above that the construction is reconstructible well back in Pama-Nyungan, the systems with the most light verbs, with classification systems, and where the CVCs have a sizable functional load in the language, are all in a single geographical area. This may be a case of language contact leading to preservation of grammatical features, or it may be a case of systemic elaboration. Second, there is some evidence that CVCs have facilitated verb borrowing (specifically coverb borrowing) in north Australia. While loan levels vary substantially across the region, they are strongly related to coverb syntax. McConvell and Bowern (2011) investigated the correlations between coverb ~ light verb nexus (see above) and

24 I assume the classification and family tree presented in Bowern (2011) and Bowern and Atkinson (2012). 25 Though Dixon (2002) argues that the presence of complex predicates in northern Pama-Nyungan languages such as Gurindji is due to areal influence, this reconstruction implies that the construction was present at an earlier stage. It may well be that subsequent contact with non-Pama-Nyungan languages has preserved and further developed the construction, but there is evidence against an argument that it was innovated de novo due to contact. 26 Of course, it is impossible to tell here whether the complex predicate structure facilitated the loaning of verbs, or whether the increase in loaned verbs led to the greater use of the complex predicate construction.

288 

 Claire Bowern

found a strong negative correlation (r = -0.78) between the number of coverb loans in a language and the freeness of its coverbs. That is, the looser the nexus between coverb and light verb, the more likely the language is to have a substantial number of coverb loans. Such languages also have coverbs which bear a large functional load in the verb system. While coverbs are freely borrowable (as would be expected in an open word class), light verbs are almost never borrowed. One exception is the light verb ‘do/say’ in the Marrngu language Karajarri. Karajarri has borrowed the Nyulnyulan verb ‘do/say’ yintanpa; the loan is most likely from Yawuru, although the form is also found in other Nyulnyulan languages. The verb is parsable in Nyulnyulan as yintan ‘he/she says/ does it’, with the filler syllable -pa. It is the most common light verb in Nyulnyulan languages, and since Karajarri has borrowed coverbs from Yawuru, it is likely that they were borrowed as a construction, along with their light verb, which was then inflected according to the morphology of the recipient language. (24) yi- n- t -an 3- tr- do/say -cont ‘He/she says/does it.’

5 Conclusions Complex predicates, particularly light verb constructions, are an important part of the syntax of Australian languages. They are internally diverse, varying in composition of the light verb inventory, in the role of the coverb, in the degree of integration with other predicates, and in their syntax. While descriptive work on these constructions has often either lumped them together or has overlooked their uniqueness (or difference from constructions such as adverbial constructions or auxiliated constructions), there is, nonetheless, a great deal of information from existing sources which allows us to draw conclusions about cross-linguistic variation in the syntax of complex verb constructions.

Appendix Languages in the sample: Family

Subgroup

Bunuban Marran

Mara-Alawa

Language

Source

Bunuba Gooniyandi Alawa Marra

Rumsey (2000) McGregor (1990) Sharpe (1972) Heath (1981)

Complex Predicates in Australian languages 

Mangarrayi Jarrakan

Mirndi Wagiman Nyulnyulan

Eastern Nyulnyulan

Western Nyulnyulan

Pama-Nyungan

Central NSW Marrngu Ngumpin-Yapa

Paman Macro-NSW Tiwi Western Daly

Worrorran

Northern Western Eastern

Warndarang Mangarrayi Gajirrabeng Miriwoong Kija Jaminjung Jingulu Wagiman Nyikina Warrwa Yawuru Bardi Nyulnyul Ngiyambaa Karajarri Nyangumarta Gurindji Jaru Mudburra Walmajarri Warlpiri Kuku-Yalanji Muruwari Tiwi Young People’s Tiwi Bachamal Emmi Malak Malak Maranunggu Marrithiyel Gunin/Gwini Ungarinyin Worrorra

 289

Heath (1980) Merlan (1982) Kofod (1976) Kofod (1978) Blythe (2001) Schultze-Berndt (2000) Pensalfini (2003) Wilson (1999) Stokes (1982) McGregor (1994) Hosokawa (1991) Bowern (2004a; 2008c; 2010) Nekes, Worms and McGregor (2006); McGregor (1996) Donaldson (1980) Sands (1989); Sharp (2010) Sharp (2004) McConvell (fieldnotes) Tsunoda (1981) McConvell (fieldnotes) Hudson (1978) Nash (1982) Patz (2002) Oates (1988) Osborne (1974) Lee (1987) Ford (1990) Ford (1998) Birk (1976) Tryon (1970) Green (1989) McGregor (1993) Rumsey (1984) Clendon (2000b)

References Alpher, Barry 1991 Yir-Yoront Lexicon: Sketch and Dictionary of an Australian Language. Berlin; New York: Mouton de Gruyter. Alpher, Barry 2001 Ideophones in interaction with intonation and the expression of new information in some indigenous languages of Australia. In: F. K. Erhard Voeltz and Christa Kilian-Hatz (eds.), Ideophones (Typological Studies in Language 44), 9–24. Amsterdam: John Benjamins. Alsina, Alex. 1997. Causatives in Bantu and Romance. In: Alex Alsina, Joan Bresnan, and Peter Sells (eds.), Complex Predicates, 203–245. Palo Alto: CSLI Publications. Amberber, Mengistu, Brett Baker and Mark Harvey 2010 Complex Predicates: Cross-linguistic Perspectives on Event Structure. Cambridge: Cambridge University Press.

290 

 Claire Bowern

Austin, Peter 1981 A Grammar of Diyari, South Australia. Cambridge: Cambridge University Press. Austin, Peter (ed.) 1988 Complex Sentence Constructions in Australian Languages. (Typological Studies in Language 15) Amsterdam: John Benjamins. Austin, Peter 1998 “Crow is sitting chasing them”: Grammaticalisation and the verb “to sit” in the Mantharta languages, Western Australia. In: Anna Siewierska and Jae Jung Song (eds.), Case, Typology, and Grammar: In Honor of Barry J. Blake, 19–35. Amsterdam: John Benjamins. Austin, Peter and Joan Bresnan 1986 Non-configurationality in Australian Aboriginal languages. Natural Language and Linguistic Theory 14(2): 215–268. Baker, Brett and Mark Harvey 2010 Complex predicate formation. In: Mengistu Amberber, Brett Baker and Mark Harvey (eds.), Complex Predicates: Cross-linguistic Perspectives on Event Structure, 13–47. Cambridge: Cambridge University Press. Birk, D. B. W. 1976 The Malakmalak Language, Daly River (Western Arnhem Land). (Pacific Linguistics B-45) Canberra: Australian National University. Blevins, Juliette 2001 Nhanda: An Aboriginal language of Western Australia. Honolulu: University of Hawai’i Press. Blythe, Joseph 2001 Yuwurriyangem Kijam (Our language Kija): A Phrasebook of the Kija language. Halls Creek, Australia: Kimberley Language Resource Centre. Bowern, Claire 2004a Bardi verb morphology in historical perspective. Cambridge MA: Harvard University PhD thesis. Bowern, Claire 2004b (Some notes on) complex predicates in Turkic. In: Heejeong Ko and Maryanne Walter (eds.), Proceedings of the First Workshop on Altaic Formal Linguistics, 33–48. Cambridge, Mass: MIT Working Papers in Linguistics. Bowern, Claire 2005 Review of Bininj Gun-wok: A Pan-Dialectal Grammar of Mayali, Kunwinjku and Kune, by Nicholas Evans, and The Non-Pama-Nyungan Languages of Northern Australia, edited by Nicholas Evans. Linguistic Typology 9: 341–350. Bowern, Claire 2008a The diachrony of complex predicates. Diachronica 25(2): 161–185. Bowern, Claire 2008b Bardi complex predicates as a challenge to monotonicity. In: Miriam Butt and Tracy Holloway King (eds.), Proceedings of LFG08, 190–197. Stanford CA: CSLI Publications. Bowern, Claire 2008c The reconstruction of Nyulnyulan complex predication. Diachronica 25(2): 161–185. Bowern, Claire 2010 The typological implications of Bardi complex predicates. Linguistic Typology 14(1): 45–61. Bowern, Claire 2011 Computational phylogenetics and the internal structure of Pama-Nyungan. Plenary paper presented at the Symposium on Historical Linguistics, Linguistic Society of America annual meeting, Pittsburgh PA, 6–9 January. Bowern, Claire 2012 A Grammar of Bardi. Berlin: Mouton de Gruyter. Bowern, Claire and Quentin Atkinson 2012 Computational Phylogenetics and the Internal Structure of Pama-Nyungan. Language 88(4): 817–845 Butt, Miriam and Willem Geuder 2001 On the (semi) lexical status of light verbs. In: Norbert Corver and Henk van Riemsdijk (eds.), Semi-lexical Categories: The Function of Content Words and the Content of Function Words, 323–369. Berlin: Mouton de Gruyter. Butt, Miriam 1995 Complex Predicates in Urdu. Standford CA: CSLI Publications. Capell, Arthur and H. E. Hinch 1970 Maung grammar. The Hague: Mouton. Cardinaletti, Anna and Ur Shlonsky 2004 Clitic positions and restructuring in Italian. Linguistic Inquiry 35(4): 519–557. doi:10.1162/0024389042350523. Clendon, Mark 2000a Topics in Worrorra Grammar. Adelaide: University of Adelaide PhD thesis. Clendon, Mark 2000b A Grammar of Worrorra. Halls Creek, Australia: Kimberley Language Resource Centre.

Complex Predicates in Australian languages 

 291

Cook, Anthony 1988 Participle sentences in Wakiman. In: Peter Austin (ed.), Complex Sentence Constructions in Australian languages (Typological Studies in Language 15), 69–90. Amsterdam: John Benjamins. Croft, William 2007 Intonation units and grammatical structure in Wardaman and in cross-linguistic perspective. Australian Journal of Linguistics 27(1): 1–39. doi:10.1080/07268600601172934. Dixon, R. M. W. 1972 The Dyirbal Language of North Queensland. (Cambridge Studies in Linguistics 9) London: Cambridge University Press. Dixon, R. M. W. 2002 Australian Languages: Their Nature and Development. (Cambridge Language Surveys) Cambridge: Cambridge University Press. Donaldson, Tamsin 1980 Ngiyambaa: The Language of the Wangaaybuwan of New South Wales. (Cambridge Studies in Linguistics 29) Cambridge: Cambridge University Press. Douglas, W. H. 1976 The Aboriginal Languages of the South-west of Australia. 2nd edn. Canberra: Australian Institute of Aboriginal Studies. Evans, Nicholas 2003 Bininj Gun-wok: A pan-dialectal Grammar of Mayali, Kunwinjku and Kune. 2 volumes. (Pacific Linguistics 541) Canberra: Australian National University. Foley, William A. and Mike Olson 1985 Clausehood and verb serialization. In: Johanna Nichols and Anthony Woodbury (eds.), Grammar Inside and Outside the Clause, 17–60. Cambridge: Cambridge University Press. Folli, R., H. Harley and S. Karimi 2005 Determinants of event type in Persian complex predicates. Lingua 115(10): 1365–1401. Ford, Lysbeth J. 1990 The phonology and morphology of Bachamal (Wogait). Canberrra: Australian National University MA thesis. Ford, Lysbeth J. 1998 A description of the Emmi Language of Northern Territory of Australia. Canberrra: Australian National University PhD thesis. Goddard, Cliff 1988 Verb serialisation and the circumstantial construction in Yankunytjatjara. In: Peter Austin (ed.), Complex Sentence Constructions in Australian Languages (Typological Studies in Language 15), 177–192. Amsterdam: John Benjamins. Green, Ian 1989 Marrithiyel: A language of the Daly River region of Australia’s Northern Territory. Canberrra: Australian National University PhD thesis. Green, Rebecca 1987 A sketch grammar of Burarra. Canberrra: Australian National University BA Honours thesis. Grimshaw, J. and Armin Mester 1988 Light verbs and θ-marking. Linguistic Inquiry 19(2): 205–232. Hale, Ken L. and Samuel J. Keyser 2002 Prolegomenon to a Theory of Argument Structure. Cambridge, Mass.: The MIT Press. Haspelmath, Martin and Uri Tadmor 2009 Loanwords in the World’s Languages: A Comparative Handbook. Berlin: Mouton de Gruyter. Heath, Jeffrey 1980 Basic Materials in Warndarang: Grammar, Texts and Dictionary. (Pacific Linguistics B-72) Canberra: Australian National University. Heath, Jeffrey 1981 Basic Materials in Mara: Grammar, Texts, and Dictionary. (Pacific Linguistics C-60) Canberra: Australian National University. Hosokawa, Komei 1987 Compound verb formations in Australian languages: with special reference [Kyōtodaigaku gengo-gaku kenkyūkai, Kyoto University to Yawuru preverbs. Linguistics Society] 6: 115–115. Hosokawa, Komei 1991 The Yawuru language of West Kimberley: A meaning-based description. Manuscript. Hudson, Joyce 1978 The Core of Walmatjari Grammar. Canberra: Australian Institute of Aboriginal Studies. Jackendoff, Ray 1992 Semantic Structures. (Current Studies in Linguistics 18) Cambridge, Mass.: The MIT Press.

292 

 Claire Bowern

Koch, Harold 1984 The category of “associated motion” in Kaytej. Language in Central Australia 1: 23–34. Kofod, Frances M. 1976 Simple and compound verbs: Conjugation by auxiliaries in Australian verbal systems: Miriwung. In: R. M. W. Dixon (ed.), Grammatical Categories in Australian Languages, 646–653. Canberra/Atlantic Highlands, NJ: Australian Institute of Aboriginal Studies/ Humanities Press. Kofod, Frances M. 1978 The Miriwung language (East Kimberley): A phonological and morphological study. Armidale NSW: University of New England MA thesis. Lanz, Linda A. 2009 Diachrony of complex predicates in Japanese. Rice Working Papers in Linguistics. Vol 1. http://hdl.handle.net/1911/21856 Laughren, Mary 2010 Warlpiri verbs of change and causation: The thematic core. In: Mengistu Amberber, Brett Baker and Mark Harvey (eds.), Complex Predicates: Cross-linguistic Perspectives on Event Structure, 167–236. Cambridge: Cambridge University Press. Lee, Jennifer 1987 Tiwi Today: A Study of Language Change in a Contact Situation. (Pacific Linguistics C-96) Canberra: Australian National University. McConvell, Patrick and Claire Bowern 2011 Reconstructing the syntax of complex verb constructions in Ngumpin-Yapa languages (Northern Australia). Paper presented at ICHL XX, Osaka, Japan. McGregor, William B. 1990 A Functional Grammar of Gooniyandi. (Studies in Language Companion Series 22) Amsterdam: John Benjamins. McGregor, William B. 1993 Gunin/Kwini. München: Lincom Europa. McGregor, William B. 1994 Warrwa. München: Lincom Europa. McGregor, William B. 1996 Nyulnyul. München: Lincom Europa. McGregor, William B. 2002 Verb Classification in Australian Languages. Berlin: Mouton de Gruyter. McGregor, William B and Tamsin Wagner 2006 The Semantics and Pragmatics of Irrealis Mood in Nyulnyulan Languages. Oceanic Linguistics 45(2): 339–379. Merlan, Francesca 1982 Mangarayi. (Lingua Descriptive Series 4) Amsterdam: North Holland Publishing Company. Merlan, Francesca 1994 A Grammar of Wardaman: A Language of the Northern Territory of Australia. (Mouton Grammar Library 11) Berlin: Mouton de Gruyter. Morphy, Frances 1983 Djapu, a Yolŋu dialect. In: R. M. W. Dixon and Barry J. Blake (eds.), Handbook of Australian languages, vol. 3, 1–188. Canberra: Australian National University Press. Nash, David 1982 Warlpiri verb roots and preverbs. In: Stephen M. Swartz (ed.), Papers in Warlpiri grammar: In memory of Lothar Jagst (Work Papers of SIL-AAB Series A Volume 6), 165–216. Darwin: Summer Institute of Linguistics. Nekes, Hermann, Ernest A. Worms and William B. McGregor (eds.) 2006 Australian Languages. Berlin: Mouton de Gruyter. Nordlinger, Rachel 1998 A Grammar of Wambaya, Northern Territory (Australia). (Pacific Linguistics C-140) Canberra: Australian National University. Nordlinger, Rachel 2001 Wambaya in motion. In Jane Simpson, Mary Laughren, Peter Austin, David Nash and Barry Alpher (eds.), Forty Years On: Ken Hale and Australian languages (Pacific Linguistics 512), 401–413. Canberra: Australian National University. Nordlinger, Rachel 2010 Complex predicates in Wambaya: Detaching predicate composition from syntactic structure. In: Mengistu Amberber, Brett Baker and Mark Harvey (eds.), Complex Predicates: Cross-linguistic Perspectives on event structure, 237–258. Cambridge: Cambridge University Press. Nordlinger, Rachel 2014 Serial verbs in Wambaya. In: Robert Pensalfini, Myfany Turpin and Diana Guillemin (eds.), Language Description Informed by Theory, 263–282. Amsterdam: John Benjamins.

Complex Predicates in Australian languages 

 293

Oates, Lynette F. 1988 The Muruwari Language. (Pacific Linguistics C-108) Canberra: Australian National University. Osborne, C. R. 1974 The Tiwi Language. Canberra: Australian Institute of Aboriginal Studies. Patz, Elisabeth 2002 A Grammar of the Kuku Yalanji Language of North Queensland. (Pacific Linguistics 527) Canberra: Australian National University. Pensalfini, Robert 2003 A Grammar of Jingulu: An Aboriginal language of the Northern Territory. (Pacific Linguistics 536) Canberra: Australian National University. Reid, Nicholas 2003 Phrasal verb to synthetic verb: structural change in Ngan’giwutimirri. In: Nicholas Evans (ed.), The Non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region (Pacific Linguistics 552), 95–123. Canberra: Australian National University. Rumsey, Alan 1984 An Intra-sentence Grammar of Ungarinjin, Northwestern Australia. (Pacific Linguistics B-86) Canberra: Australian National University. Rumsey, Alan 2000 Bunuba. In: R. M. W. Dixon and Barry J. Blake (eds.), The Handbook of Australian Languages, vol. 5, 34–153. Oxford: Oxford University Press. Sands, Kristina 1989 A grammar of Karajarri. Canberra: Australian National University BA Honours thesis. Schultze-Berndt, Eva 2000 Simple and Complex Verbs in Jaminjung: a study of event categorisation in an Australian language. (MPI Series in Psycholinguistics 14) Nijmegen: E. SchultzeBerndt. Schultze-Berndt, Eva 2003 Preverbs as an open word class in Northern Australian languages: Synchronic and diachronic correlates. In: Geert Booij and Jaap Marle (eds.), Yearbook of Morphology 2003, 145–177. Dordrecht: Springer Netherlands. http://www.springerlink.com/ content/mthn7k7616021043/ (accessed 6 February, 2012). Seiss, Melanie 2009 On the difference between auxiliaries, serial verbs and light verbs. In: Miriam Butt and Tracy Holloway King (eds.), Proceedings of the LFG09 Conference, 501–519. Stanford CA: CSLI Publications. Sharp, Janet 2004 Nyangumarta: A Language of the Pilbara Region of Western Australia. (Pacific Linguistics 556) Canberra: Australian National University. Sharp, Janet 2010 Karajarri Grammar. Manuscript. Sharpe, Margaret C. 1972 Alawa Phonology and Grammar. Canberra: Australian Institute of Aboriginal Studies. Silverstein, Michael 1986 Classifiers, verb classifiers, and verbal categories. Proceedings of the Annual Meeting of the Berkeley Linguistics Society 12: 497–514. Simpson, Jane H. 1991 Warlpiri Morpho-syntax. Dordrecht: Kluwer Academic. Stokes, Bronwyn 1982 A description of Nyigina, a language of the west Kimberley, Western Australia. Canberra: Australian National University PhD thesis. Tryon, Darrell T. 1970 An introduction to Maranungku. (Pacific Linguistics B-15) Canberra: Australian National University. Tsunoda, Tasaku 1981 The Djaru Language of Kimberley, Western Australia. (Pacific Linguistics B-78) Canberra: Australian National University. Tunbridge, Dorothy 1988 Affixes of motion and direction in Adnyamathanha. In: Peter Austin (ed.), Complex Sentence Constructions in Australian languages (Typological Studies in Language 15), 267–283. Amsterdam: John Benjamins Wilkins, David 1988 Switch-reference in Mparntwe Arrernte (Aranda): Form, Function, and Problems of Identity. In: Peter Austin (ed.), Complex Sentence Constructions in Australian Languages (Typological Studies in Language 15), 141–176. Amsterdam: John Benjamins. Wilkinson, Melanie 1991 Djambarrpuyngu. A Yolŋu variety of Northern Australia. Sydney: University of Sydney PhD thesis.

294 

 Claire Bowern

Wilson, Aidan 2009 Negative evidence in linguistics. Sydney: University of Sydney BA Honours thesis. Wilson, Stephen 1999 Coverbs and Complex Predicates in Wagiman. Chicago: University of Chicago Press. Wood, Esther 2007 The semantic typology of pluractionality. Berkeley CA: University of California at Berkeley PhD thesis.

Alice Gaby and Ruth Singer

7 Semantics of Australian Languages 1 Introduction Genetic relatedness among all of Australia’s languages will probably never be proven.¹ Nevertheless, recurring patterns of meaning – in the absence of lexical cognates – reveal cultural connections stretching right across the continent. There is meaning to be found at almost every level of grammar, from sublexical morphemes and processes, to individual words and the relationships between them, to larger constructions. As such, this chapter touches on a wide range of phenomena, revealing the interconnectedness of language and its contexts of use. Among these heterogeneous topics, there are nevertheless a number of recurring themes. The first of these themes is how people use language to divide the world and their experience of it into categories. This includes ontologies and classification of the natural world (§ 2.1 & § 2.2), familial relations and kinship (§ 4.1), as well as the overt and covert meaning of grammatical systems of noun and verb classification (§ 2.2, § 2.3, cf. Bowern, this volume). A second theme concerns the organisation of meaning within categories themselves. This includes the identification of prototypes and delineation of categorical boundaries (§ 2.2, § 3.3, § 4.1). A third theme explores the meaning relationships that obtain between categories, including antonymy, metonymy, meronymy, hyponymy, metaphor and the distinction between polysemy and vagueness (§ 3.1–3.3, § 4.1, § 4.3, § 4.4). A fourth theme is the mapping between distinct semiotic systems, for instance between language and gesture, different linguistic registers and so on (§ 3.2, § 4.1, § 4.5, § 7). And finally, there emerges a theme of the embedding of semantics in communication, such that meaning structures cannot be understood in isolation from language use and language change (§ 3, § 4.1, § 5.1.1, § 6, § 7, § 8). We begin in § 2 by considering classification of the natural world, choosing ethnobiology as a quintessentially Australian illustration. § 2.2 considers the semantic implications of the grammatical system of nominal classification. Here we see metaphor and metonymy exploited to justify the system’s macro-categories. This thread is picked up and elaborated on in §  3, which considers the nature of the lexical sense relations antonymy, metonymy, metaphor, polysemy and vagueness. § 4 goes on to consider five conceptual domains that have received particular attention from Australian semanticists, whether because of their exceptionally high or low degree of elaboration (e.g. kinship and mathematics respectively). The thread of metaphori-

1 For their suggestions for literature to include in this survey, thanks to Jenny Green, Harold Koch and Joe Blythe. They should not be blamed for omissions. Neither should the authors – there’s a lot out there!

296 

 Alice Gaby and Ruth Singer

cal and metonymic linkages continues to run through this section. §  5 revisits the semantic significance of grammatical systems introduced in § 2. § 5.1 considers how speakers exploit case marking to express secondary meanings, while § 5.2 emphasises the difficulty of analysing subtle semantic categories such as mood, especially as a non-native speaker. In § 6 we take a diachronic perspective, highlighting the role of semantics and pragmatics in language change. Finally, § 7 zooms in to look at these pragmatic processes in synchronic interaction, considering how these may best be studied. At a number of points in the chapter we note strikingly common patterns found across Australia, which seem to derive from a uniquely Australian Indigenous worldview (Hale 1986; Hiatt 1978).

2 Categorisation and classification 2.1 Ethnobiology Most Australian Aboriginal groups were traditionally hunter-gatherers who moved from place to place within an area around their clan territory. This way of life, adapted to the vagaries of the Australian climate, required an intimate understanding of the land and its cycles. The significance of flora and fauna is highlighted by their central place in Aboriginal religion, song, dance and stories. The names given to plants and animals in Australian languages reflect a system of classification of the natural world. As elsewhere these operate at three main levels: the species level (e.g. ‘Silver Bream’) and one level one up from this; the generic level (e.g. ‘bream’) and above these the superordinate terms such as ‘fish’, ‘animals’ etc. Systems of classification in Australian languages present a number of challenges to Berlin’s (1992) claims about universals in ethnobiological classification. Firstly, Berlin claims that binomial terms are most commonly used to refer to species crosslinguistically. Binomial terms combine a species level term with a genus level term such the English common name Spotted Gum or its scientific name Corymbia maculata. However, among Australian languages monomial species terms are the norm and genus level terms rare (Baker 2007; Heath 1978). Baker (2007) explains this with reference to the large genera that dominate the Australian flora; Eucalyptus and Acacia. Brown (1985) on the other hand argues that monomials are used more commonly by hunter-gatherers, binomials by agriculturalists. The Australian data generally support this theory, though McKnight (1999) points out one exception: while monomial terms are the norm in ordinary Lardil, the associated initiation language Damiin uses binomial terms.²

2 See § 3.2 for more discussion of the Damiin special register.

Semantics of Australian Languages 

 297

Berlin (1992) also claims that the genus level is psychologically primary in all systems of ethnobiological classification. However, genus level terms are rather rare in Australian languages (Heath 1978). For example in Mawng, there is a superordinate term referring to turtles and dugongs, inyarlgan, and at least twenty specific terms for turtle species. And yet there is no generic term for turtles in Mawng. Instead the species name of the most psychologically salient turtle manpiri ‘Green turtle’ is used much like what Heath (1978) describes as a “quasi-generic” in Nunggubuyu (Wubuy). These terms, spanning two levels of the taxonomic hierarchy, are common in Australian languages but rare elsewhere. A handful of genuine generics are found in each language, such as Mawng arukin ‘snake’ and kiyap ‘fish’, but these tend not cover the full range of specific terms, leaving the majority without a corresponding generic. For example, there is no generic term for shellfish, although these are an important source of food for Mawng speakers.³ This does not necessarily mean, however, that speakers do not conceptualise higher level categories – other areas of the language can reveal covert patterns of classification. However even covert categories tend to distinguish categories at the family level or above rather than the genus level. For example in Mawng there is a specific idiom for ‘hunt yams’ – the verb -wu ‘hit, kill’ is used with Edible gender object agreement – although there is no generic for yams. The category ‘yam’ reflects traditional interactions with the natural world rather than a genetic family or genus. Nominal classification systems of Australian languages also usually distinguish superordinate categories; in Mawng all yams and edible fruit are Edible gender, crabs are Feminine gender and most other animals are Masculine gender (Singer 2010). The account of classification in Yanyuwa in Bradley et al. (2006) is one of the most detailed descriptions of a system of ethnobiological classification available for an Australian language. For example, they show a large number of terms for male and female dugongs of different ages. They also give a detailed account of superordinate terms for types of fish. Bradley et al. (2006) argue that it is the interactions people have with their environment that have the primary influence on Yanyuwa plant and animal classification rather than any universal principles. The classification of plants and animals can also relate directly to social organisation. Waddy (1982), Povinelli (1990), Bradley et al. (2006) inter alia, discuss how plants and animals are classified relative to kinship and clans. In addition to work on classification, many more practically-oriented books on traditional ethnobiological knowledge have been compiled by Northern Territory botanist Glenn Wightman in collaboration with linguists, anthropologists and Aboriginal communities. See Puruntatameri et al. (2001) and Wightman et al. (1992) for examples.

3 But see Douglas (1976) for a discussion of the Western Desert language, which appears to have quite a large number of generics for snakes and insects.

298 

 Alice Gaby and Ruth Singer

2.2 Nominal Classification Among Australian languages we find a wide variety of systems of nominal classification (Dixon 1968, 1982, 1986; Harvey and Reid 1997; Sands 1995). These systems divide the world up into different classes which must be treated differently by the grammar (Aikhenvald 2006). The North Queensland language Dyirbal uses four groups, which Dixon (1982) refers to as “noun classes”. When using a noun, speakers of Dyirbal must select the correct noun class marker to go with each noun. (1)

Dyirbal noun classes Masculine Feminine Vegetable Neuter

bayi jaja balan jaja balam mirrany bala gajin

‘boy baby’ ‘girl baby’ ‘black bean tree’ ‘yamstick’

Nominal classification systems vary in terms of how transparently they are organised. In Australian languages they typically have a strong semantic basis, which means that the membership of each class can be described with reference to a list of semantic properties. This contrasts with the nominal classification systems of European languages, usually referred to as gender systems, whose membership is better explained with reference to form rather than meaning. However, nominal classification systems are rarely completely tidy; there are always many classifi cations that are difficult to explain. Metaphorical links between members of a class can explain some of the exceptions. For example, Dixon (1982) argues that the Feminine class in Dyirbal includes both women, fire, snakes, scorpions and the ‘hairy mary grub’ because these are all considered “dangerous”.⁴ The semantics of the Dyirbal system is used as a key example in Lakoff ’s (1987) Women, fire and dangerous things to demonstrate the importance of metaphor in language and cognition. The nominal classification systems of Australian languages can be divided into those which exhaustively classify everything – generally referred to as genders or noun classes⁵ – and those which do not, generally referred to as “classifier systems”. Languages with classifier systems may have a large number of classes but some nouns do not co-occur with a class marker. For example, Murrinh-Patha has ten classifiers (Walsh 1993) whereas Mawng has five genders. Within classifier systems it is normal for there to be some cross-classification, i.e. the same noun can take more than one classifier. Cross-classification and generally more creative uses of classification are also found

4 See Plaster and Polinsky (2010) for a recent reanalysis of Dyirbal classification. They argue that only a small set of simple semantic features are required to explain Dyirbal classification, in combination with some morphophonemic features. 5 The terms “noun class” and “gender” are used interchangeably in the literature (see Corbett 1991).

Semantics of Australian Languages 

 299

in semantically-based gender systems as Singer (2010) shows for Mawng. An example of cross-classification in a gender system from Bininj Gun-wok is given in (2). (2)

Bininj Gun-wok (Nicholas Evans, pers. comm.) Masculine Feminine Vegetable Neuter

na-bang al-bang an-bang gun-bang

dangerous man dangerous woman dangerous food, poison grog

A key difference between gender systems and classifier systems is that classifiers are often homophonous with generic nouns whereas gender markers tend to be further grammaticalised. In Mparntwe Arrernte, all classifiers also occur as generic nouns, so Wilkins (2000) argues the language is better analysed as having “classifier constructions” which the generic nouns participate in, rather than having a distinct part of speech “classifier”.

2.3 Verb classification In addition to the elaborate nominal classification systems found in many Australian languages we can also find systems of verb classification. These involve complex verbs which have two parts: one plays the role of the classifier while the other combines with it to provide more specific information. The reader is referred to Bowern (this volume) for a discussion of the semantics of complex predicates and the extent to which light verbs may be considered to classify the coverb.

3 Lexical sense relations The meanings of words are only understood in relation to the meanings of other words. In each language, meanings are distributed among word forms in slightly different ways, which is one reason that attempts at perfect translations will always fail. Even if a language appears to divide up the domain of experience similarly to English, there will always be subtle differences once we look in depth. For example in the central Australian language Pitjantjatjara there are three separate words for ‘sit’, ‘stand’ and ‘lie’ to describe the most common human postures, just as in English. However in Pitjantjatjara the word ‘lie’ can also mean ‘sleep’ (Goddard and Harkins 2002). Interestingly, the verb meaning ‘lie’ -u also means ‘sleep’ in Mawng, thousands of kilometres away. Despite the low level of lexical cognates between Australian language families, common patterns of lexical sense relations can stretch right across the continent.

300 

 Alice Gaby and Ruth Singer

The categorisation of the colour spectrum is notoriously variable across languages. Australian languages tend to have very few basic colour terms – terms which are only ever used in reference to colour. Kay (2003) gives the Warlpiri term walyawalya as a good example of a non-basic colour term. It is derived from the term walya ‘earth’ by reduplication and covers the range of colours found in the earth of the central desert: brown, red, yellow etc. Davis (1982) argues that the colour terms of the Yolngu-Matha language group contrast mainly in terms of brightness, rather than hue, which is the primary contrast between English colour categories. Hargrave (1982) supports this, comparing studies of four traditional Australian languages, which suggest that the only true basic colour terms are ‘black’ and ‘white’, which differ in brightness but not in hue. Hill (2011) points to work which discusses the importance of luster, iridescence and illuminance in Aboriginal perception of the visual domain. These properties are closely allied to brightness and their presence or absence seems to be much more important to visual perception than a notion of colour that is based on hue. Applied semantic analysis is illustrated by the many dictionaries produced (e.g. Kilham et al. 1986, Laughren et al. forthcoming) and several grammars provide a detailed analysis of at least some semantic domains. Dixon (1972), for example, has a chapter devoted to the “semantics” of Dyirbal, while Dixon (1977) has a detailed discussion of Yidiny lexical semantics in a chapter dedicated to the “lexicon”. The literature also abounds in detailed, insightful analyses of various verbal domains in Australian languages. In addition to Goddard and Harkins’ (2002) work on Pitjantjatjara and Yankunytjatjara posture verbs mentioned above, Reid (2002) investigates how the language Ngan’gityemerri divides up the domain of “posture”. Riemer (2005) shows that Warlpiri verbs of percussion and impact can be more simply defined with reference to the result they produce such as ‘produce point-like depression in a surface’, rather than the action they involve (c.f. English hit, punch, slap). Rumsey (2001) analyses the semantics of ‘trying’ in Ungarinyin (Ngarinyin) and Yidiny, Wilkins (2004) looks at the expression of motion in Mparntwe Arrernte narratives, Evans (2007b) discusses expressions for remembering in Dalabon, and Gaby looks at verbs for ‘cutting’ and ‘breaking’ (2007) and ‘putting’ and ‘taking’ (2012) events in Kuuk Thaayorre. The next few subsections look at recurring patterns in lexical sense relations found across Australian languages.

3.1 Polysemy, metonymy and metaphor: recurring linkages Patterns of polysemy found among distantly related Australian languages capture common metaphors rooted in the culture of their speakers. Evans (1992a) argues that cultural similarities combined with widespread multilingualism have led to deep parallels in semantic structure right across the continent. One common type of polysemy found in Australian languages has been referred to as actual/potential polysemy

Semantics of Australian Languages 

 301

(Dixon 1980: 102–103). Examples include the Yolngu term gurtha ‘fire, firewood’ (Lowe 2004: 95). The same word is used to refer both to something in one stage of transformation and also in a later stage. This pattern is found not only in the nominal domain but also in the verbal domain, where it can be seen in verbs that refer not only to a process but also to its result, such as the much discussed Warlpiri verb pakarni ‘hit, kill by hitting’ (e.g. Riemer 2005). Patterns of polysemy found in Australian languages are often strikingly different to those found elsewhere. In English we can ask Do you see what I mean? – using the verb see to mean ‘understand’ or ‘know’. Sweetser (1990) suggests the extension of the meaning ‘see’ to ‘know’ reflects a universal tendency to prioritise visual perception as a source of knowledge over other modes of perception. However, Evans and Wilkins (2000) show that the extension of the meaning ‘see’ to ‘know’ is very rare among Australian languages; instead the most common source for ‘know’ is ‘hear’. Evans and Wilkins use synchronic polysemy, pragmatic implicatures and diachronic data as evidence. For example, they point to reflexes of the Proto-Pama-Nyungan term *pina ‘ear’ such as Warlpiri pina ‘wise, knowing experienced’. Evans and Wilkins argue that a stage in which there was polysemy between ‘hear’ and ‘know’ provided the bridging context for this pathway of semantic change. The underlying motivation for the Australian metaphor hear=know is a cultural conceptualisation that places hearing as the primary source of knowledge rather than vision (Evans 2002). This is not to say, however, that the eye and seeing do not also feed knowledge metaphors in Australia (cf. Warlpiri ngurrju=rna nyangu ‘I saw/realised that it was good’). Future research in linguistics and related fields (such as anthropology and cognitive psychology) is required to evaluate the relative importance of visual and aural modalities in the conceptualisation of knowledge in Indigenous Australian cultures. Another sense relation that is common among Australian languages is that of “sign metonymies” (Evans 1992a), usually involving pairs of plant and animal species which share the same name. For example in Kunwinjku, a dialect of Bininj Gunwok, the Spangled Grunter fish and the native White Apple tree both have the name bokorn. In Ndjebbana the unrelated word nja-murdibidj is used to refer to the same two species. The metonymic relation between the two species – i.e. their proximity in time and/or place – is well understood by speakers of these languages: the fruits of the White Apple tree are eaten by the Spangled Grunter. The tree is a sign that the Spangled Grunter may be present, for example in a watercourse which a White Apple tree overhangs. § 4.3 and § 4.5 introduce several patterns of metaphorical linkage that are common in the Australian context.

302 

 Alice Gaby and Ruth Singer

3.2 Relationships between registers The majority of Australian languages have associated varieties such as avoidance registers, initiation languages and auxiliary sign languages. Avoidance registers are used in the presence of relatives – such as a mother-in-law or son-in-law – with whom one must be particularly respectful. As well as being interesting in their own right, these registers can be invaluable tools for semantic analysis. For example Damiin, the avoidance register used by Lardil speakers, has only around 250 words. The relationship between most Damiin and Lardil words is one of hyponymy (Evans 1992a). Thus the Damiin term n!2u⁶ ‘liquid’ includes the meanings of a large range of Lardil terms such as nguka ‘water’, mela ‘sea, seawater’ and kaldirr ‘urine’. There is no corresponding general term for liquids in ordinary Lardil. The meaning of terms in the special registers of other languages can also be related to the ordinary term by metaphor, metonymy and synecdoche (Evans 1992a). Hale (1971) discusses an initiation register used by Warlpiri speakers – known as Tjiliwirri – which is formed through a process of antonymy, using Warlpiri words with (for the most part) the opposite meaning. So to say ‘I am sitting on the ground’ speakers say something that would mean ‘You are standing in the sky’ in ordinary Warlpiri. Other work on avoidance registers includes Dixon (1971), Haviland (1979), McGregor (1989).

3.3 Vagueness Australian languages are known for their classificatory kinship systems in which kinship terms are extended even to those who are not genetic relatives (cf. § 4.1 below and McKnight 1999). Kin terms extend in certain characteristic ways in almost all Australian languages. For example, in Mawng the term for ‘mother’, kamu, is also used to refer to ‘mother’s sister’. This works reciprocally too, so while I call my mother and her sisters all kamu they also all refer to me as their child. The rather broad denotational ranges of basic kinship terms – combined with various observations such as that sisters sometimes breastfed one another’s children – was taken by late nineteenth century anthropologists as evidence for the existence of “group marriage”. Group marriage was an idea popular briefly as a system anthropologists thought was a primitive form of family structure, intermediate between an original system of total promiscuity and the ideal system: monogamous marriage (Hiatt 1996: chapter 3). They believed sets of sisters who were married to sets of brothers raised their children together. However, it is clear that children do know who their biological parents are, and the lifelong bond between a parent and their child is manifested in multiple ways. This raises an inter-

6 Damiin has unusual phonology in that it is the only Australian variety which has click consonants, in fact nasal clicks.

Semantics of Australian Languages 

 303

esting question regarding the lexical semantics of the kinship terms. Are terms that denote multiple biological categories (e.g. FATHER as well as FATHER’S BROTHER) simply vague regarding the differences between them (much as the English term sister is vague with respect to the difference between elder and younger sisters), or are they polysemous, with the prototypical sense of the word (FATHER) extended to refer to others who occupy a socially, biologically and/or culturally analogous role (FATHER’S BROTHER, much as the English term uncle might be used to address a male friend of one’s parents)? The difference between polysemy and vagueness – as well as tests for teasing the two apart – is considered in detail by Riemer (2005), who considers the semantics of Warlpiri (and English) verbs in the semantic domain of percussion. (Cf. Eades, this volume, for discussion of a courtroom misunderstanding fed by differences between the Aboriginal English and Standard Australian English semantic ranges of father).

3.4 Lexicography Developing dictionaries for Australian languages presents a number of challenges. Not least is the challenge of capturing “meaning”, which persists regardless of developments in technology. As Evans and Sasse (2007: 66) note, “the search for meaning in any language is best seen as a never-ending stringing together of hypertextual commentary which gradually leads to a better understanding of the utterances under study”. An additional challenge in the Australian context is making dictionaries of Australian languages accessible to speakers of the language and their descendants. Simpson (1993) gives a historical perspective on dictionaries of Australian languages. She observes changes in the purposes dictionaries are intended for. Early dictionaries were designed for White adults with an interest in Aboriginal languages. More recently dictionaries have been designed for use by Aboriginal children learning to read and write their own languages in bilingual schools. Dictionaries may also be designed for language revival, or a range of other purposes. McConvell et al. (1983), for instance, discuss how a project developing a Meriam Mir dictionary was combined with providing linguistics training to a group of Meriam Mir speakers. Along with this shift in intended audience, there has been a concomitant shift in format. Recent paper dictionaries (including Turpin and Ross 2012, Mackman and Irra Wangga Language Centre 2011) are being joined by an increasing number of dictionaries produced for online and mobile phone platforms (e.g. S. Wilson and Harvey 2001, cf. A. Wilson 2010; McElvenny 2008). Corris et al. (2004) report on a study of how dictionaries of Australian languages are actually used by speakers and their descendants. They discuss some of the possibilities of electronic dictionaries, such as offering a range of interfaces for different groups of users. However, as both Simpson (1993) and Corris et al. (2004) point out, many issues remain for Aboriginal dictionary users regardless of the technology in-

304 

 Alice Gaby and Ruth Singer

volved. Depending on the Aboriginal community involved, these may include: low levels of literacy, few computer skills, the effects of the codification of language variation and issues of credibility: whether a dictionary is seen as a legitimate source of information by the community. For a historical perspective on the production of dictionaries of Australian languages and specific mention of some dictionaries of Australian languages see Goddard and Thieberger (1997), Evans (2007a) and Austin (2003), an updated version of Austin (1983). Walsh’s chapter of this volume also discusses some limitations of dictionaries, particularly vis-à-vis information on the semantic ranges of lexemes.

4 Conceptual domains Among the languages of Australia there is a tendency for particular semantic fields to be highly elaborated in particular ways. §  2.1 gave examples of the fine distinctions made in lexifying the natural world in taxonomies, with many Australian languages possessing large numbers of specific terms for plant and animal species. In § 4.1 we will likewise see the precision with which kin relationships are traced. On the other hand, the numerical system has traditionally been cited as a gap in the lexica of Australian languages. § 4.2 considers some recent literature that seeks to move beyond a deficit model of traditional mathematical systems, acknowledging their original complexity and future potential. In § 4.3 we discuss the human body’s value as a model for the investigation of semantic extension, meronymy, vagueness and polysemy, as well as a ubiquitous source domain for metaphor and metonymy. In § 4.4 we consider the realm of emotion, one of the target domains most often metaphorically described in bodily terms. Finally, § 4.5 considers the culturally central domain of the land and location, which has been a particular focus of work exploring the parallelism between semantic and cognitive categories.

4.1 Kinship Reflecting its social importance, kinship is arguably the most highly elaborated semantic domain in many Australian languages (Heath et al. 1982). In most cases, both blood relatives and unrelated social contacts are integrated into a complex classificatory kinship system, in which every member of the social universe has a place. So someone speaking Kuuk Thaayorre would use the term nganin to refer to their own father, but they would also use nganin to refer to their father’s brother(s), their mother’s sisters’ husbands, their wife’s father’s sister’s husband, and so on, as well as any unrelated outsiders who are calculated to best fit this “father” category. In addition to the terms used to address kin directly, Kuuk Thaayorre (like many Australian

Semantics of Australian Languages 

 305

languages) also has a set of terms used in conjunction with a propositus term to refer to people in the third person (e.g. ‘X’s father’ as opposed to vocative terms in which the propositus may be suppressed, such as ‘hey, Dad!’), a set of kinship hand signs (for use when hunting, over long distances, or when speech is otherwise proscribed), and a set of kin terms used to refer to people bereaved of particular categories of kin (analogous to English orphan and widow, but far more extensive). Speakers of other languages have ways of dividing kin into even broader categories. These categories are commonly referred to as “skins”, and every member of a society that uses this system is assigned a “skin name” which is commonly used in place of a personal name (Stanner 1979). Skins are organised in a paradigm that determines marriageability and descent, as well as conferring certain rights and responsibilities vis-à-vis land management, language, songs, stories, totems and ceremonies (cf. Breen 2002). Thus a Pintupi woman with the skin name Napaltjarri should preferentially marry a man with the Tjakamarra skin. Their son would then have the skin Tjupurrula, and would inherit certain rights and responsibilities through the matriline (female descent line) of his mother and others through the patriline of his father. The system is cyclical, in that – given the preferred marriage – a man should always have the same skin as his son’s son, and a woman the same skin as her daughter’s daughter’s daughter’s daughter. The Pintupi skin system has eight different categories (each of which has a male and a female version of the skin name, totalling sixteen skin terms), also known as a “subsection” system. It is also common for skin systems to distinguish just four “sections”. Indeed, McConvell (1985) proposes that subsection systems were born out of contact between speakers of languages with different section systems. Still more general kin groupings are also possible. For instance, the Yolngu divide the social and natural world into two groups (the Yirritja and Dhuwa patrimoieties) in addition to their subsection system (Morphy 1977). Membership in one or the other moiety has linguistic implications as well as cultural and material ones, since there are distinct Yirritja and Dhuwa Yolngu dialects. There are other complex and unusual ways of expressing kin relationships. Nyangumarta, Bininj Gun-wok, Iwaidja, Mawng, Warlpiri, Banyjima, Pintupi, Gurindji, Wardaman and Yir Yoront, for instance, possess a class of “trirelational” kin terms (Evans 2012; Heath 1982; Laughren 1982; McConvell 1982; Garde 2003, O’Grady and Mooney 1973; Singer 2006). These terms specify how the person referred to is related to two other people, usually the speaker and addressee. Thus a Gurindji speaker would use the term kuyarri-marnany to refer to the addressee’s brother if and only if they were speaking to their own mother’s mother (McConvell and Obata 2006). If they were speaking to their sister about their brother, they would instead use the term ngarumpa-marnany (where -marnany is a special trirelational suffix meaning ‘your’). In some languages – for instance Arrernte, Martuthunira, Lardil, Wangkangurru and Noongar – kinship finds its way even into the grammatical paradigm of pronouns (cf. Hale 1966; Hercus and White 1973; Schebeck 1973; Dench 1982; Koch 1982; McConvell and Obata 2006). For example, in Lardil all pronouns referring to more

306 

 Alice Gaby and Ruth Singer

than one person must also indicate whether the people in question are separated by an odd or even number of generations. So to refer to a father and daughter in Lardil, you would need use the pronoun for two people separated by an odd number of generations (i.e. one), niinki (Ngakulmungan Kangka Leman and Hale 1997). To refer to a grandmother and grandson, you would use the pronoun for two people separated by an even number of generations (i.e. two), birri. To talk about three sisters, you would use the pronoun for three or more people separated by an even number of generations (i.e., zero), bili. Many languages have a more expansive set of “dyadic” forms (usually morphologically derived from kin terms themselves) which are used to refer to a pair of individuals while also specifying the kin relationship that obtains between them. So the Gooniyandi dyadic form garingilangi refers to a husband and wife pair, or more precisely, two people, one of whom calls the other one garingi ‘wife’ (McGregor 1996a; cf. Merlan and Heath 1982; Evans 2006). Much of the work done on the language of kinship in Aboriginal Australia has been somewhat anthropological in orientation, explicating the intricate web of relations that make up the classificatory system, including any “skewing” rules which collapse particular categories together in a systematic way (e.g. Meggitt 1962; Burling 1970; Scheffler 1978; Wafer 1981; Rumsey 1981; Dench 1982; Geytenbeek 1982; Heath 1982; Laughren 1982; Yallop 1982; Dixon 1989; McKnight 1999). There is great scope here for a finer semantic analysis of the category-internal make-up of the terms themselves. For example, in the case of nganin ‘father’ above, is this word vague across all the biological and social relations it subsumes, perhaps with a prototypical meaning of ‘biological sire’? Or are the various senses semantically distinct, related through polysemy? This kind of fine-grained semantic research remains to be fully pursued, notwithstanding some excellent work addressing such questions (e.g. Wierzbicka 1986; Dixon 1989; and Tonkinson’s 1978 exploration of the meaning of ‘father’ vis-àvis competing biological and spiritual explanations of conception). There is also call for further variationist research in the model of Morphy’s (1977) research on the moiety-based dialects of Yolngu-Matha, Smith and Johnson’s (1986, 2000) descriptions of the Kugu Nganhcara “patrilects” (dialects inherited through the male descent line), and Garde’s (2008b) study of the formation of new Bininj Gunwok patrilects through intentional language change. As the pressures of post-colonial language contact and language obsolescence often favour standardisation, it is difficult to tell just how widespread clan- or moiety-based dialectal differences might have been.

4.2 Numbers It is often claimed that Australian languages lack labels for numbers beyond three, four or five (e.g. Blake 1987). This in turn has fostered a misapprehension that traditional Aboriginal mathematics was unsophisticated or “primitive” (e.g. Trewin 1971).

Semantics of Australian Languages 

 307

However, Harris (1982, 1987) points out several flaws in such a view. Firstly, there are languages that do label higher numbers (e.g. ogripulung ‘twenty’ in Anindilyakwa), and/or possess productive quinary (base five) numerical systems (Wurm 1972: 64). Secondly, Harris argues that it is a mistake to dismiss numerical terms derived from or etymologically related to terms for body parts (typically hands, feet or fingers), or non-verbal techniques of counting and abstractly representing number (such as body-marking). Moreover, as Hale (1975: 296) points out, though “conventionalized counting systems, i.e., numerals, are for the most part lacking, […] counting itself is not lacking, in the sense that the principle of addition which underlies the activity of exact enumeration is everywhere present”. Taking a rather different tack, Cooke (1996) makes the point that the intricate kinship system of the Yolngu is akin to a “mathematics without numbers”, following Kenneth Hale’s observation that the Warlpiri’s “algebra of kinship plays an intellectual role similar to that which mathematics plays in other parts of the world” (quoted in Dixon 1980: 108). The fact remains, however, that the fit between traditional Australian languages and the current-day Australian school curriculum is seldom seamless, and many linguists and educationalists have pondered how traditional numeracy should best inform classroom teaching of mathematics (see, e.g., Sayers 1982; Stokes 1982; Watson 1988; Graham 1990; Wilkinson 2011). These papers underline the importance of understanding the traditional use of mathematical concepts, rather than the simple translation of number terms. It is likewise increasingly evident that how numerals were used – and how quantities were otherwise judged and recorded – likely represents a gap in Australianist research, more than a gap in Australian languages.

4.3 Body The human body offers a rich vein for semantic miners. As a (more-or-less) universal of human experience, it offers fieldworkers a readily-available, concrete tool for examining questions of extension (what words map onto in the real world) and intension (how the meaning of one word relates to the meanings of other words). So, for example, how do the physical boundaries of the English arm map onto those of its Wagiman translation, lari (Wilson and Harvey 2001)? The uppermost border with the shoulder varies from language to language, as does its inclusion or exclusion of the hand. Some semanticists (e.g. Schebeck 1978; Laughren 1984; Gaby 2006) have thus focused on cataloguing the inventory of body part terms, examining their respective extensions to actual portions of the body, and/or the meronymic (part:whole) relationships between them. Yet another analytical opportunity afforded by the corporal semantic field is the investigation of metaphor. Being so central to human experience of the world, the body serves as a source domain for the metaphorical description of more abstract ideas in quite possibly every language of the world (cf. Lakoff

308 

 Alice Gaby and Ruth Singer

and Johnson 1999). To give just a few examples, body part terms are invoked in: both verbal reference to and manual signs for kin (e.g. pointing to the shin to refer to a sibling in many parts of Australia, cf. Kendon 1988); expressions of cognition (particularly the ear, cf. §  3.1); labelling features of the landscape (e.g. lari ‘arm’ above also being used to refer to creeks and roots); or even unexpected mappings like ‘eye’ > ‘no good’, which nevertheless recur in a number of Australian languages (Sommer 1978). Body part terms also occupy a privileged position in the grammatical systems of many Australian languages (cf. Walsh 1996; McGregor 1996b; Hosokawa 1996; McKay 1996; Evans 1996; Leeding 1996; Harvey 1996). For instance, they are frequently among the only nouns that may be incorporated into verbs (e.g. in Kayardild, Evans 1996), or may appear as an apparently transitive direct object in otherwise intransitive reflexive/reciprocal clauses (as in (3)). (3)

nganhi mara rdama-thadi-rna warra-yi 1sg.s hand.abs cut-refl-ptcp aux-prs ‘I cut my hand’. (Diyari: Austin 1981: 152)

4.4 Emotions The meanings of emotion terms vary widely from language to language. Though their intangibility makes them difficult to study, emotion terms provide key insight into the culture of their speakers. A common pattern found in Australian languages is to use a single term for both ‘fear’ and ‘shame’. Hiatt (1978) first documented this polysemy in Gidjingali (a Burarra dialect) and Warlpiri.⁷ The two emotions ‘shame’ and ‘fear’ have in common that they create a desire to remove oneself from the situation at hand. Hiatt observes the Gidjingali verb -gurakadj being used for situations of fear, situations that involve shame or embarrassment, as well as situations that involve both. For example, to behave inappropriately towards one’s mother-in-law would invoke shame and embarrassment but also often fear of the repercussions of breaking traditional rules. Thus the meaning of emotion terms does not simply relate to how people feel and how they respond to different situations but also how people are expected to feel and respond to different situations from a cultural perspective. White teachers often describe Aboriginal students as “shy” because they seem uncomfortable when singled out in class, irrespective of whether it is to be admonished or praised. Harkins (1990) analyses the meaning of the concept of shame in Aboriginal culture with a view to improving intercultural understanding in the classroom. She argues that shame is not simply a negative emotion that people feel after

7 See also Myers (1986) on shame.

Semantics of Australian Languages 

 309

doing something wrong, but is also a positive concept guiding proper behaviour, rooted in the egalitarian ethos of Aboriginal society. She writes that the concept of shame is a key to understanding how Aboriginal students approach classroom behaviour (cf. Eades, this volume, for further discussion of the cultural underpinnings of “shame” and the pragmatics of silence). Also in a practical vein, Morice (1978) surveys some of the main Pintupi terms for emotions, with a view to facilitating better diagnoses of psychiatric conditions in Aboriginal communities. Like shame, anger is an emotion that comes from an evaluation of social relations. Anger contrasts with shame in that it is a feeling that others have not behaved as they ought to. Myers (1988) argues that anger in Pintupi culture is best understood in relation to the importance of being compassionate, ngaltutjarra, towards one’s kin. When somebody is not compassionate towards a family member, that person can feel rejected and become angry. Similarly, when a person feels angry, they lack compassion towards the object of that anger. Drawing on discussions with Dalabon speakers, Ponsonnet (2010) also gives culturally specific accounts of Dalabon emotions. But she contests a number of Myers’ claims, including his idea that there is always a clear distinction between private feelings and publicly expressed emotions. Goddard (1991) argues that although situations in which people become angry differ from culture to culture, the core meaning of emotion terms can be described in basic, universal terms. Drawing on the theory of Natural Semantic Metalanguage, Goddard and a growing number of researchers have striven to capture both the universal and the culturally particular aspects of how the experience of emotion is lexicalised by explicating emotion words in terms of a restricted set of semantic primes. Australian languages so analysed include Arrernte (Harkins 2001), Yankunytjatjara (Goddard 1990, 1991), Burarra (Wierzbicka 1999) and Kayardild (Evans 1994; Wierzbicka 1999). Other researchers, meanwhile, have observed that emotions are often metaphorically described in terms of the parts of the body where they are felt to be experienced, often the belly or the throat (cf. Turpin 2002; Gaby 2008; Walsh 1996, § 4.3).

4.5 The land and spatial relations The land looms large in Australian Indigenous cultures, and the interconnectedness of people, language and the land is a leitmotif that runs through much of the anthropological and linguistic literature. Stories abound of ancestral (or “Dreamtime”) beings who created the landscape, imbuing it with the languages that would come to be spoken thereon (Merlan 1981; Rigsby and Sutton 1982; Rumsey 1993; but cf. Bowern 2009: 335 for a counter-example). In the present era, then, people are linked to the languages they speak through the tracts they occupy and/or hold ritual obligations toward. Given the cultural primacy of land and people’s interaction with it, it is hardly surprising that this has also been an area of great interest to researchers.

310 

 Alice Gaby and Ruth Singer

Placenames have been a prominent focus, with two recent books assembling papers that detail: (i) the hierarchical relationships between names of different sized tracts of land; (ii) the etymology and spiritual meaning behind traditional placenames; and (iii) how those placenames have been appropriated, interpreted and restored in post-colonial Australia (cf. papers in Hercus et al. 2002; Giacon 2004; Furphy 2001, 2002; and papers in Koch and Hercus 2009). Placenames are of course an invaluable navigational aid, especially in traditionally nomadic and/or hunter-gatherer societies. But in addition to the lexicon of named places, all societies employ some number of more general terms that allow them to navigate the surrounding terrain and to locate the relative position of people, places and things in space. Speakers of Australian languages are renowned for employing the same geographically-based, “absolute” directional terms (often translating to English north, south, east and west) for both large-scale navigation (e.g. Sydney is to the north of Melbourne) and to locate objects on a small scale (e.g. the scissors are in the eastern drawer), whereas for the latter case most English-speakers would use viewer-based “relative” terms (e.g. the scissors on the shelf to your left) or reference object-based “intrinsic” terms (e.g. the scissors are in the top drawer of the bureau). Most languages, like English, furnish their speakers with a choice of absolute, relative and/or intrinsic spatial terms, but one system tends to predominate over the others in everyday use. Proponents of linguistic relativity (who argue that language influences thought) have amassed a wealth of experimental data that demonstrate broader cognitive consequences of speaking a language with a predominately relative/egocentric spatial reference system, as opposed to an absolute/geocentric one (cf. Pedersen et al. 1998; Levinson 2003; Levinson and Wilkins 2006). Australian Aboriginal languages by and large privilege the absolute frame of reference, and have been instrumental in developing the typology of spatial reference. Absolute spatial reference systems include both directional systems based on compass points (usually anchored by the sun’s trajectory and a perpendicular axis, as with north, south, east and west) and the similarly geo-centric riverine systems (e.g. Jaminjung manamba ‘upstream’, buya ‘downstream’, thangga ‘up’, thamirri ‘down’: Schultze-Berndt 2006: 67). Though such terms may also include relative frame of reference information (e.g. referring to a location ‘nearby’ or ‘far away’) and/or intrinsic frame of reference information (e.g. referring to a location ‘inside’ or ‘below’ some reference point), the organising principle of the system is usually clearly absolute. Speakers of these languages are renowned for their “dead reckoning” abilities in navigation, a fact often credited to the language they speak. Indeed, people who speak languages with a dominant absolute frame of spatial reference have been shown to draw on this absolute system in both experimental memory tasks and spontaneous co-speech gesture (Haviland 1993, Levinson 1997, 2003). Though there are Australian languages that appear not to privilege an absolute frame of reference (e.g. Murrinh-Patha), there is very little research detailing the use and cognitive consequences of these systems. This is an obvious avenue for future research.

Semantics of Australian Languages 

 311

Like the body, physical space is commonly drawn on as a concrete source domain for more abstract metaphorical targets. We might therefore expect differences in the ways in which target domains such as time are conceptualised by speakers of languages with dominant absolute and relative spatial systems respectively. Though research in this domain is still in its infancy, initial results would seem to bear out this expectation. For instance, Boroditsky and Gaby (2010) find Kuuk Thaayorre speakers represent time as flowing from east-to-west when asked to arrange photo cards depicting stages in a temporal sequence (e.g. a chick hatching from an egg), or when drawing in the sand to represent sequentially-related points in time (cf. Gaby 2012).

5 The semantics and pragmatics of grammar This section looks at three examples of the insights that semantics and pragmatics can provide in analysing aspects of morphosyntax. The two following subsections look at sublexical semantics: case systems and verbal inflections. It is worth noting, however, that not all semantic phenomena respect a neat division of syntax, morphology and the lexicon. Take for example the semantic notion of reciprocity, where two or more individuals direct their actions towards one another. If we begin with a functional – rather than formal – definition of the domain of investigation, we find quite different structures encoding a more or less similar semantic range. For instance, Australian languages are found to mark reciprocity by means of verbal morphology (e.g. Kuuk Thaayorre: Gaby 2011; Murrinh-Patha: Nordlinger 2011), special auxiliary (e.g. some Nyulnyulan languages: McGregor 2000) or a clause-level construction that is transparently a contraction of a biclausal construction (e.g. Mawng: Singer 2011; and Iwaidja: Majid et al. 2011). Another area that is not neatly divided between verbal morphology and syntactic constructions is quantification. Australian languages may express quantification through verbal morphology or through free quantifiers (Bittner and Hale 1995; Evans 1995). There is great scope for further research on the meaning components of morphological and syntactic processes, such as Wilkins’ (1984) tantalising study of the semantics and pragmatics of reduplication, Austin’s (1982) paper on cognate objects in Australian languages and Evans’ (1997) paper on whether it is the semantic role or ontological properties of arguments that determine whether they can be incorporated into verbs.

5.1 Case systems In addition to marking core syntactic functions such as “object”, case systems can have a range of semantic and pragmatic functions. For example, ergative case marking – prototypically used to mark the subject of a transitive clause – is optional in a

312 

 Alice Gaby and Ruth Singer

range of Australian languages. In these languages, not all transitive subject NPs carry ergative case marking and/or some subjects of intransitive clauses do (McGregor 2010 see also Nordlinger, this volume). Optional ergativity may be sensitive to semantic factors such as animacy and pragmatic factors such as topicality (Verstraete 2010). The non-syntactic functions of case are sometimes considered as evidence of grammaticalisation in progress. But, as the level of detail in descriptions of Australian languages improves, it may turn out to be the norm for cases to combine a range of syntactic, semantic and pragmatic functions. There may be multiple pathways leading to this polyfunctionality in ergative case markers. Gaby (2010) suggests that a pragmatic focalising function predates the syntactic function of some Kuuk Thaayorre ergative markers, while Pensalfini (1999) and Meakins and O’Shannessy (2010) report on grammaticalisation in the opposite direction: ergative case markers which have recently developed pragmatic functions (see also Meakins, this volume, for a more detailed exploration of case-marking systems being restructured under the pressure of language contact). These examples coming from distant parts of the continent illustrate that optional ergative marking is not restricted to any one area or family and is not necessarily the result of contact between traditional Australian languages and English. The prevalence of case among Australian languages, particularly among the Pama-Nyungan languages and their neighbours, has led to many unusual developments, not frequently attested elsewhere. See Nordlinger (this volume) for a discussion of the re-purposing of case morphology to mark different types of subordinate clause.

5.2 Verbal inflections Verbal inflections are found in most Australian languages and they can be relatively complex, particularly in non-Pama-Nyungan languages. This section focuses on the expression of tense, aspect and mood (TAM) in the verb. The meaning of TAM inflections can be notoriously difficult for non-native speaker linguists to pin down. Granites and Laughren (2001) discuss the late Ken Hale’s technique for eliciting semantic descriptions from native speakers. Their examples attest to the complex semantics that guides native speakers’ usage of Warlpiri verbal morphology. One way to deal with the insecurity of being a non-native speaker working on such subtle semantics is to look at what is common across a number of languages. Taking this path, Verstraete (2005, 2006) applies his substantial insight into the semantics and pragmatics of mood to the task of understanding the expression of mood in non-Pama-Nyungan languages. He describes mood marking in non-Pama-Nyungan languages as “composite” because it is spread between verbal prefixes and suffixes such that it is the combination of prefix and suffix morphology that is associated with particular mood meanings. For example in a sentence expressing a meaning such as

Semantics of Australian Languages 

 313

“They wanted to get the fire back.” as in (4) from Bunuba, the category “irrealis” is marked in the verb prefix and past tense is marked in the verb suffix.⁸ (4)

winthali baburru bada iywirrunugu fire below seize irr.3s