Keeping Languages Alive : Documentation, Pedagogy and Revitalization 9781107703711, 9781107029064

Citation preview

Keeping Languages Alive

Many of the world’s languages have diminishing numbers of speakers and are in danger of falling silent. Around the globe, a large body of linguists are collaborating with members of indigenous communities to keep these languages alive. Mindful that their work will be used by future speech communities to learn, teach, and revitalize their languages, scholars face new challenges in the way they gather materials and in the way they present their findings. This volume discusses current efforts to record, collect, and archive endangered languages in traditional and new media that will support future language learners and speakers. Chapters are written by academics working in the field of language endangerment and also by indigenous people working ‘at the coalface’ of language support and maintenance. Keeping Languages Alive is a must-read for researchers in language documentation, language typology, and linguistic anthropology. m a r i c . j o n e s is Reader in French Linguistics and Language Change at the University of Cambridge and Fellow in Modern and Medieval Languages at Peterhouse, Cambridge. A highly experienced fieldworker, she has published extensively on language obsolescence and revitalization in relation to Insular and Continental Norman, Welsh, and Breton. Her publications include Language Obsolescence and Revitalization (1998); Jersey Norman French (2001); The Guernsey Norman French Translations of Thomas Martin (2008). s a r a h o g i l v i e works at Amazon Kindle on languages, dictionaries, and content. Previously she was Reader in Linguistics at the Australian National University, and was Alice Tong Sze Research Fellow at Lucy Cavendish College, Cambridge. She lived and worked with an Australian Aboriginal community to write a grammar and dictionary of their language, and her current research focuses on how innovative technologies can help maintain and revitalize endangered languages. Her publications include Words of the World: A Global History of the OED (Cambridge University Press 2013).

Keeping Languages Alive Documentation, pedagogy, and revitalization Edited by

Mari C. Jones and Sarah Ogilvie

University Printing House, Cambridge CB2 8BS, United Kingdom Published in the United States of America by Cambridge University Press, New York Cambridge University Press is part of the University of Cambridge. It furthers the University’s mission by disseminating knowledge in the pursuit of education, learning and research at the highest international levels of excellence. www.cambridge.org Information on this title: www.cambridge.org/9781107029064  C Cambridge University Press 2013

This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 2013 A catalogue record for this publication is available from the British Library Library of Congress Cataloguing in Publication data Jones, Mari C., editor of compilation. Keeping languages alive : documentation, pedagogy and revitalization / Mari C. Jones and Sarah Ogilvie. pages cm. Includes bibliographical references and index. ISBN 978-1-107-02906-4 (hardback) 1. Endangered languages. 2. Language obsolescence. 3. Language maintenance. 4. Typology (Linguistics) 5. Anthropological linguistics. I. Ogilvie, Sarah, editor of compilation. II. Title. P40.5.E53K44 2014 306.44–dc23 2013023078 ISBN 978-1-107-02906-4 Hardback Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party internet websites referred to in this publication, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.

Contents

List of figures List of tables List of contributors Preface

page vii viii ix xiii

I Documentation 1 Language documentation and meta-documentation peter k. austin 2 A psycholinguistic assessment of language change in eastern Indonesia: Evidence from the HALA project amanda hamilton, jawee perla, and laura c. robinson

3

16

3 Documentation of endangered sign languages: The case of Mardin Sign Language ulrike zeshan and hasan dikyuva

29

4 Re-imagining documentary linguistics as a revitalization-driven practice david n athan and meili fang

42

5 Language documentation and community interests john henderson 6 American Indian Sign Language documentary linguistic fieldwork and digital archive jeffrey e. davis 7 Purism in language documentation and description michael riessler and elena karvovskaya

56

69 83

v

vi

Contents

8 Greek-speaking enclaves in Pontus today: The documentation and revitalization of Romeyka ioanna sitaridou

98

II Pedagogy 9 New technologies and pedagogy in language revitalization: The case of Te Reo M¯aori in Aotearoa/New Zealand tania m. ka‘ai, john c. moorf ield, and muiris o´ laoire

115

10

Teaching an endangered language in virtual reality hanna outakoski

128

11

A nomadic school in Siberia among Evenk reindeer herders alexandra lavrillier

140

12

Task-based language teaching practices that support Salish language revitalization arieh (ari) sherris, tachini pete, lynn e. thompson, and erin flynn haynes

155

III Revitalization 13

14

15

16

Speakers and language revitalization: A case study of Guern´esiais (Guernsey) yan marquis and julia sallabank

169

On the revitalization of a ‘treasure language’: The Rama Language Project of Nicaragua colette grinevald and b e´ n e´ dicte pivot

181

Whistled languages: Including Greek in the continuum of endangerment situations and revitalization strategies maria kouneli, julien meyer, and andrew nevins

198

What is language revitalization really about? Competing language revitalization movements in Provence james costa and m e´ d e´ ric gasquet-cyrus

212

Bibliography Index

225 249

Figures

2.1 The thirty-one test items. Asterisks indicate items removed in the Agreement subgroup (see Section 5) page 18 2.2 Per cent accurate responses in Adang and Indonesian by age group 22 2.3 Mean RT in Adang and Indonesian by age group 24 3.1 Family tree diagram produced by consultants during a workshop 36 3.2 Revised family tree diagram created by researchers 37 3.3 The physical environment in Mardin 38 3.4 Screenshots taken during the map activity 39 4.1 Eli Timan’s collection on Jewish Iraqi 47 4.2 Ngarrindjeri learner wordlist with words known by elders marked 51 6.1 Map of cultural areas of Native North America 81 7.1 The traditional Saami settlement area on the Kola peninsula and adjacent parts to Fennoscandia 84 8.1 The three remaining Greek-speaking enclaves of Pontus: Of (C ¸ aykara), S¨urmene, and Tonya 99 8.2 A taxonomy of Asia Minor Greek 101 9.1 The Te Whanake homepage displaying its seven separate websites, all of which are bilingual and free to access 121 9.2 Part of a results page taken from the online dictionary 122 123 9.3 The iPhoneTM dictionary app 9.4 The Te Whanake animations homepage 124 9.5 The Te Whanake TV homepage 125 9.6 The homepage of the T¯amata Toiere website 127 10.1 A North Saami lesson through the medium of Second Life (SL) 129 10.2 At the first station the students work in pairs and are trained in conversation (with manual) and vocabulary 136 10.3 Teachers’ meetings in Second Life (SL) 138 12.1 Salish SOPA results 2009–10 162 12.2 Salish ‘spot-the-difference’ information-gap task 165 15.1 Mrs K, a ‘very good whistler’ in Antia 199 15.2 Spoken (top) and whistled (bottom) equivalents of the Greek word ema ‘blood’ 201 vii

Tables

2.1 2.2 3.1 4.1 6.1 7.1 8.1 12.1 12.2

viii

Testing schedule page 19 Accuracy codes 21 Linguistic backgrounds of the research team 34 Overview annotations 49 Documentary linguistic fieldwork: Native signers identified and filmed (2009–12) 77 Cognate sets of borrowed exclusive focus particles in different Saami languages 92 The typological status of Romeyka within Asia Minor Greek varieties 100 Rankings for three of the six Salish imperative forms 161 Junior novice sublevels for oral fluency, Salish Rating Rubric, 2011 161

Contributors

peter k. austin SOAS, University of London, UK james costa ´ Ecole Normale Sup´erieure, Lyons, France jeffrey e. davis University of Tennessee, USA hasan dikyuva University of Central Lancashire, UK meili fang Ochanomizu University, Japan m e´ d e´ ric gasquet-cyrus Universit´e Aix-Marseille, France colette grinevald Universit´e Lyon 2, France amanda hamilton Wangka Maya Pilbara Aboriginal Language Centre, Australia erin flynn haynes American Institutes for Research, Waltham, MA, USA john henderson University of Western Australia tania m. ka‘ai Auckland University of Technology, New Zealand elena karvovskaya University of Potsdam, Germany

ix

x

List of contributors

maria kouneli Yale University, USA alexandra lavrillier Universit´e de Versailles St-Quentin-en-Yvelines, France yan marquis Guernsey, Channel Islands julien meyer Museu Goeldi, Belem, Brazil john c. moorfield Auckland University of Technology, New Zealand david nathan SOAS, University of London, UK andrew nevins University College London, UK muiris o´ laoire Institute of Technology, Tralee, Ireland hanna outakoski Ume˚a University, Sweden jawee perla American University, Washington DC, USA tachini pete Flathead Reservation, Arlee, Montana, USA b e´ n e´ dicte pivot Universit´e Lyon 2, France michael riessler University of Freiburg, Germany laura c. robinson University of California, Santa Barbara, USA julia sallabank SOAS, University of London, UK arieh (ari) sherris Teachers College, Columbia University, USA

List of contributors

ioanna sitaridou Queens’ College, Cambridge, UK lynn e. thompson Center for Applied Linguistics, Washington DC, USA ulrike zeshan University of Central Lancashire, UK

xi

Preface

Increasing numbers of the world’s languages have ever fewer speakers and are in danger of falling silent. To tackle this problem, scholars are collaborating with members of indigenous communities all around the globe to document and describe these endangered languages and cultures so that future speech communities may learn, teach, and revitalize their languages. Documentation is often considered an essential part of the work of any field linguist. It is self-evident that, if teaching (whether in schools or within the wider community) is a key component in keeping a language alive, then for this to occur some form of record of that language needs to exist to provide a basis for that teaching. Because of its associations with standardization, written documentation can provide much-needed linguistic bolstering, as a language becomes codified in dictionaries and grammars, both of which serve to facilitate the production of pedagogical material by forming the linguistic basis for revitalization efforts. Over and above this, however, documentation carries with it a vital symbolic force. It serves important needs and functions in support of language maintenance by raising the status of endangered languages and fostering a sense of unified identity (especially in areas where a dispersed population or dialectal fragmentation have not been conducive to this). It can even provide a partial means of repackaging a hitherto stigmatized identity, as speakers see their ancestral language being used in modern domains. Documentation clearly represents a critical first step in any process of language revitalization, and efforts to maintain a language may well succeed or fail on the basis of the quality and range of material gathered. And yet the all too familiar clock that is ticking loudly for many endangered languages means that the process of documentation may necessarily be rapid and dependent on sometimes arbitrary decisions by linguists – who may be analysing a language for the first time. This volume aims to explore the three themes of language documentation, pedagogy, and revitalization and how they interface in different communities to help keep languages alive. As we see in the various chapters of this book, given the world’s rich linguistic diversity a ‘one size fits all’ approach is simply not viable – for example, keeping whistled languages or sign languages alive represents a completely different challenge from that of revitalizing a spoken xiii

xiv

Preface

tongue. Moreover, the sociopolitical and/or cultural context in which these processes take place can either impose particular constraints or bring increasing freedoms for language planners. The twenty-first century has also brought a whole new tool-kit to the linguist’s workbench in the form of new technologies with the ability to transform the documentation of endangered languages and also their dissemination within a community. Digital and audiovisual archives are increasingly commonplace, and avatar teachers and mobile phone apps are becoming ever more prominent. Such technology has had a considerable impact upon the ways in which documentation, pedagogy, and revitalization interact with and complement each other. The volume discusses current efforts to record, collect, and archive endangered languages in both traditional and new media, and assesses the different kinds of support that these can offer future language learners and speakers. Keeping languages alive ultimately depends on languages being spoken by children in the home and especially the kitchen, but getting there often requires a combination of quite different approaches – frequently involving collaboration between academics working in the field of language endangerment and members of indigenous communities who are working ‘at the coalface’ of language support and maintenance. The volume accordingly gives voice to both types of contribution. The data and case studies presented here are drawn from many different languages that are currently being kept alive across four of the world’s continents. The sixteen chapters are divided into three Parts. Part I addresses both theoretical and practical issues relating to language documentation and illustrates them with material from: Australian Aboriginal languages; Adang (spoken in Indonesia); the Tungusic languages of Siberia; Kildin Saami; Romeyka; and sign languages in Mexico, Bali, Turkey, and North America. Part II, on pedagogy, focuses on methodologies and practices in Aboriginal languages of eastern Australia, Montana Salish in North America, Maori in New Zealand, and North Saami in Scandinavia. Part III examines different aspects of language revitalization and draws on material from current projects in France, Nicaragua, Greece, Turkey, the Channel Islands, and Spain.

Part I

Documentation

1

Language documentation and meta-documentation Peter K. Austin

1

Introduction

The past fifteen years have seen the emergence of a new sub-field of linguistics that has been termed ‘language documentation’ or ‘documentary linguistics’ (Himmelmann 1998, 2002, 2006, Lehmann 2001, P. Austin 2010a, Grenoble 2010, Woodbury 2003, 2011a). Its major goal is the creation of lasting multipurpose records of languages or linguistic practices through audio and video recording of speakers and signers, and annotation, translation, preservation, and distribution of the resulting materials. It is by its nature multi-disciplinary and draws on theoretical concepts and methods from linguistics, ethnography, folklore studies, psychology, information and library science, archiving and museum studies, digital humanities, media and recording arts, pedagogy, ethics, and other research areas. The term ‘language documentation’ historically has been used in linguistics to refer to the creation of grammars, dictionaries, and text collections for undescribed languages (the so-called ‘Boasian trilogy’; for discussion see Woodbury 2011a: 163). However, work defining language documentation as a distinct sub-field of linguistics emerged around 1995 as a response to the crisis facing the world’s endangered languages, about half of which might disappear in the twenty-first century (the crisis was identified and popularized in such publications as Robins and Uhlenbeck 1991, Hale et al. 1992, Wurm 2001). Linguists drew attention to an urgent need to record and analyse language materials and speakers’ linguistic knowledge while these languages (or threatened special registers and varieties within them) continued to be spoken, and to work with communities on supporting threatened languages before

Versions of this paper have been presented as talks at the Kioloa Aboriginal Languages Workshop 2010, the Linguistic Society of America Annual Meeting 2011, the International Conference on Language Documentation and Conservation 2011, and the seventh European Australianists Workshop 2012, as well as in classes at SOAS and Tokyo University of Foreign Studies. I am grateful to Lisa Conathan, Lise Dobrin, Andrew Garrett, Geoff Good, Heidi Johnson, Anthony Jukes, Stuart McGill, David Nathan, Julia Sallabank, and Tony Woodbury for discussion of the ideas included; I alone am responsible for the material presented here.

3

4

Peter K. Austin

opportunities to do so became reduced. The emergence of language documentation was also prompted by developments in information, media, communication, and archiving technologies which make possible the collection, analysis, preservation, and dissemination of documentary records in ways which were not feasible previously. In addition, it was facilitated by large levels of research funding support from three main sources: the DoBeS (Dokumentation Bedrohter Sprachen ‘Documentation of Endangered Languages’) programme sponsored by the Volkswagen Foundation in Germany (2000–13), the Endangered Languages Documentation Project (ELDP) supported by the Arcadia Trust in the United Kingdom (2002–16), and the Documenting Endangered Languages (DEL) interagency initiative of the United States National Science Foundation and the National Endowment of the Humanities (2005 onwards). Language documentation concerns itself with principles and methods for the recording and analysis of primary language and cultural materials, and metadata about them, in ways that are transparent and accountable, and that can be archived and disseminated for current and future generations to use. Some researchers have emphasized standardization of data/metadata and analysis and ‘best practices’ (e.g. E-MELD, OLAC), while others have argued for a diversity of approaches which recognize the unique and particular social, cultural, and linguistic contexts within which individual languages are used (sele Dobrin, Austin, and Nathan 2009, Dobrin and Berson 2011). This chapter is concerned with the role of metadata in language documentation and argues for a broad approach to observation and documentation of the methods, processes, and outcomes of language documentation projects, which we refer to as ‘meta-documentation’ (or ‘meta-documentary linguistics’). It argues that a theory of meta-documentation does not (yet) exist and discusses some techniques that could be adopted for developing such a theory, as well as proposing some of the components that may make it up. The need for fuller reflexivity on the part of language documenters, and linguists more generally, is particularly emphasized. 2

Language documentation (or documentary linguistics)

Language documentation (or ‘documentary linguistics’) is defined by Himmelmann (2006: v) as ‘concerned with the methods, tools, and theoretical underpinnings for compiling a representative and lasting multipurpose record of a natural language or one of its varieties’. A similar definition is given by Woodbury (2011a: 159) as ‘the creation, annotation, preservation, and dissemination of transparent records of a language’. Woodbury (2011a: 161) also notes that the term ‘language documentation’ is used in another sense, as well namely the outcomes of documenting languages, but proposes to clarify the terminology: ‘The sets of records, coherent or not, are often called language

Language documentation and meta-documentation

5

documentations; but since that is what we are calling the activity as a whole, I will call such sets language documentary corpora (or just corpora).’ The form and uses of documentary corpora have been explored somewhat by linguists (e.g. within the DoBeS programme – see DoBeS 2005) and, as Dobrin and Berson (2011: 188) argue: It does seem clear that documentary linguists have been on relatively comfortable ground in thinking about the products of linguistic research: conceptually distinguishing an annotated corpus or documentation of a language from a higher order description of its patterning . . . reasserting the intellectual value of vocabulary . . . and oral discourse (as represented in texts) alongside grammar, extending the range of documentary outputs to include items like primers and orthographies that are targeted directly at non-academic audiences . . . They have also enriched the inventory of digital data models, formats, and software tools that facilitate documentary research and enable the preservation and dissemination of its results.

There has been rather less discussion about what Woodbury (2011a: 161) calls ‘corpus theorization’: ‘I will call the ideas according to which a corpus is said to cohere or “add up” its (corpus) theorization. Corpus theorizations, and even principles for corpus theorization, can both offer a space for invention and become a matter of contention and debate.’ Corpus theorization has been only weakly developed within documentary linguistics. Seifart (2008) attempts to address some aspects of corpus theorization, namely representativeness and sampling, and L¨upke (2010) discusses data collection methods, but few scholars have been explicit about why and how they are collecting and organizing their particular corpora, other than for some vague notion of ‘documenting the language’ or ‘saving the data’. In addition to corpus theorization, Woodbury (2011a: 161) also mentions wider issues of what he calls ‘project design’: Of special interest is the range of concerted, programmed documentary activities motivated by impending language loss and aimed at creating a final record. These activities . . . raise questions about the participants, their purposes, and the various stakeholders in the activity or program of activity or project: we may refer to this set of questions as the project design . . . of a language documentation activity.

We see corpus theorization and project design as part of meta-documentation, which we explore and elaborate in Section 3. For some speculations about a possible typology of project designs see Section 4. 3

Meta-documentation (or meta-documentary linguistics)

From the outset, those who have been working on language documentation have been clear that alongside collecting and analysing data (typically audio

6

Peter K. Austin

and video recordings, but also still images)1 it is necessary to record and analyse metadata, data about the data, to ensure that its context, meaning and use can be properly determined. As Nathan (2010b: 196) states: ‘[M]etadata is the additional information about data that enables the management, identification, retrieval and understanding of that data. The metadata should explain not only the provenance of the data (e.g. names and details of people recorded), but also the methods used in collecting and representing it.’ Notice that metadata is required not only for archiving but also for the very management, identification, retrieval, and understanding of the data within the documentation project once processing and value-adding is to be done. The way files are named and structured in folders is itself a type of metadata (see Nathan 2010b), and as Nathan and Austin (2004) argue, any knowledge added to the recordings (including transcription, translation, annotation, summary, index etc.) should be seen as ‘thick metadata’ (contrasted with the ‘thin’ cataloguing metadata often promoted in discussions of language documentation, e.g. by the E-MELD ‘School of Best Practices in Language Documentation’). Nathan (2010b: 196) also proposes that: ‘[A]nother way to think of metadata is as meta-documentation, the documentation of your data itself, and the conditions (linguistic, social, physical, technical, historical, biographical) under which it was produced. Such meta-documentation should be as rich and appropriate as the documentary materials themselves.’ It is my contention that alongside language documentation we need to develop a theory (and related practices) of language meta-documentation (or Meta-documentary Linguistics), the focus of which would be (to adapt the definition of Himmelmann 2006) the methods, tools, and theoretical underpinnings for setting up, carrying out, and concluding a documentary linguistics research project. It would be the documentation of the documentation research itself. Some work on particular issues that are relevant here has been published in the language documentation literature, especially in relation to research ethics (Grinevald 2003, Dwyer 2006, Rice 2006, Thieberger and Musgrave 2006, Macri 2010), reciprocity and exchange (Yamada 2007, CzaykowskaHiggins 2009, Glenn 2009, Guerin and Lacrampe 2010, Leonard and Haynes 2010, Crippen and Robinson 2013), and researcher and community motivations (Dobrin 2008), which are part of what Dobrin and Berson (2011: 189) call ‘the social processes set in motion by . . . research, from the conceptualization of fieldwork to the dissemination of its products’; however, no wider approach

1

The documentary linguistics literature pays little attention to the role of still images in corpus creation. However, evidence from materials archived by documenters, e.g. at ELAR, suggests that they take large numbers of photographs and scans for a range of purposes (including documenting their field sites, recording setup, ceremonies, objects, and consultants and other people, and for copying fieldnotes and other documents, etc.).

Language documentation and meta-documentation

7

or theorization has been undertaken. There are several reasons it would be valuable to do so: r to develop good ways of presenting and using language documentations (what Woodbury 2011b calls ‘making language documentations people can read, use, understand and admire’) r for future preservation of the outcomes of current documentation projects r to assist with sustainability of the field of language documentation in terms of ensuring continuity of projects, people, and products r helping future researchers learn from the successes and failed experiments of those currently grappling with issues in language documentation (see Gawne 2012 and comments by James Crippen) r to document intellectual property contributions to projects, including those of community members, researchers, and others, along with their career trajectories, especially for more junior researchers (Conathan 2011a). There are at least three possible directions that could be explored to strive towards a theory of meta-documentation: 1. deductive approaches: the postulation of axioms and theorems 2. inductive approaches: examination of current and past documentations (socalled ‘legacy materials’) to analyse practices and identify operating principles (as well as lacunae) 3. comparative approaches: examination of what other relevant and related fields have done in their meta-documentation to see what is applicable, and what not, to language documentation. We discuss each of these in turn. 3.1

Deductive approaches

Since its establishment in the late 1990s language documentation has been dominated by declarative deductive approaches to recommendations for creating metadata which have been primarily influenced by library concepts (e.g. Dublin Core). Key metadata notions have been interoperability, standardization, discovery, and access (OLAC,2 E-MELD, Good 2002, Farrar and Lewis 2007). However, the wider goals of language documentation (including the wider social goals relating to speaker community involvement) mean this is not powerful enough and we need, as P. Austin (2010a: 29) argues, to ‘extend the concept of meta-documentation to include as full as possible documentation of the documentation project itself’. It appears that meta-documentation of at least the following aspects should be covered: r the identity of the stakeholders and their roles in the project beyond the, so far, restricted concern to document people and roles such as ‘speaker’ and 2

See the Appendix to this chapter for a listing of OLAC metadata terms and their definitions.

8

Peter K. Austin

‘recorder’ (for a fuller but still incomplete listing see the OLAC roles given in Conathan 2011b: 245). For many projects other people, organizations, and institutions play a crucial role, e.g. funders, gate-keepers, validators of the research etc., but they and their roles tend to be neglected in metadata creation. r the attitudes of language consultants, both towards their languages and towards the documentation project. These can of course change and develop over time and have a vital impact on the success or failure of a project, as well as the nature of the materials which can be collected and disseminated. r the methodology of the researcher, including research methods and tools (see L¨upke 2010), and any theoretical assumptions encoded through abbreviations or glosses, as well as relationships with the consultants and the community (Good 2010 mentions what he called ‘the 4 Cs’: ‘contact, consent, compensation, culture’)3 r the biography and history of the project,4 including the background knowledge and experience of the researcher and the main consultants5 (e.g. how much fieldwork the researcher had done at the beginning of the project and under what conditions, what training the researcher and consultants had received), and how the project emerged and developed. For a funded project, the project biography would include the original grant application and any amendments, reports to the funder, e-mail communications with the funder, and/or any discussions with an archive, such as the reviews of sample data mentioned by Nathan (2010b). Both successful and unsuccessful aspects of the project biography should be included. r any agreements entered into, whether formal or informal (such as a Memorandum of Understanding, payment arrangements, and any promises and expectations issued to stakeholders). This kind of information is invaluable not only for the researcher and others involved in a project but also for any other future parties wishing to make sense of the project and its history and context. Unfortunately, linguists have typically been poor at recording and encoding this kind of information, meaning that work is often difficult with so-called ‘legacy data’, especially materials that only become available once the researcher has died (see Subsection 3.2 and Bowern 2003, P. Austin 2010b, Innes 2010, O’Meara and Good 2010). This is an area for further development and experimentation within language documentation theory and practice. 3 4

5

This seems to correspond to Woodbury’s (2011a, b) ‘corpus theorization’. Note that OLAC (see the Appendix to this chapter) allows a date specification in the metadata for individual resources but is vague about the significance of such dates, defining it merely as ‘a date associated with an event in the life cycle of the resource’. Conathan (2011b: 248) mentions biographical information about project participants but not the historical biography of the research project itself.

Language documentation and meta-documentation

3.2

9

Inductive approaches

An inductive approach to meta-documentation would involve exploring current and past practices of language documenters to see what types of metadata they collect and notate within their projects. Here we report on two examples of such an approach: (1) a review of metadata practices in the Endangered Languages Archive (ELAR) at the School of Oriental and African Studies (SOAS) carried out by Nathan (2011), and (2) the main points from P. Austin (2010b), which look at the challenges of working with Australian Aboriginal legacy materials. Nathan (2011) is a survey of metadata practices in forty-nine deposits in ELAR. He found that about 80 per cent of the most frequently occurring categories can be mapped to OLAC labels (see the Appendix to this chapter). However, depositors added richer specifications of other kinds of metadata information, including such things as parents’ and spouse’s mother tongues, speaker education levels, workflow status of materials, and terms in the researched languages (such as song titles or place names), or in other locally significant languages. Across the deposits examined some of these terms appeared frequently (e.g. 1 occurred in 20 of the 49 deposits); however, there were 613 terms which were unique and only occurred once in all the deposit descriptions, giving a ‘long-tail’ distribution (Anderson 2006). Nathan (2011) concludes that ‘for endangered languages documentation, the metadata framework is to be discovered, not predefined, and the principle of the Long Tail is the opposite of focusing on the top 10–20 keywords . . . if supported and encouraged, documenters do produce diverse and more comprehensive metadata’ (my emphasis). Nathan’s review is suggestive of documenter practices for one archive, but needs further elaboration if it is to serve as a counterpoint to the deductive approaches which have dominated the field so far and which have emphasized standardization and metadata templates. A second example of induction comes from P. Austin (2010b), which looks at issues arising from working with legacy materials on the Guwamu language from southern Queensland, Australia, collected in 1955 by the late Stephen Wurm. There are practical, technical, ethical, and political issues that this legacy data raises because of a lack of meta-documentation. Exploring these gives insights into what current documenters might wish to take into account for future users. The Guwamu materials consist of: (1) fieldnotes of language elicitation (translations from English to Guwamu) collected from Willy Willis at Goodooga and comprising forty double-sided pages of notes with phonetic transcription and glosses in Hungarian shorthand, and (2) a short tape recording. At my request, the glosses were decoded and translated into English by Wurm and recorded onto tape in 1977. I copied the fieldnotes and added the English glosses (by transcribing Wurm’s tape recording), resulting in a

10

Peter K. Austin

138-page manuscript, a copy of which was deposited with the Library of the Australian Institute of Aboriginal and Torres Strait Islander Studies (AIATSIS).6 The following is a sample of data from the notes: jama inda goammu ŋalgaŋanda? bađarinj ŋalla balgaru ŋunan ugwɛ:ilɛja balunj ŋadju ilu iđamanjgija juraŋu-nda

Do you speak Guwamu? He is sick. A few days ago I camped there. I will leave my axe here with you all.

Very little metadata was recorded with these materials, with the result that there are difficulties with them of several types: 1. Problems with the form of the original: a. The handwriting in the notes is sometimes difficult to decipher. b. Orthography – Wurm’s transcription is not documented anywhere in the notes but appears to be similar to the International Phonetic Alphabet. It is quite surface phonetic but appears both to overdifferentiate (e.g. by recording gemination for consonants) and underdifferentiate (e.g. by failing to distinguish apico-alveolar and lamino-dental nasals). c. Word boundaries are sometimes incorrectly represented. d. There is sometimes cryptic glossing, or apparently wrong glossing.7 e. Changing understandings over time of the language being recorded – Wurm was clearly working out the structure of Guwamu as he went along (and there are some comments in the fieldnotes which indicate his guesses about particular morphemes), so his transcription (and translation) varies from the first page to the last.8 2. Problems with the lack of context – we know nothing of how the material was recorded, what sessions took place, the background of the speaker and his involvement in and attitudes towards the project (on tape he sounds enthusiastic, at least when singing). No information is available about agreements entered into or any compensation or dissemination arrangements. 3. Problems of unclarity about protocol, i.e. access and usage rights to the materials in their various forms. The copy of Austin’s notes at AIATSIS have the following access restrictions applied to them: ‘Closed access – Principal’s permission. Closed copying & quotation Principal’s permission. Not for 6 7

8

See www.aiatsis.gov.au/library/docs/langbibs/Guwamu Kooma July07.pdf. P. Austin and Crowley (1995: 60) give examples from work on legacy materials of such errors arising because the collector could not understand the consultants’ accent or pronunciation, or because the semantics were misunderstood; we find instances of the latter in Wurm’s notes but not the former. Bowern 2003 mentions that Gerhardt Laves began to analyse Bardi material collected in the 1930s as he was writing it down and made mistakes as a result, i.e. he did not write what he actually heard but what he thought he had heard. Also, Steele (2005: 84) comments on William Dawes’ notebooks on the Sydney language: ‘In order to be in a position to make some assessment of the soundness of an interpretation of a word, expression or sentence provided by Dawes, it is useful to have an idea of at which stage of his language learning an entry was created.’

Language documentation and meta-documentation

11

Inter-Library Loan’. However, the notes have also been re-transcribed and typed up by Jeanie Bell and myself in Toolbox format. The status of these derivative works is unclear. 4. Stakeholder and political issues – there is a community in southern Queensland who identify themselves as Guwamu, although no one today speaks the language. What their relationship is to Willy Willis and any interest or rights they might have in the materials collected by Wurm are unclear. Similar problems with legacy materials have been identified by Bowern (2003), Schmidt and Benn¨ohr (2008), Innes (2010), and O’Meara and Good (2010), and these all point to the need for as rich a meta-documentation as possible at the time material is collected and processed, especially for future researchers and users.9

3.3

Comparative approaches

An area for exploration for the development of a theory of meta-documentation is comparison between the place and function of metadata in language documentation and its role and application in other allied fields, including social and cultural anthropology, physical anthropology, and archaeology (the so-called ‘four fields of anthropology’). The beginnings of such comparisons are made in Ember and Good (2011) and Hanks (2011). As Ember and Good (2011) point out, many of the data types used across the fields of linguistics, cultural anthropology, physical anthropology, and archaeology are similar or identical, and that ‘some, but not much, metadata is now shared across the four subfields’. They suggest that linguistics could learn from cultural anthropology, which ‘has devised rich, semi-standardized vocabularies to describe cultural ideas and behavior’. In addition, archaeology and physical anthropology have rich ways to deal with spatial data and taxonomy. Hanks (2011) shows that archaeology (especially that influenced by Hodder 1999) has been more reflexive about its practices in the past fifteen years than language documentation has (e.g. by regularly encoding daily field diaries, debating the role of publication of ‘raw data’ (field reports) versus ‘cooked analyses’ (academic papers), and working on ways to share comparative datasets). Its concerns have also differed from language documentation as well in that it tends to be practised almost exclusively by teams of specialists (e.g. bone specialists, ceramics experts, usewear analysts etc.), resulting in fragmented practices among those involved in 9

Schmidt and Benn¨ohr (2008) discuss a number of issues with recovering data and metadata from several types of digital legacy materials; one of the recommendations at the end of their paper is ‘[d]ocument transcription and annotation conventions as well as the design, structure, and technical realization of your corpus as early on and in as much detail as possible. Publish this documentation in a form in which it will still be accessible 50 years from now’ (127).

12

Peter K. Austin

a given project, with greater emphasis on recording and analysing the excavation itself (for linguistics this would be the documentation of the recording event(s) – see Nathan 2010a on the importance for sound recording of attention to, and documentation of, spatiality), and so ‘post-excavation analysis is in fact at the periphery of the interpretive process’ (Hanks 2011, see also A. Jones 2002). Perhaps language documentation could benefit from exploring this separation between meta-documentation and interpretation of observable events and processes on the one hand, and meta-documentation and interpretation of corpora and other outputs on the other. 4

A possible typology of language documentation project designs

Over the past nine years I have been involved at various times in the assessment panels for grant applications for language documentation projects submitted to the Endangered Languages Documentation Programme, the Volkswagen Foundation DoBeS project, and the Documenting Endangered Languages inter-agency programme of the National Science Foundation and the National Endowment for the Humanities. Over this period I estimate that I have assessed over 1,500 grant applications; from this I believe it is possible to obtain a sense of the kinds of project design and planned activities and outcomes that applicants consider would receive positive evaluations from the grant selection panels. Both successful and unsuccessful applications can reveal much about what the applicants wish to do (within the parameters of the application specifications of each individual funding agency).10 In terms of overall project design, Grinevald (2003: 58–60), drawing on Cameron et al. (1992), presents a typology of fieldwork orientations by linguists over various time periods that is a useful starting-point: r fieldwork on a language – this is the traditional model associated with Boasian-type linguistic description and (later) ‘ethical research’ r fieldwork for the language community – this model emerged in the 1960s under the title ‘advocacy research’ r fieldwork with the language speakers – this was developed in the 1980s as ‘action research’ or ‘negotiated fieldwork’ r fieldwork by trained language speakers – this can be labelled ‘empowerment research’ and, while found sporadically in the nineteenth and twentieth centuries, only became practised more generally from the 2000s onwards. Himmelmann (2006: 15) identifies as a key feature of language documentation ‘work in interdisciplinary teams – documentation requires input and expertise from a range of disciplines and is not restricted to mainstream (“core”) 10

Although in a number of cases applications do not conform to the application guidelines issued by the funders.

Language documentation and meta-documentation

13

linguistics alone’. We may contrast this with ‘lone-wolf’ research (P. Austin 2005, 2007, Crippen and Robinson 2013), where a single individual carries out all aspects of a project (what Dwyer (2006: 54) describes as ‘the go-italone model of research: go in, get the data, get out, publish’). My impression, gained from reviewing documentation grant applications, is that, despite Himmelmann’s proposal, the majority of projects are planned as lone-wolf research on a language or research for the language speakers. Further, by analogy with the typology of human land use developed within geography (see O’Neil 2011), we can identify the following general project design types:11 1. Hunter-gatherer projects. In land-use studies, hunter-gatherers are typified by a ‘primary subsistence method that involves the direct procurement of edible plants and animals from the wild, by foraging and hunting, without significant recourse to the domestication of either’ (Familypedia, huntergatherers).12 Within linguistics this kind of project typically involves rapid surveys relying on questionnaires (primarily lexical) for language identification and classification, and collecting basic typological data, never staying in one place for any length of time (Simpson 2007 refers to this as ‘FiFo (fly-in fly-out) linguistics’). 2. Slash-and-burn swidden projects. In land use, slash-and-burn swidden is characterized by the ‘cutting and burning of forests or woodlands to create fields for agriculture . . . or for a variety of other purposes. It is sometimes part of shifting cultivation agriculture, and of transhumance livestock herding . . . and operates on a cyclic basis’ (Wikipedia, slash and burn)13 . In language documentation these are typically 3–5 year projects aimed at the creation of a Boasian trilogy (grammar, texts, dictionary), after which the researcher moves on to the next language to be studied. 3. Sedentary intensive cultivation projects. In land use, this is a kind of fixed-location agriculture that ranges from feudal to communal sociopolitically, with employment of local serfs and artisans in temporary or specialist roles (and the necessary application of fertilizer and pesticides, or crop rotation). For language documentation, these are typically long-term projects on a single site, often with Christian missionary connections, e.g. projects carried out by SIL International (formerly the Summer Institute of Linguistics) – see Dobrin (2009) for a discussion of the scope, institutional underpinnings, and wider implications for academia of the work being carried out by missionaries. 11 12 13

These are meant to be broad types for general classificatory purposes; any given project may well be a mixture of types, or change from one to another over the course of its lifespan. Available online at http://familypedia.wikia.com/wiki/Hunter-gatherer (accessed 16 April 2012). Available online at http://en.wikipedia.org/wiki/Slash-and-burn (accessed 16 April 2012).

14

Peter K. Austin

4. Plantation projects. In agriculture, plantations typically involve using thirdworld local residents to grow consumable products according to a specified form which are then extracted to be transported to a first-world context where they are refined and expensive value-adding takes place for distribution and sale. In language documentation, such a project typically involves training native speakers to transcribe and gloss language data (in a local orthography and lingua franca) using such software tools as ELAN14 and Toolbox.15 The outside linguist then takes the resultant files ‘home’ to process them further (by ‘cleaning up’ transcriptions and glosses, and adding further analytical labels, or translations into an international language), and publishes academic papers based on the analysed data. Note that the skills acquired by the native-speaker artisans typically have no local application. 5. Sustainable projects. In land use, this requires an ecologically driven holistic approach that may include reforestation and recuperation of damaged land. It typically combines social, ecological, and economic objectives (Munasinghe 1993), and involves careful resource assessment and mixed production systems with close management and control, taking a long-term perspective. For language documentation, few exemplars of what could be sustainable projects have so far been developed. While we generally understand sustainability in a language-archiving context (Nathan 2010b), it is unclear how to develop and sustain documentation projects and their necessary relationships beyond the three- to five-year cycle that typifies research and academic life. It is also unclear what kind of research project models can best contribute to sustaining endangered languages and the communities who want to maintain and develop them. This rough typology, based on impressions gained from reviewing research grant proposals, suggests that language documentation projects can be categorized into a set of general design and organizational types. If the aim of granting agencies, and researchers more generally, is to promote sustainable projects, there is a need to move beyond our current models, take a longer time-scale perspective, and develop (and meta-document) a range of different approaches involving more stakeholders taking different roles and contracting different relationships with each other. 5

Conclusion

The practice of language documentation over the past fifteen years suggests that we need a new development in this sub-field of linguistics, namely metadocumentation (or meta-documentary linguistics), which aims to document 14 15

Available online at http://tla.mpi.nl/tools/tla-tools/elan/ (accessed 7 November 2012). Available online at www.sil.org/computing/toolbox/ (accessed 7 November 2012).

Language documentation and meta-documentation

15

the goals, processes, methods, and structures of language documentation projects. We can develop this field by theorization and by investigation of current and past practices, and by exploring comparative approaches. By creating meta-documentation for projects now, we shall hopefully reduce the legacydata problems for future researchers compared with those that we face today (because such meta-documentation was not thought about very deeply in the past). Appendix: OLAC metadata The following is a list of the basic metadata categories proposed by the Open Language Archives Community (OLAC): Contributor Coverage Creator Date Description Format Format.encoding Format.markup Identifier Language Publisher Relation Rights Source Subject Subject.language Title Type Type.linguistic

An entity responsible for making contributions to the content of the resource The extent or scope of the content of the resource An entity primarily responsible for making the content of the resource A date associated with an event in the life cycle of the resource An account of the content on the resource The physical or digital manifestation of the resource An encoded character set used by a digital resource A mark-up scheme used by a digital resource An unambiguous reference to the resource within a given context A language of the intellectual content of the resource An entity responsible for making the resource available A reference to a related resource Information about rights held in and over the resource A reference to a resource from which the present resource is derived The topic of the content of the resource A language which the content of the resource describes or discusses A name given to the resource The nature or genre of the content of the resource The nature or genre of the content of the resource from a linguistic standpoint

2

A psycholinguistic assessment of language change in eastern Indonesia: Evidence from the HALA project Amanda Hamilton, Jawee Perla, and Laura C. Robinson

1

Introduction

Experience reveals that a language in decline is unlikely to reverse its condition until (at the very least) stakeholders discover that this attrition is occurring. But while this is straightforward enough, detecting and quantifying language loss, especially during its crucial early stages, is notoriously inexact. So far, language vitality assessment has largely focused on subjective consideration of sociolinguistic indicators through surveys, questionnaires, and ratings scales (Fishman 1991, UNESCO 2003, Hinton 2006, Otsuka and Wong 2007, among others). While the trends in language attitudes and usage explored by these measures are indisputably crucial to understanding language change, these trends also possess the unfortunate tendency of remaining largely imperceptible until they have become quite pervasive. W. O’Grady et al. (2009) developed the Hawai’i Assessment of Language Access (HALA) with hopes of improving the sensitivity of language vitality assessment. Rather than relying on sociolinguistic measures, the HALA project comprises a set of psycholinguistic tests which the project’s developers expect will be better able to detect subtle language dominance effects in the minds of speakers before the corresponding sociolinguistic signs become evident. By employing the HALA instrument in communities where languages are under potential threat, they aim to help these communities recognize encroaching language loss years earlier than they otherwise would – that is, before it becomes ‘a decline that is already firmly entrenched, perhaps irreversibly, for lack of earlier indications of endangerment’ (Schafer et al. 2009: 1). Specifically, the developers’ stated goal is to provide a simple, portable fieldwork tool offering sensitive measures of language strength in bilingual

We thank the following for their generous assistance with this project: William O’Grady, Amy Schafer, Marlon Adang, and the residents of Pitung Bang, Alor, Indonesia.

16

A psycholinguistic assessment of language change

17

speakers of any language pair (W. O’Grady et al. 2009: 110). The HALA project was developed and trialled, however, at an American university, and so this goal remained to be verified. In encouraging others to ‘test our test’ (110), O’Grady et al. caught the attention of the current authors. And so, in 2010, the HALA project travelled to Pitung Bang, an agrarian village in eastern Indonesia. The current study has two main objectives. First, we present and discuss the results of the Pitung Bang HALA experiment; second, we consider the extent to which the HALA instrument seems, in this instance, to have reached its developers’ goals, outlined above. We will begin with a brief introduction to the HALA project itself, followed by a word about Adang and Indonesian, the languages on which the current study focused. Then we will discuss the administration and results of the Pitung Bang HALA experiment, followed by a critique of the HALA project itself. Finally, we shall suggest possibilities for further study. 2

The HALA project

The HALA project’s theoretical foundation is the well-attested relationship between word frequency and accessibility (Jescheniak and Levelt 1994, Bates et al. 2003, Szekely et al. 2005, among others). Briefly, ‘increased use leads to improved lexical accessibility’ (Gollan et al. 2008: 788); that is, a speaker can produce a word she uses frequently with faster speed and better accuracy than she can a word she uses less often. It follows that if a bilingual speaker begins to favour one language over the other, lexical accessibility (as evidenced by speed and accuracy of recall) will become relatively lower in the dispreferred language. When we consider that language loss is really just this phenomenon multiplied over an entire community, a connection between relative lexical accessibility and language death – i.e. the connection on which the HALA project is based – becomes clear. The empirical goal of the HALA project, then, is to compare the relative accessibility of the two languages in the mind of the bilingual in order to detect even very minute differences between her strengths in the two. In practice, the HALA project is a body-part picture-naming task; Figure 2.1 presents a list of the thirty-one test items. (The eight starred items are those that were eliminated for the ‘Agreement’ subgroup, as discussed below.) Participants are tested in both their languages with at least one day intervening, and then each participant is compared with herself: her accuracy and reaction times (RTs) in one language are compared with her accuracy and RTs in her other language, with the results taken to indicate her relative strength of access in these languages.

18

Amanda Hamilton, Jawee Perla, and Laura C. Robinson

Figure 2.1 The thirty-one test items. Asterisks indicate items removed in the Agreement subgroup (see Section 5)

3

Adang and Indonesian: two languages of Pitung Bang

Adang is an Alor-Pantar (non-Austronesian) language spoken in a dozen or so agrarian villages surrounding the regional capital city, Kalabahi, on the island of Alor in eastern Indonesia’s Nusa Tenggara Timur province (Haan 2001: 1). Estimates of the number of speakers vary considerably, from over 31,000 (Lewis 2009) to just 7,000 Haan (2001: 1). Adang has no standardized orthography, and Pitung Bang children are educated in Indonesian. Haan (2001) describes Adang’s sociolinguistic environment as being in a dire state. He writes that pressure from Indonesian is particularly strong because of the villages’ proximity to the city, and that parents now prefer to speak Indonesian to their children because it is seen as more prestigious. By contrast, Indonesian (Austronesian family, Malay subgroup) is the official language of the Indonesian archipelago and claims over 23 million native speakers (Lewis 2009). It has a standardized writing system and is used in education, media, and government throughout the vast island chain. 4

Method

4.1

Participants (n = 16)

Sixteen participants (seven females and nine males), ranging in age from 18 to 69 years (M = 42), participated in the Pitung Bang HALA experiment. They were selected by the chief Adang consultant, himself a Pitung Bang resident, and they all received a small gift (a bag of coffee, sugar, or betel nut) for their contribution. 4.2

Materials

Because participants were located by walking from house to house through the village with the head consultant, the experimental ‘lab’ was necessarily small and portable. Trials were performed using a laptop computer and a Marantz audio recorder with an attached external microphone positioned on a tabletop stand. The picture stimuli were presented via two shockwave flash (.SWF)

A psycholinguistic assessment of language change

19

Table 2.1. Testing schedule Session 1

Adang Indonesian

Session 2

Bod1

Bod2

Bod1

Bod2

Group 1 Group 3

Group 2 Group 4

Group 4 Group 2

Group 3 Group 1

videos, referred to as Bod1 and Bod2. Participants’ responses were recorded as .WAV files on the Marantz, as was the sound of the focusing beep that accompanied each image’s appearance on the screen. 4.3

Procedure

The HALA project requires two separate testing sessions, spaced by at least a day, so that each participant is tested once in each of the two languages under investigation. For the current experiment, thirteen of the sixteen participants were tested twelve days apart, while the three who were not available on the second testing day were tested two days later, i.e. fourteen days after their initial session. In order to avoid possible confounds resulting from language testing order and/or video order (Bod1 versus Bod2), the sixteen participants were divided into four groups and then balanced for language and test as shown in Table 2.1. Participants were tested in their homes, with the laptop and recorder set up on convenient pieces of furniture. Before beginning the trials, an informal conversation took place between the participant and the head consultant or one of the two researchers, in which the experiment was explained and instructions provided. Initially, the researchers read out the supplied HALA instructions script (translated into Indonesian), but it quickly became apparent that participants benefited more from hearing concise, tailored instructions and then asking questions. Hence, no formalized instructions were used for this experiment. Both Bod1 and Bod2 began with six optionally repeatable practice items (black-and-white photos of single objects, like a bowl, hat, or T-shirt, each with a red circle around it). The practice items were followed by the thirtyone test items, which differed in order between the two tests. These main items comprised a black-and-white photo of a person (the same adult male in all instances) or a part of that person on a plain white background. A red circle was drawn around a subsection of the photograph, marking the specific body part that the participant was to name. Items remained on the

20

Amanda Hamilton, Jawee Perla, and Laura C. Robinson

screen between 3,500 and 4,500 ms.1 The total duration of each body part naming test was about 2.8 minutes when the practice items were viewed only once. Finally, at the end of each participant’s first session, the researchers administered a shortened version of the Language Experience and Proficiency Questionnaire (LEAP-Q; Marian, Blumenfeld and Kaushanskaya 2007). This questionnaire, completed orally, asked participants about their language use and general demographic background. It was included in order to provide the sociolinguistic context essential to understanding why and how any observed language shifts might be occurring in the first place.

5

Results

The data from the two experimental sessions was coded for both accuracy and RT. First, the chief consultant and one of the researchers worked together to assign accuracy codes to the majority of the data; the researcher then completed the accuracy coding herself based on the judgements the consultant had already provided. All responses received one of seven codes, as shown in Table 2.2. Coding followed a modified version of the guidelines provided with the HALA materials (A. Schafer, pers. comm., 22 September 2010). While coding for accuracy, researchers were struck by the variety of different names that participants provided for some items. The overall mean number of discrete responses per item, not counting unintelligible or no-response instances, was 3.18 (SD = 1.30, range 1–6). Furthermore, the ‘most popular’ or dominant response for each item was provided, on average, in just 69.35 per cent (SD = 19.75 per cent, range 18.75–100 per cent) of trials.2 Together, these two factors – number of responses and response agreement – comprise what C. Johnson, Paivio, and Clark (1996: 119) call ‘linguistic codability or uncertainty . . . a robust predictor of naming difficulty’. The high uncertainty level associated with some items seems to indicate that they were poorly suited to the languages or population under study. To avoid introducing confounds due to inappropriate test items, the researchers set a name agreement minimum of 50 per cent and formed an analysis subgroup 1

2

Wieting (2010) and Schafer (pers. comm., 17 January 2012) both report that all beeps occur at a set interval. At least for the videos used in the current study, however, this was not the case. While an exact list of times between beeps was not reviewed, the variance was observed in multiple participants’ recordings, both for Bod1 and Bod2, and in a manner that appeared consistent – indicating that this variation is likely to be a property of the videos themselves rather than an artifact of a particular recording setup. Compare, for instance, Snodgrass and Vanderwart (1980: 205–10), who report 86.47 per cent name agreement in their timed English picture-naming task involving written responses.

A psycholinguistic assessment of language change

21

Table 2.2. Accuracy codes Correct

Broader reference

Incorrect

Hypernym

Disfluent Repair Other

Lexical items considered appropriate responses for the stimulus by the chief consultant, and produced without disfluency or repair. Includes responses preceded by hesitation sounds (‘umm . . . ’) as long as these did not cause disfluencies within the response itself. Does not include overly general (see ‘Hypernym’) or phrasal responses. Situations in which the depicted body part must be named with a term whose actual reference encompasses more than just the circled part (e.g. tangan, the Indonesian term for both ‘hand’ and ‘arm’). A subset of correct items. Responses not considered accurate or complete by chief consultant. Also includes phrasal responses (e.g. ‘person’s arm’) and failures to provide any response. Responses that were both disfluent and incorrect were coded as incorrect, as were attempted repairs in which the second term supplied was incorrect. Responses considered overly general in situations where speakers usually use a more specific term (e.g. ‘body’ for ‘leg’ or ‘head’ for ‘nose’). A subset of incorrect items. Responses exhibiting stuttering, hesitation, or prolongation. Responses that begin with an incorrect term but are repaired to a correct one. Any situations not covered by the above, including unintelligible responses and instances in which bystanders prompted the participant’s response.

(called ‘Agreement’) containing only those twenty-three items that met this minimum in both languages (see Figure 2.1).3 Reaction times, defined as the time between picture onset and onset of the naming utterance, were recorded and analysed using Praat software. Hesitation noises (‘umm . . . ’) that did not cause disfluencies in the actual response were counted as part of the RT. Stuttering, prolongation, or self-correction within the response item itself, however, resulted in the item being marked as disfluent or repaired and removed from the analysis. A total of 992 responses were coded (16 participants x 31 items x 2 languages). They comprised 68.45 per cent correct tokens (n = 679, with ‘broader reference’ included), 26.01 per cent incorrect tokens (n = 258, including hypernyms and ‘no response’), 2.32 per cent disfluencies (n = 23), 2.12 per cent 3

Note that ‘agreement’ was determined independently of ‘correctness’, although in most instances the two coincided. The agreement minimum was used only to eliminate poorly suited test items; accuracy codes were determined separately based on the chief consultant’s intuitions. The eight items eliminated due to low agreement were ‘ankle’, ‘ear’, ‘elbow’, ‘foot’, ‘hand’, ‘palm’, ‘shoulder’, and ‘wrist’.

22

Amanda Hamilton, Jawee Perla, and Laura C. Robinson

Figure 2.2 Per cent accurate responses in Adang and Indonesian by age group

repairs (n = 21), and 1.11 per cent ‘other’ (n = 11). Disfluencies, repairs, and ‘others’ were not included in either accuracy or RT measures. 6

Accuracy analysis

To begin, results were screened for outlying participants or items, defined as those with an average Adang or Indonesian accuracy score more than 2.5 standard deviations from the overall mean for that language. One item, ‘wrist’, was eliminated from further accuracy analysis due to the very low rate at which participants correctly named it in Indonesian. Furthermore, one participant was identified who had been marked with an accuracy score of zero in Adang because he had responded to every item in that language with a phrase. In contrast to ‘wrist’, he was removed from both the accuracy and RT analyses because it was clear that he had not fully understood the task. First, Adang and Indonesian overall accuracy scores were compared. No significant differences were found, either when all items were included or when the analysis was restricted to the Agreement subset. For a more detailed comparison, participants were then divided into three age groups: those under the age of 30, called the young adults (n = 4), those between 30 and 49, referred to as the adults (n = 6), and those aged 50 years and up, called the older adults (n = 5). Because the small n of these three groups makes them unlikely to reveal significant differences, they were not submitted to statistical analysis; rather, they were observed numerically. As Figure 2.2 shows, the young adults were more accurate in Indonesian than they were in Adang, while the opposite was true for the older adults. Young adults’ Indonesian responses were 10.16 per cent more accurate than

A psycholinguistic assessment of language change

23

their Adang ones; older adults showed 6.16 per cent greater accuracy in Adang. 7

Reaction time analysis

Although the HALA project is designed to measure both relative accuracy and speed of access, the test developers consider speed to be the key measure (W. O’Grady et al. 2009: 102, 108). To help ensure the RT data’s validity, an additional subgrouping method was introduced for this analysis, after Tang (2011). The ‘Matched by Participant’ group retained only those items a given participant identified correctly in both languages (Tang 2011: 60). Despite the fact this results in a slightly different list of items (and different number of items) being included for each participant, it improves the ability of the test to compare participants with themselves accurately. This Matched subgroup comprised 422 (53.90 per cent) of the 783 responses retained for the overall RT analysis. As above, results were screened for outlying participants or test items, defined as those with an average Adang or Indonesian RT more than 2.5 standard deviations from the overall mean for that language. One test item, ‘ankle’, was removed from further RT analysis due to the very long naming times it displayed in Indonesian.4 First, reaction times were compared between the two languages over all 783 responses, then over all items in the Agreement subset, and finally over only Matched Agreement items. No significant differences were found. The Matched Agreement subset was then divided into the same three age groups as before5 and analysed numerically. As Figure 2.3 displays, a pattern similar to that seen in the accuracy calculation emerged: young adults’ Indonesian RTs were shorter than their Adang ones, while for the older adults’ group the opposite was true. The young adults performed on average 9.68 per cent (134 ms) faster in Indonesian than in Adang, while older adults performed 14.79 per cent (197 ms) faster in Adang than in Indonesian. Finally, one additional analysis was performed to help eliminate confounds introduced by mismatches between the body parts displayed in the stimuli and the languages’ individual strategies for naming the body. Szekely et al. (2005: 14) report that shared-name test items (i.e. items for which the most common response provided is also the most common response for another test item) are associated with lower name agreement and longer RTs in some cases. To test 4

5

As mentioned above, one participant was excluded from the entire test because of task-based difficulties. Additionally, two participants included in the accuracy calculations were excluded from RT analysis due to loud background noise in their recordings. With the two additional participants removed, young adults = 3; adults = 5; and older adults = 5.

24

Amanda Hamilton, Jawee Perla, and Laura C. Robinson

Figure 2.3 Mean RT in Adang and Indonesian by age group

whether this might also apply to the current study, the researchers eliminated from the Matched Agreement dataset all items for which the most commonly given name was also the most common name for any other item.6 While this did in fact result in slightly shorter RTs and slightly higher mean accuracies in both languages, the effect was once again not statistically significant. 8

Use and domains questionnaire

Finally, data from the informal language use questionnaire was compiled to provide a glimpse at Pitung Bang’s linguistic landscape. Respondents reported Adang as their main language of the home and farm – informal domains within the village – and Indonesian as the primary language of the church and market, domains characterized by greater formality and extra-village contact. Of course, in a multilingual society like Indonesia, this could be due largely to necessity rather than attitudes about language status. Nevertheless, as Ingram (2006:11) states, ‘if the languages are to survive, there is need for them to be used in all the domains in which the people use language’. Furthermore, while the majority of participants name Adang as the primary or only language they use with their siblings and parents (80 and 93 per cent, respectively), only 70 per cent claim that the same is true with their spouses. In other words, relationships formed during childhood are maintained in Adang, but those formed during adult life – and, crucially, those which facilitate the 6

This resulted in the removal of ‘hand’ and ‘arm’ (shared in both languages) and ‘face’ and ‘forehead’ (shared in Adang). Of course, to preserve the integrity of the dataset, any test item removed from one language was also removed from the other.

A psycholinguistic assessment of language change

25

upbringing of modern children – rely more on Indonesian. More worrisome still, though, is that only one-third of participants with children report speaking mainly Adang to them. Nearly 60 per cent, on the other hand, report using only Indonesian; the remaining 8 per cent use both. Additionally, the mean age of the four participants claiming to use primarily Adang with children is 60 years – nearly twenty years older than the average participant in this study – while the mean age of those who report using Indonesian with their children is just under 38 years. Clearly, today’s parents have almost categorically switched to rearing their children in Indonesian. 9

Discussion

While the dataset was small and the results inconclusive in many areas of this experiment, it can nevertheless provide a foundation on which to base dialogue about, and future research into, the vitality of Adang. Furthermore, this field test of the HALA project – one of the first – can generate important critiques of the theoretical and practical utility of the test itself. 9.1

The Pitung Bang HALA results

Overall accuracy and RT measures did not show significant differences between Adang and Indonesian. When the results are further divided by participant age, however, they seem to be suggestive of language decline. Specifically, the younger speakers appear both slower and less accurate in Adang than Indonesian, while just the opposite is true for the older speakers. While these effects did not reach statistical significance with this small dataset, they are nevertheless worthy of further inquiry. Furthermore, the language use data supports these observations and provides an important sociolinguistic context for them. First, Adang’s reported domains of usage appear limited to informal, within-community spheres; second, and more alarmingly, younger parents increasingly choose to speak Indonesian to their children. With such contracted language use patterns emerging, it is reasonable to expect a decline in the younger generation’s language strength like that suggested by analysis of the HALA data. 9.2

The HALA instrument

While the HALA project goals emphasize simplicity and ease of use, the concepts underlying it are not simple at all. Some linguistic and sociocultural challenges – both those foreseen for the test in general and those experienced during the current project – are outlined below. While the HALA tests certainly have potential as valuable research tools, they do not yet meet their stated goals

26

Amanda Hamilton, Jawee Perla, and Laura C. Robinson

of providing ‘easy-to-interpret results . . . for any language’ (W. O’Grady et al. 2009: 110). r Varying cross-linguistic frequency: While the facilitative effect of word frequency on lexical access has been reliably replicated, it will be near impossible to determine baseline norms of word frequency in the understudied languages that the HALA project targets. Although Bates et al. (2003: 368) report ‘surprisingly high cross-language correlations among our word frequency measures’ for the seven languages examined in their analysis, the true representativeness of their dataset must be questioned. Five of their seven languages were Indo-European; all were from highly modernized cultures. It seems premature to assume, based on such an unrepresentative dataset, that word (or indeed conceptual) frequencies will generally align cross-linguistically – a major presupposition of the HALA test. r Word structure variables: An essential assumption of the HALA tests is that ‘speakers are accessing a stored linguistic unit’ (A. Schafer, pers. comm., 22 September 2010). Adang body parts, however, are obligatorily inflected for possessor (Haan 2001: 130), and both languages exhibited widespread use of compound or multiword responses (e.g. Indonesian ‘toes’, jari kaki, literally ‘leg digits’). The potential need for morphosyntactic processing in examples like these may reduce the test’s validity. r Age of acquisition: Multiple studies have argued that age of acquisition influences lexical access independently of frequency (e.g. Szekely et al. 2005, Dent, Johnston, and Humphreys 2008). If this is so, then it would prove especially relevant for studies of bilinguals, who may not have begun learning both languages simultaneously. In the current experiment, all participants who provided their age of Adang acquisition reported learning it since birth. The average age at which they reported starting to learn Indonesian, however, was 5.56 years. Hence, age of acquisition presents another possible confound in the HALA experiment. r Unfamiliar graphic symbolism: It quickly became clear during the trials that not all participants were familiar with the convention of using a circle to draw attention to a part of a whole. More than one participant began the test by responding ‘body’ or ‘person’ (or even ‘white person’!) to every photo. The use of culture-specific graphic symbolism may inhibit the proposed widespread applicability of the HALA instrument. r Unfamiliar medium: It is unlikely that many participants had had much previous experience of looking at digital images. Computer usage (and even television ownership, for that matter) is very limited in Pitung Bang. Participants may therefore have been made anxious by the unfamiliar medium; certainly, despite the researchers’ best efforts at angling the computer screen, many appeared to struggle to see the images.

A psycholinguistic assessment of language change

27

r Comparatively low formal education: Participants’ low levels of formal education mean that most of them were probably quite unfamiliar with research and testing environments. Their average number of years at school was 8.75; only five had completed high school, and one had only one year of schooling. Education is known to influence performance in experimental trials; as Tainturier, Tremblay, and Lecours (1992: 469) write, ‘Different patterns of [RT] performance can be observed, as a function of relatively small differences in schooling level . . . level of school education should always be explicitly considered as a potentially influential variable in the interpretation of language-related experimental data.’ Indeed, such schooling-based differences could prove particularly relevant in studies like the current one, where participants received their education in only one of the two languages under investigation. r Culturally unsuited methodology: Finally, the experimental design itself simply did not align well with participants’ cultural norms. Attempting to isolate subjects for the duration of the test in such a communal society was not only fruitless but bordered on offensive; trying to convince bystanders not to help by calling out answers would have turned the researchers into shushing disciplinarians. Participants clearly wanted to complete the tests together. Rather than struggling to make participants’ behaviour conform to the mandates of the test, the researchers would undoubtedly gain more valuable data by altering the experiment to fit within the cultural norms of the community (see Chung, Borja, and Wagers 2012 for a discussion of similar difficulties encountered while conducting linguistic experiments in Guam). The challenges, particularly the sociocultural ones, encountered during the Pitung Bang HALA experiment suggest that the measure is not as well suited for use in all fieldwork as its developers intended. At least in the current situation, the experiment would have had more success if it had been: r collaborative, rather than independent, in line with participants’ strong desire to complete the test together r interactive, rather than unidirectional, allowing participants more control over the format of the testing situation r flexible, rather than strict, in the conditions required to conduct valid trials, as it was nearly impossible to create the kind of experimental environment (a quiet setting, a lone, undistracted subject, participants who had not seen or overheard the test before taking it) that HALA ideally requires. Conforming to these recommendations would require fundamental changes to the HALA project, even to the point of its no longer being an RT investigation. In order for the developers to fulfil their goal of creating a universally usable instrument, though, these points must necessarily be considered.

28

Amanda Hamilton, Jawee Perla, and Laura C. Robinson

10

Conclusion and further research

While the results of the Pitung Bang HALA experiment did not reach statistical significance, they nevertheless suggest that Adang may be in decline among younger speakers. Continued investigation of Adang’s vitality, therefore, will undoubtedly prove instructive. Furthermore, a linguistic environment such as Pitung Bang lends itself to fruitful investigations of such matters as language contact and change, diglossia, and the effects of national language policies on minority speech communities. Research in these areas will contribute valuable knowledge to both areal and general linguistic studies. Additionally, the current project’s goal of ‘testing the test’ has led to observations on methodological changes that could render instruments like the HALA project more effective in communities like Pitung Bang. With the growing emphasis within linguistics on understudied languages and language endangerment, the usefulness of research instruments that can be tailored to fit a variety of sociocultural contexts will only grow. Hence, perhaps our strongest recommendation is for development of experimental techniques that can be adapted more readily to the existing norms of the environments in which they are used. Such methods will benefit the field as a whole by collecting better data and building better working relationships in countless communities such as Pitung Bang.

3

Documentation of endangered sign languages: The case of Mardin Sign Language Ulrike Zeshan and Hasan Dikyuva

1

Introduction

Since its inception in the 1960s and 1970s, sign language research has demonstrated that, at all linguistically relevant levels, sign languages are entirely on a par with spoken languages. Following from this initial breakthrough, it was natural that sign language research should expand into all the areas that are covered in research on spoken languages. Therefore, there has been a steadily growing stream of literature not only on the linguistic structures of various sign languages but on the whole spectrum of sub-disciplines in linguistics, such as psycho- and neurolinguistics, translation and interpreting studies, applied linguistics, sociolinguistics, and so forth (see Joachim and Prillwitz 1993). Sign languages thus often open up new perspectives on issues that are familiar from research on spoken languages but appear in a different light when seen from the point of view of visual-gestural languages. However, the realization that sign languages too are subject to processes of language endangerment is still very recent. For instance, the UNESCO Atlas of the World’s Languages in Danger does not include any data on sign languages,1 and not all textbooks and seminal publications on language endangerment make reference to sign languages. This chapter presents a case study of ongoing work on a severely endangered sign language in Turkey. Mardin Sign Language (MarSL) is a previously undocumented rural minority sign laguage from south-eastern Turkey (the first

This research has been funded by the UK Arts and Humanities Research Council under the EuroBABEL programme of the European Science Foundation’s EUROCORES scheme (project ‘Endangered Sign Languages in Village Communities’) and by the Endangered Languages Documentation Programme (Major Documentation Programme, grant no. MDP0200). We are grateful to our consultants Hasret Dilsiz, S¸u¨ kr¨u Dilsiz, Veysiye C ¸ elen, Veysi Dilsiz, S¸u¨ kran Fidancı, and Nadire Turan for sharing with us their unique sign language and their cultural heritage. We also thank all other members of the Dilsiz family for their help with this project. 1 Available online at www.unesco.org/culture/languages-atlas/ (accessed 22 March 2012). A project coordinated at the International Institute for Sign Languages and Deaf Studies, University of Central Lancashire, is now under way to remedy this situation and make data on endangered sign languages available for the Atlas. This will include Mardin Sign Language.

29

30

Ulrike Zeshan and Hasan Dikyuva

publication is Dikyuva and Dilsiz 2009). After introducing the language and its setting in the remainder of Section 1, we summarize the documentation process in Section 2, and then draw more general conclusions in Section 3. 1.1

Endangerment of sign languages

Like spoken languages, sign languages can become endangered for a variety of reasons. For instance, developments in medicine such as cochlear implants and genetic screening may reduce the number of deaf children, particularly in the industrialized countries. In other cases, for instance in West Africa and in parts of south-east Asia, the ‘importation’ of foreign sign languages, most prominently American Sign Language, has threatened indigenous sign language varieties. Educational policies and institutional settings such as special schools for deaf education often play an important role in these developments. A different scenario, which also applies to MarSL, concerns small-scale communities with a high incidence of hereditary deafness, concentrated in a tightly constrained rural setting and therefore sometimes referred to as ‘village sign languages’ (Zeshan 2007). This includes some of the world’s currently most critically endangered sign languages, and their systematic documentation is a very recent undertaking (see Zeshan and de Vos 2012). Currently no systematic overview of sign language endangerment is available. 1.2

Sign languages in rural communities

Most research on sign languages has previously focused on communities of deaf people who constitute linguistic and cultural minorities within the larger hearing, non-signing society. A critical mass of deaf people coming together in institutions such as deaf schools, deaf clubs, and so forth gives rise to sign languages in these cases, which are most commonly associated with larger, often urban, populations. Typically, there are significant communication barriers for deaf individuals, and collective experiences of exclusion and disadvantage, but there may also be dedicated legislation to protect the languages and rights of these ‘deaf communities’ with their own linguistic and cultural identities (see Wheatley and Pabsch 2010 for an overview focusing on EU countries). By contrast, ‘village sign languages’ in rural communities exist in a radically different setting. They arise due to a localized very high incidence of hereditary deafness over several generations (Zeshan and de Vos 2012). In such communities, the sign language is typically co-created and co-used on a daily basis by all the deaf and the great majority of hearing people, and therefore some researchers speak of a ‘shared signing community’ (Kisch 2008). This means that communication barriers are of lesser or no importance, and often the daily lives of deaf people are quite similar to those of hearing villagers

Documentation of endangered sign languages

31

(Marsaja 2008). Typically, these rural sign languages have no official status or associated institutional implementation. The linguistic structures of these sign languages are now beginning to be documented (e.g. Sandler et al. 2005, Nyst 2007), and some of their structures differ radically from urban sign languages. They are therefore very important for a proper understanding of possible typological variation across sign languages, and subsequently for more informed typological comparisons between sign and spoken languages. The endangerment of village sign languages is usually due to pressure from larger urban or national sign languages (see Nonaka 2004 for a case study from Thailand). In the case of MarSL, the pressure stems from the national sign language of Turkey, T¨urk ˙Is¸aret Dili (T˙ID), which itself has attracted the attention of linguistic research only recently (see Zeshan 2003, Dikyuva and Zeshan 2008). 1.3

The sociolinguistic situation of Mardin Sign Language

1.3.1 The community context Mardin is located in south-eastern Turkey close to the border with Syria. From around the 1930s onwards, a local small-scale sign language developed in this town as a result of genetic deafness in an extended family. Although the setting cannot really be called a ‘village’, the dispersion of deafness and sign language use are in line with other sign languages of rural communities as described in Subsection 1.2. Interestingly, the family where the sign language arose has Dilsiz as their family name, which means ‘deaf’ in Turkish (literally ‘tongue/language-less’). The name ‘Mardin Sign Language’ was coined by the researchers documenting the language. The users of MarSL do not seem to have a particular name for their variety of signing but in conversation referred to it by various descriptive terms including dilsizce (Turkish for ‘deaf language’) and eski is¸aretler (Turkish for ‘old signs’). MarSL is currently used by an estimated forty individuals, all of whom are members of the extended Dilsiz family. Families in rural Turkey are typically large, with up to a dozen children in previous generations of the Dilsiz family. The initial version of the Dilsiz family tree that was constructed as part of this project includes over 200 individuals (see Subsection 2.2). The main consultants for this project include both deaf and hearing family members, and the fluent users who are providing the linguistic data are all in their 50s, 60s, and 70s. Therefore, it is clear that this language is currently on the brink of extinction, although it is still being used in the home and at family gatherings such as weddings. Our research so far has documented use of the sign language in Mardin going back to the 1930s, and there are at least four successive generations of deaf individuals in the Dilsiz family (Dikyuva and Dilsiz 2009). This

32

Ulrike Zeshan and Hasan Dikyuva

does not seem like a long period from the point of view of spoken languages, but in the context of newly emerging sign languages, it is a considerable time depth. There are documented cases of rural sign languages going back no further than the 1970s (see, for example, Nonaka 2009). During the first decades when the sign language developed in Mardin, there is very little evidence of any contact with Turkish Sign Language (T˙ID), since, unlike in other parts of Turkey, Mardin did not have any schools, clubs, or associations of deaf people using T˙ID. The oldest living user of MarSL, who is a hearing woman in her 70s, is illiterate, although a fluent user of both MarSL and the local Mardin variety of Arabic. The ongoing documentation project is now beginning to reveal a fascinating picture of unique social, linguistic, and cultural practices from the 1930s onwards. About twenty-five years ago, family members, including the deaf members, began moving away from Mardin to Istanbul and Izmir because of better employment prospects. Some family members remain in Mardin, but this does not include any of the deaf individuals. The geographic dispersal of the core community of MarSL users is one of the causes of the language’s endangerment, in addition to increasing contact with T˙ID.

1.3.2 The language context In its relatively short history, MarSL has existed in two quite distinct settings: a) the original setting in Mardin where the language first arose and developed (c. 1930–85) and b) the current setting where MarSL is in decline (c. 1985–present). The linguistic environment in which MarSL arose, although not including any other sign languages, was multilingual with respect to spoken languages, as Turkish, Kurdish, and Arabic are spoken in Mardin. In the past, the Dilsiz family’s main home language was the local variety of spoken Arabic, and there is evidence of contact between MarSL and spoken Arabic, particularly in the occurrence of mouth movements accompanying some of the signs. The setting of this initial period is described in more detail in Section 2 below, on the basis of our fieldwork with the Dilsiz family. Currently, all deaf members of the Dilsiz family are either monolingual in T˙ID (the younger generations) or bilingual in T˙ID and MarSL (the older generation). Curiously, MarSL is maintained mainly due to the presence of hearing family members, who are essentially ‘sign monolingual’. That is, hearing family members only know MarSL, but not T˙ID, as they have no incentive to associate with the urban deaf community in Turkey. Therefore, the deaf family members use MarSL in communication with their hearing relatives. This also implies that, currently, the younger monolingual T˙ID signers have limited communication with their hearing MarSL-using relatives, thus approaching the situation of communication barriers that is typical of urban communities of

Documentation of endangered sign languages

33

signers (see Subsection 1.2). The younger deaf family members have joined the urban deaf community, including deaf schools and clubs where T˙ID is used. The maintenance of a sign language by hearing people seems paradoxical at first sight, given that signing originated due to the needs of deaf people, but is not unique in this particular setting (see Lanesman 2011 documenting a parallel case in Israel). In addition, language shift is also evident in the spoken language used within the family. Except for our oldest informants, hearing family members have been shifting from speaking Arabic to speaking Turkish. This particular mix of sign and spoken languages creates special challenges for fieldwork, which are discussed in Subsection 2.1.

2

Documenting Mardin Sign Language: A case study

In this section, we provide an overview of how the research team has been working together with this particular micro-community of MarSL users. As we shall see, the involvement of deaf researchers is of particular importance in such a project. This section covers the setup, composition, and communication in our research group (Subsection 2.1) and the creation of multimedia resources documenting the language and its setting (Subsection 2.2). Although the documentation of MarSL includes aims and outcomes related to theoretical linguistics, for the purpose of this chapter we focus on those aspects that are of direct relevance to the MarSL community in terms of securing the linguistic and cultural heritage of MarSL.

2.1

The Mardin Sign Language research group

Our first contact with MarSL was established through Hasret Dilsiz, a deaf teacher of T˙ID who participated in one of the T˙ID sign language teacher training courses organized and taught by the present authors. Hasret is one of the younger family members, all of whom attended special schools for deaf children and who have shifted to using T˙ID. Hasret is a research assistant in the MarSL documentation project, and the main consultants we have been working with are close relatives, including her deaf father, S¸u¨ kr¨u Dilsiz, and his hearing siblings. One of the major challenges in working with this particular group is the array of languages to be negotiated during fieldwork. The interactions are often multimodal (signing, speaking, writing) as well as multilingual. Table 3.1 shows the linguistic backgrounds of some of the participants in one of our project workshops, and all these languages were used in one way or another during the workshop.

34

Ulrike Zeshan and Hasan Dikyuva

Table 3.1 Linguistic backgrounds of the research team

T˙ID Hasan Dikyuva Ulrike Zeshan

Hasret Dilsiz S¸u¨ kr¨u Dilsiz

Veysiye C¸elen

Veysi Dilsiz

Researcher, deaf Principal Investigator, hearing Research assistant, deaf MarSL consultant, deaf MarSL consultant, hearing MarSL consultant, hearing

√ √

√ √

–

–

MarSL some –

some √

√

√

International Sign

Turkish

English

Arabic

written only √

written only √

–

some

written only

–

–

–

–

–

–

–

some; spoken only √

–

√ √

–

–

some

√

√

This complex array of linguistic competences has many consequences for the way in which the research team has been working together. S¸u¨ kr¨u Dilsiz and Hasret Dilsiz work intensively on the MarSL data, using T˙ID as the metalanguage to discuss recordings of MarSL users. Translations and transcriptions into English (using ELAN) represent a time-consuming group effort and may need to go through several translation stages: r MarSL to T˙ID by S¸u¨ kr¨u r T˙ID to written Turkish by Hasret r written Turkish to written English by Hasan. One of the deaf consultants (S¸u¨ kr¨u) is not literate in any written language, while one of the hearing consultants (Veysiye) in addition does not have any language in common for fluent communication with any of the researchers. Some workshop sessions therefore included translation into spoken Arabic for Veysiye, or support by a sign language interpreter using spoken Turkish and T˙ID. The deaf researchers in the group are able to mediate between T˙ID and MarSL, which also serves to increase their metalinguistic skills. In addition to this substantial linguistic diversity, our group members live far apart in Ankara, Izmir, Istanbul, and the UK, so communication requires additional efforts. In a culture where face-to-face interactions are very important,

Documentation of endangered sign languages

35

a lot of travel within Turkey is needed. Communication about sensitive issues such as payments for consultants can be particularly challenging, and misunderstandings inevitably arise. Establishing mutual trust and respect as the basis for such a project has therefore been especially important, and the development of ethical practices is a continuous process rather than something that can be negotiated and agreed from the outset (see Section 3). We have found it very useful to issue regular newsletters to all consultants every three to six months (in written Turkish) about the progress of the project. Newsletters inform the participants about the project’s progress, any special events that have taken place, and future plans. Producing something tangible has been important, so that the members of the MarSL community, in particular the older family members, can experience some of the outcomes of our work at a practical level. 2.2

Language documentation resources

One of the aims of the MarSL project is to preserve this unique sign language as part of Turkey’s linguistic and cultural heritage in general and part of the Turkish deaf community’s heritage in particular. Turkey has a particularly rich history in relation to the use of sign languages, going back to the sixteenthcentury Ottoman court (Miles 2000). If MarSL cannot carry on as an actively used language, it can at least be preserved as part of the ‘deaf linguistic heritage’ in Turkey. For this purpose, the research group decided to compile multimedia materials, including the following resources: r videotaped conversations in MarSL, including recollections by the elderly informants about the ‘olden days’ in Mardin r videotaped vocabulary lists r documentation of the cultural background of MarSL, including genealogical diagrams of the Dilsiz family, maps of Mardin, photographs, and so forth. Video filming is carried out on a continuous basis, but for documenting the cultural background the project team organized workshops with MarSL consultants in Ankara in April 2011 and in Izmir in October 2011. These workshops have been very important both academically and for team-building, as they illustrate the way in which the consultants from the Dilsiz family actively shape the methodologies and outcomes of the project. Figures 3.1–3 illustrate some of the outcomes from these workshops. Given that the occurrence of deafness in the family has been the driving force behind the development of MarSL, one of our aims has been to assess the incidence of deafness in successive generations. The result of one of the workshop sessions on family history is shown in Figure 3.1, an extended family tree created by two female consultants from the Dilsiz family (S¸u¨ kran Fidancı and Nadire

36

Ulrike Zeshan and Hasan Dikyuva

Figure 3.1 Family tree diagram produced by consultants during a workshop

Turan). They reported that Mardin has a local tradition of family tree drawings in a particular style. Therefore, they were able to rely on a reference point for creating the very substantial family tree of their own family. Such a task would usually be quite challenging for most people and might require many false starts, particularly given the size and complexity of the family. However, the tree diagram displayed in Figure 3.1 was created without any hesitation, and although there was discussion about some of the individuals at times, as well as a problem with space on the paper for one branch of the family, the tree was essentially completed within one hour. If the researchers from outside

Documentation of endangered sign languages

37

Veysiye f Abdüsselam m

Veysi m Sükran ¸ f

Fatma f

Sükrü ¸ m Davut m Nadire f Emine f

???

ibrahim m

Mehmet m

Hasret f Songül f ¸ Seyhmus m Cahide f Sabah m

Fahima f Halise f Halil m

Fakkiye f Memire f Sultan f Zahide f ¸ m Besir Lütfiye f

Nahide f

Figure 3.2 Revised family tree diagram created by researchers. Reproduced from Dikyuva (2012)

had imposed a particular format unfamiliar to the consultants, the process and result would certainly have been compromised. Instead, relying on a culturally accessible framework and trusting in the validity of the method as initiated by the consultants themselves provided a much better output.2 This initial diagram was then used by the researchers to create more detailed family trees that show the distribution of deaf people in the extended family over several generations. Figure 3.2 shows a section of such a revised family tree diagram (bold and shaded names represent deaf individuals). The family tends to have more male (m) than female (f) deaf members, and the oldest known deaf people (to the far left) are a brother and a sister (Fatma and Halil). 2

This approach brings to mind the principles of ‘community-based participatory research’ (CBPR), as advocated, for instance, in Walters et al. (2009). Many of these principles are broadly in line with our approach, at least in terms of aspiration, although the specifics have not been worked out explicitly for our research on MarSL.

38

Ulrike Zeshan and Hasan Dikyuva

Figure 3.3 The physical environment in Mardin. Reproduced from Dikyuva (2012)

Figure 3.3 is a representation of the town of Mardin at the time when our main consultants were growing up. The map shows the location of their family home (in the centre of the map under the word kilise) and other relevant locations within the town. In addition to understanding who was involved in the creation and development of MarSL, it is also essential to understand the context of the physical locations, not least because this has an important impact on the daily interactions between sign language users who lived in Mardin from the 1930s onwards. Thus, the map shows the locations of households of other relatives and friends who were sign language users, the built environment, with only a single circular road for car traffic, the distance between relevant households (measured in minutes of walking, e.g. 5 dak. = 5 minutes), and other relevant places where sign language users would gather, such as the public cafes on the main road (c¸ay evi, kahve). A detailed analysis of this rich and layered information is beyond the scope of this chapter. At this point, we would like to focus on the process that resulted in the map displayed in Figure 3.2. As in the case of the family tree diagram, it was the consultants’ initiative to produce the map together as a genuine team effort. The external researchers again did not provide any specific guidance or form beyond a generic request for a map of the area.

Documentation of endangered sign languages

(a)

39

(b)

Figure 3.4 Screenshots taken during the map activity

During fieldwork, we videotaped many of the workshop sessions, and, in this case, analysing the video recordings will be particularly helpful because there was a lot of communication in both sign and spoken languages during the construction of the map. Four siblings of the family household shown in the centre of the map participated in this activity, and each person contributed additional features or modifications, often with accompanying comments that will help us analyse the map as fully as possible. Figure 3.4 shows two successive screenshots taken during this map activity. The researchers and the MarSL community have decided to produce a multimedia DVD and host a website where all these resources are brought together in an educational package. The consultants are taking an active interest in the structure and content of the DVD and website, and they have been adamant that information about their language should be widely accessible. For instance, they have suggested that a separate educational package should be provided for use in schools for the deaf where the deaf students communicate in T˙ID. As MarSL is a part of the national sign language heritage, deaf children should have access to information as part of their taught curriculum. It is not certain whether this will really be viable and when it could be done, but it shows clearly the involvement and motivation of the MarSL community in this project. 3

Conclusion

MarSL is a severely endangered sign language in a peculiar situation. Language shift has taken place not in favour of a high-status majority language but in favour of another low-status minority language, T˙ID. Moreover, although the language has arisen as a response to deafness, somewhat paradoxically it is now being kept alive largely owing to the presence of hearing users of MarSL. T˙ID itself has only very recently achieved official recognition, as part of disability legislation in 2005, and the T˙ID-using community is still struggling for implementation of basic linguistic rights, such as access to sign language

40

Ulrike Zeshan and Hasan Dikyuva

interpreting (Dikyuva and Zeshan 2008). This creates a situation of competing priorities, that is, the more successful T˙ID will be in its bid for status and recognition, the more pressure will be exerted on MarSL. Given that MarSL is under pressure from T˙ID, it is particularly important to inform the T˙ID-using community about the existence, status, and importance of MarSL. Involving deaf researchers in our team as well as T˙ID-MarSL bilinguals is therefore particularly important, as they are able to bring this unique sign language to the attention of the larger signing community in Turkey in a way that the hearing project members are unable to achieve. Whether or not revitalization has any chance of succeeding depends on the attitudes and actions of the T˙ID community. Government departments and official institutions are unaware of even the existence of MarSL, and, since they are struggling to implement legislation relating to T˙ID, it is highly unlikely that there will be any systematic response to the endangerment of MarSL. Before this project, the MarSL community was in an advanced stage of disintegration owing to its diaspora and to disrupted intergenerational transmission. Nobody in the urban communities of T˙ID users had ever heard of MarSL in the sense of being a minority sign language in Turkey. The organizations of deaf people, such as the Turkish National Deaf Federation, were as unaware of this sign language as the international community of sign language linguists was. Our project has made members of these various target communities aware of the existence of MarSL. For instance, the authors have presented information about MarSL at an important conference on sign language endangerment organized by the World Federation of the Deaf and the European Union of the Deaf, two of the most important international deaf community organizations (Dikyuva 2011). We have also conducted research into aspects of the grammar of MarSL, most notably the highly unusual numeral system, and are comparing this with other sign and spoken languages. The materials being collected now will allow a detailed reconstruction of sign language use and related social networks with deaf and hearing signers both within and outside the Dilsiz family in Mardin. Together with the stories about the childhood of our elderly informants, a clearer picture is now emerging of the social and linguistic situation in which MarSL developed and thrived. This is essential for considering an intriguing theoretical research question: what does it take for a language to emerge? It seems that the early history of MarSL comes close to the minimal requirements of a language community, both in terms of number of users (i.e. how many people it takes to constitute a language) and in terms of time depth (i.e. how much time it takes to constitute a language). The early development of MarSL has been an instance of an incipient linguistic community, and this is of interest to theoretical linguistics, as spoken languages do not offer such natural laboratories (see Zeshan 2011).

Documentation of endangered sign languages

41

Research on endangered sign languages requires particular attention to be paid to questions of ethics, and this is a topic under increasing discussion in the literature. For instance, Harris, Holmes, and Mertens (2009) explore the meaning of ethical research for sign language communities and make reference to the notions of trust and respect (see Section 2 above). However, the specific case of sign language endangerment has barely entered the discourse on research ethics yet (see Zeshan 2007), and it remains to be seen how approaches such as community-based participatory research may best be implemented in working with endangered sign languages. Finally, sign language endangerment and its associated issues clearly need to be examined seriously. The concept of linguistic heritage, commonly discussed with regard to hearing communities, is not yet widely considered in the context of sign languages, maybe because many deaf communities are still engaged in the struggle for proper recognition of their current sign languages, and historical information about sign languages is sketchy in many cases. However, as we become more sensitive to sign languages that are at risk of extinction, we need to devise strategies for their preservation in the same way as is currently being done for spoken languages. We hope that our work on MarSL makes a small contribution to this field.

4

Re-imagining documentary linguistics as a revitalization-driven practice David Nathan and Meili Fang

1

Introduction

This chapter poses the question: if the goals of documentary linguistics were driven primarily by considering the needs of language revitalization, what would the methodology, process, and form of language documentation look like? We ask this question with three goals: to raise awareness of the potential of documentary linguistics for supporting language teaching and learning – and to make concrete suggestions for doing so; to challenge the current ‘flat earth’ documentary linguistics which fears that if it strays from the epicentre of formal linguistics it will fall into an abyss; and in the hope that new syntheses between language documentation and revitalization can materially contribute to maintaining languages. Many language documenters have already started to advocate revitalization, and this chapter follows from papers by Yamada (2011) and Czaykowska-Higgins (2009), among others. Documentary linguistics is the single field that has chosen for itself the mission of ‘responding to language endangerment’ (Hale et al. 1992). However, the field has not held itself in any concrete sense accountable for the health of languages. If it does have this responsibility, evidence for success remains minimal, and a second question arises: should language endangerment be addressed through a more varied set of perspectives and activities, or should documentary linguistics be a more flexible umbrella for a range of responses? This question is for others to answer because both present authors work in areas adjacent, but somewhat peripheral, to documentary linguistics.1 We see responding to language endangerment not as solving linguistic questions but as helping to

This paper has benefited from the authors’ discussions with Gary Holton, Julia Sallabank, and teachers at the Warm Springs Language Program (Oregon) and Punana Leo School (Hawai’i), as well as the comments of an anonymous referee. We thank them, but take all responsibility for any errors that remain. 1

Nathan is based at the Hans Rausing Endangered Languages Project, which has a strong allegiance to documentary linguistics, but is a data archivist and multimedia producer. Fang is a practising language teacher, as well as a researcher in both pedagogy and formal linguistics.

42

Re-imagining documentary linguistics

43

meet human needs arising from loss of culture, identity, and language, whether those needs are intellectual, aesthetic, social, emotional, or economic. The current proposals are not, therefore, made in criticism of existing approaches or activities, but in the hope of seeing the evolution of methods and outcomes that can slow the rate at which languages continue to die. The chapter is organized into five themes. The first identifies examples of messages and techniques of ‘mainstream’ documentary linguistics which might actually be unhelpful for revitalization. The second theme is metadata, which enables the discovery, access to, and understanding of resources. Metadata is a crucial element in mobilizing documentation for use by language teachers and learners. The third theme notes gaps in the documentary approach which leave revitalization less well served than it might be. Then we briefly consider teaching and learning processes, and communities and ethics, before offering summary recommendations for a re-imagined documentary linguistics. 2

Reviewing messages and techniques of documentary linguistics

2.1

The tyranny of interlinearization

Interlinearization is typically presented as the sine qua non of language documentation (P. Austin 2010a: 18, Drude et al. 2011: 3). After recording a language event, a documenter typically interlinearizes its formal linguistic content (interlinearization involves segmenting and transcribing an utterance into morphemes and providing additional lines which gloss or translate those morphemes). Transcription reduces the speech stream to symbols (thereby losing around 99 per cent of its acoustic information; see Nathan 2009), and interlinearization implements a shift from uttered tokens to types in a putative linguistic system. This process is at the heart of the ‘one-way journey’ described below. Interlinearization requires a huge investment in time and consumes the bulk of the documentation effort (an hour of recording may take well over twenty hours to transcribe). Interlinearization is indispensable because it makes documentation accessible to linguists, and it may be useful to learners, especially in the absence of suitable learning materials. However, it is but one of the documentation outputs useful for revitalization, and its usefulness may be limited. Mithun (2011: 5) provides the following interlinearized example (a single word of Mohawk) to illustrate the Mohawk language’s polysynthetic typology: (1)

Thia ’tonsakani ’ ts´onhkwahkwe’. th-i-a’-t-onsa-k-an-i’tsonhkw-ahkw-e’ just-there-PAST-change-back-I-own-rear-pick.up-PFV ‘I just jumped right back in there.’

44

David Nathan and Meili Fang

Mithun explains that although any Mohawk speaker ‘would immediately recognize (1) as a single word . . . most would not be able to segment it into morphemes . . . They can, however, manipulate the parts with dizzying speed.’ We assume that Mithun means that a Mohawk speaker would immediately recognize and manipulate the sound represented by (1).2 Despite this admission that Mohawk speakers do not think about or engage with their language in terms of morphemes, Mithun does not question the pedagogical value of such interlinearization. Instead, she argues that awareness of the language’s morphology when made visible through writing promotes ‘pride in the complexity’ (2011: 10) and strengthens the language’s prospects. We are not convinced by this appeal to the ‘genius’ of the language and such a heavy investment in literacy. Gary Holton (pers. comm.) recalled Inuit ladies asking at the Alaska Native Languages Archive3 for access to particular songs. He admitted that although the songs were likely to be among hundreds of hours of painstakingly interlinearized text, they were not marked or identified and thus could not be easily retrieved. Technical detail on a vast scale had blinded documenters to cultural relevance. More recently, Holton (2011: 166) notes a general failure to archive fieldnotes and attributes this to documenters’ overemphasis on data formalization, processing, and abstraction. In an insightful description of field linguists as documenters of their own language learning, he describes the field notebook as: the very foundation of linguistic fieldwork and the documentary enterprise. Filled with observations and varying levels of meta-analysis [it] spans the gulf between documentation and description. From an archivist’s point of view, field notebooks are a treasure revealing the linguist’s thought processes as the language is revealed to her. Scribbles and corrections and mistakes trace the path of the linguist’s experience . . . To take a simple example, field notes may reveal which sounds a linguist was struggling to transcribe, permitting a better exegesis of transcribed texts.

Yet, Holton points out, linguists often fail to deposit fieldnotes with archives, i.e. they fail to treat them as language documentation. Holton contrasts this with an enthusiasm for ‘typed versions of field notes which have been highly processed and redacted’ under ‘today’s emphasis on data processing’ (2011: 166). Fieldnote contents such as comments, diagrams, or pedagogical guidance – anything not regarded as formal linguistic content – are at risk of being lost to the record. 2

3

If this assumption is correct, then a claim about an acoustic event has been presented as a claim about a written and decontextualized instance of a language system (Nathan 2009). Von Sturmer (2009) argues that this slippage is how writing ends up being more ‘correct’ than the language speaker. See www.uaf.edu/anla/ (accessed 22 March 2012).

Re-imagining documentary linguistics

2.2

45

A one-way journey

Documentary linguistics is often characterized as a fieldwork-based activity whose outputs are the inputs to language description (Himmelmann 1998: 161, P. K. Austin and Grenoble 2007: 22, P. Austin 2010a: 23). This is a characterization of a one-way journey, which starts in the documented language and leads irreversibly to the language of explanation (often English but typically a national language or language of translation/wider communication).4 The one-way journey departs from a range of socially and personally contextualized language events and arrives at a single, abstract, and disembodied language system, expressed in the language of explanation. The journey’s directionality is reinforced by the representation of language events via specialized software. Once captured in these software interfaces, at least three ‘foreign languages’ – English (or other glossing language), linguistic terminology, and computer literacy – are needed to retrieve the content of what were once spoken vernacular events. Through this one-way journey, language speakers are isolated from representations of language events, the knowledge encoded in them, and the hard-won linguistic explanations and outcomes. Given the power and prestige differential between the endangered and the ‘foreign languages’, the very causes of endangerment are re-enacted. The one-way journey is an element in the devalorization of languages. 2.3

In pursuit of the pure

Documentary linguistics tends to follow the traditional concept that languages are mutually exclusive codes. Projects typically take their task as the documentation of a particular language. Although Himmelmann’s (1998) emphasis on spontaneous, naturalistic speech seems neutral in regard to language mixing and speaker ability, documentation projects usually engage consultants who are believed to speak a particular language naturally, spontaneously, and fluently. As it is often elders who are the ‘last speakers’, their speech competencies and styles draw the attention of documenters (Dobrin and Berson 2011: 195). This association can reinforce tendencies to ‘purism’, make the ‘old language’ appear sacred and appropriate only for elders, and give younger people the message that their speech is of lesser value (Hermes et al. 2011). The flip side of a ‘pure’ and singular language is language use replete with code mixing, borrowing, and dysfluencies and other ‘errors’ – all phenomena that can be expected in situations of language retreat and replacement. However, documentary linguistics does not promote methodologies for documenting 4

Examples of such languages would include Arabic, Chinese, French, and Spanish.

46

David Nathan and Meili Fang

these phenomena. Our understanding of the path of language revitalization will remain limited without ongoing documentation that admits errors, the speech of semi-speakers, code mixing, and speech between people of differing competencies.5 The Punana Leo School in Hilo, Hawai’i, is a long-standing and innovative model for community-based language revitalization. Teachers constantly look for strategies to keep students speaking Hawaiian, despite any limitations in accuracy or fluency. In science classes, for example, value is attached to Hawaiian being used as the language of investigation, explanation, knowledge, and science. Teachers and students are encouraged to use Hawaiian at all discourse levels and domains. In such a context, the ‘one-way journey’ and the quest for the pure are unhelpful messages.6

3

Mobilizing metadata

Metadata is additional data about resources that facilitates the discovery, understanding, and management of those resources. It defines who can effectively discover and access materials, and how they can understand and use them. Thus, metadata is not neutral: choices made in its formulation constrain future users and usages. The metadata schemes most influential in documentary linguistics, the Open Language Archives Community (OLAC) and ISLE Meta Data Initiative,7 provide fixed sets of categories skewed towards descriptive linguistics and typology (Nathan and Fang 2009: 137). If metadata fuels resource discovery, then screen interfaces are the engines that drive the discovery process. The Endangered Languages Archive at the London School of Oriental and African Studies (SOAS) has encouraged depositors to provide rich and diverse metadata, and it has developed interfaces that maximize the exposure of metadata and other contextualizing information in support of a wide range of users, including those teaching and learning languages. Figure 4.1, for example, shows the homepage for Eli Timan’s collection on Jewish Iraqi (a variant of Arabic) which displays background information, along with the names of speakers and the topics they spoke about, such as apples, donkeys, henna, and so forth.

5

6

7

Exceptions include the BOLD (Basic Oral Language Documentation) methodology, which uses ‘respeaking’ of texts, including by semi-speakers. See www.boldpng.info/ (accessed 22 March 2012). These observations were made during a visit to the Punana Leo School on 15 February 2011. The visit was part of an excursion from the International Conference on Language Documentation and Conservation organized by the University of Hawai’i. We record our thanks to the teachers and students who spoke to visitors and answered our many questions. See www.language-archives.org/ and www.mpi.nl/IMDI/ (both accessed 22 March 2012).

Re-imagining documentary linguistics

47

Figure 4.1 Eli Timan’s collection on Jewish Iraqi

While archives can attempt to support revitalization in such ways, their ability to do so depends on documentary linguists supplying suitable metadata. In this section we identify categories of metadata that would further facilitate discovery and usage of materials for revitalization. Documentary metadata can identify activities and events that speakers regard as markers of their culture, such as the Alaskan songs mentioned above. Examples range from songs, rituals, speeches, and prayers, to such speech acts as greetings, common expressions, and snippets of verbal flourish. These can be annotated, time-aligned, or commented on using mechanisms similar to those used for formal linguistic content, such as ELAN or Toolbox software, or provided as structured metadata in tables or spreadsheets, or simply in narrative form using plain text. The method should be chosen to best enable materials to be identified and retrieved by teachers and learners. A rich source for pedagogical resources would be metadata that links associated materials. Links could describe potential usefulness for teaching, e.g. materials that explicate, complement, or illustrate other materials. Links could indicate alternative versions: for example, versions of stories varying in detail or complexity would be useful for different ages or levels of learners (see also the discussion of intertextuality below).

48

David Nathan and Meili Fang

Metadata can be collected on the teaching status of materials: how they could be used, and how they have been used. Pedagogical value could be immensely increased simply by marking or annotating materials within a collection that are not suitable for learners (e.g. learners of certain age, sex), thereby making the collection safe for teachers and learners. Linking documentary materials with teaching notes, and with records of who has taught and learned using which material, would create an expanding corpus of use not only for ongoing language teaching and learning but also for future researchers wishing to understand the dynamics of maintenance and revitalization. Opening up language documentation in such a way to iterative collaboration with related disciplines and practitioners was part of Himmelmann’s (1998) vision for documentary linguistics, yet as far as we know no such attempts have been made to date. Such interdisciplinary collaboration would not only provide feedback to documentary linguists about the effectiveness of their materials but also provide ongoing benefit to communities. One of the authors recently attended a community-based pre-school immersion class in Ichishkiin at Warm Springs, Oregon, USA. The experience highlighted possibilities for richer documentary metadata. Metadata categories for documenting revitalization activities such as a language class, or series of classes (for a longitudinal documentation), might include: the number of classes attended and children’s attendance rates; relationships among the children (and with the teachers); family environments; parents’ aspirations; attitudes towards being recorded/filmed; health and emotional states; teachers’ evaluation of performance in games and activities; and teachers’ and parents’ descriptions and evaluation of progress. Some of these categories could be applied or adapted to more typical documentary situations. The Endangered Languages Project at SOAS encourages documenters to create ‘overview annotations’ – a quickly and easily prepared type of metadata that serves as roadmaps or indexes of the main resources of a documentary corpus. The purposes of overview annotation are: r to provide discovery of and access to more cultural and learning resources r to ‘lower the bar’ of skill and time required to make documentation accessible (i.e. compared with interlinearizing) r to prevent large amounts of recorded material being rendered inaccessible by never receiving any metadata or annotation at all. Overview annotations consist of roughly time-aligned information about what is in a recording: who is participating, topics, genres, and notes of the presence of interesting phenomena. Here is a hypothetical example: from 1 to 3 mins Auntie Freda is singing the song called ‘Fat frog’; from 3–7 mins Harry Smith is telling a story about joining the army; from 7–10 mins there is interesting use of applicative morphology; from 15–18 mins contains rude content that should not be used for teaching children

Re-imagining documentary linguistics

49

Table 4.1 Overview annotations Segment number

Time (mins)

Speaker

Topic/content

1 2

1–3 3–7

Auntie Freda Harry Smith

song: Fat frog story about joining the army

3

7–10

Harry Smith

interesting use of applicative morphology

4

15–18

Joe Brown

Comments

rude content, unsuitable for teaching children

This could be written as prose (as it is here) or, better, structured into a table:8 see Table 4.1. Such annotations can be created in ‘real time’, i.e. no longer than the duration of the recording, and perhaps much shorter. There is no special software, schema, format, or cryptic terminology needed. They allow metadata to be relative to the particular resource (and, in turn, relative to the speaker community, the individuals involved, the particular project goals, and the interests and skills of the documenters).9 They can (unlike technically glossed materials) be translated into a range of languages, including the documented language. They can be open, accessible, shared, multidisciplinary, and iterative. If overview annotations became the sine qua non for language documentation, far fewer resources would remain hidden and inaccessible to those wanting to mobilize them for revitalization. 4

Filling the gaps

A number of language phenomena that are often overlooked could be usefully documented and used in support of revitalization. Songs, including simple songs such as those sung to and by children, are outstanding resources and have provided the foundation for revitalization (Amery 2000: 148). The ‘stickiness’ of their melody and language makes them a fundamentally human way of ‘archiving’ language. The expression of emotion is rarely documented. Emotion-bearing expressions are linguistic resources that can be regularly used by speakers in a 8

9

No special software is required, but those familiar with such software as Transcriber or ELAN can create overview annotation by marking time breaks as required, and typing descriptive text into the segments between breaks. Another strategy is to simply type a number into the timealigned segment and then create a table which links each number with its overview information, as illustrated in Table 4.1. A departure from the standard approach to metadata which places higher priority on standardized category schemes.

50

David Nathan and Meili Fang

revitalization situation. Nathan (2006b) described Kamilaroi elder Auntie Rose Fernando’s frustration during a workshop attempting to ‘modernize’ her language with words for refrigerators and the like, while useful and flexible expressions such as ngarigaa (meaning something like ‘woe is me’) – that can be frequently uttered – remained undocumented. We find little discussion of methodology for eliciting, recording, or annotating emotion. Despite the considerable resources that have been put into the conception and construction of stimulus materials for investigating spatial, narrative, and other properties of speech (Luepke 2009), we find no comparative effort expended in creating stimuli for evoking expression of emotion, or indeed generating any kind of learning materials. The speech of learners has already been mentioned as an example of oftendeprecated ‘impure’ language. However, the productions of learners, for example new texts such as stories, dramas, and poems created during language classes, have been shown to provide significant ongoing motivation for young people’s involvement in revitalization. For example, our work in the Karaim community in Lithuania together with Eva Csat´o saw summer schools culminate in the creation of unique, new drama pieces by heritage learner groups, which were then performed in front of elders as judges and prize-givers (and a large number of other Karaim community members; the summer schools eventually became annual community events). Examples such as a ten-minute drama about Karaim gangsters created and performed by a family (across two generations) point at new senses of ‘intergenerational transmission’. The performances were filmed and distributed throughout the community (Fang 2011). Such work not only results in the generation of new contemporary texts but also provides longitudinal documentation of the course of revitalization, both individually and collectively. Recent trends to collaboration with, and training of, community members (Grinevald 2003, P. K. Austin and Grenoble 2007, Rice 2011) are frequently focused on engagement in transcription and the use of such software tools as ELAN and Toolbox. The value of such training might be limited because it directs community members along the ‘one-way journey’. Alternatives could include in-community training on how to make great recordings, how to be teachers, or how to create pedagogical materials (of course these would require the trainers to have the appropriate skills and experience). Documenters could help potential teachers to receive recognition or certification by formalizing or certifying locally provided linguistic training (Shaw et al. 2012). Priority could be given to documenting language needed for the teaching process, such as for organizing and explaining class activities (Julian Lang, pers. comm.). Documentation could extend to identifying who likely learners would be, domains/contexts where effective language learning might take place, and who might make great teachers.

Re-imagining documentary linguistics

COMMUNICATION 1. LEARN THE FOLLOWING WORDS: *ngapi ‘I’ (or nga:pi) ngarni = we (all) nginti = you (singular) nguni = you all nganawi = my (= ngan ‘me’ + -awi ‘possessive’) namawi = our (= nam ‘us’ + -awi ‘possessive’) *le:wun = sitting, living somewhere *ruwi = land Pomberuk = Murray Bridge Raukkan = Point McLeay -angk = to, at (a place) (–angk is a suffix placed on nouns)

51

*mimini = woman, female *ko:rni = man, male *pa:mi = young girl *po:rli = child, young one *tinyeri = child, baby yarteki = teenage girl (S) narumbi = teenage boy (T) *mi:tji = name *nga:tji = totem *krawi = big (K) *minyuwan = small1 *plombi = ear (s) *plombatji = deaf (literally: ears + -atji ‘without’)

Figure 4.2 Ngarrindjeri learner wordlist with words known by elders marked

Knowledge other than the strictly linguistic can be built into documentation materials in support of revitalization. An example comes from pedagogical materials prepared for Ngarrindjeri learners in South Australia. Mary-Anne Gale (pers. comm.) reported that in the early days of Ngarrindjeri language revival activities in schools, some teachers wanting to defer to elders’ knowledge drafted a curriculum wordlist where certain words were marked (by asterisks) to indicate those which could be taught (see Figure 4.2). This could be called a kind of ‘emic protocol’ that made materials safe and acceptable from a community perspective. Another kind of knowledge lies in intertextuality, the relations or links between texts and parts of texts on the basis of associated content or meaning. Familiar examples abound: a dictionary links to texts in a language by listing its lexical types; a grammar lists syntactic types and patterns; and a concordance links lexical types or tokens in combination. Such intertextual resources provide navigable knowledge structures that strongly benefit many kinds of learner (Snyder 1997). While documenters construct and exploit intertextuality when they use software such as Toolbox to write dictionaries and to gloss and analyse texts, many of those intertextual properties can be lost when documentation resources are handed to communities (see Section 6 below). Paradoxically, descriptive rather than documentary materials may have greater pedagogical value. Interactive multimedia, delivered on CD, online, or on mobile devices, is a form of documentation which can provide language learners with accessible intertextuality through richly linked content. Well-known examples which

52

David Nathan and Meili Fang

are widely distributed in their communities and have made positive contributions to language revitalization include Spoken Karaim (Csat´o and Nathan 2004) and Gayarragi, Winangali (Giacon and Nathan 2009), both created using the games platform (Macromedia/AdobeTM ) Director.10 Two recent examples, using (AdobeTM ) Flash, can be found on Stuart McGill’s and Eli Timan’s websites.11 Such resources are frequently dismissed by the message that documentation must avoid proprietary linguistic formats (Bird and Simons 2003). But useful materials for language maintenance and revitalization must be distinguished from preservable materials for archiving. While some proprietary formats may not be usable in the future, documentation materials held in Toolbox or ELAN are frequently not usable right now. The latest mobile or iPadTM apps, even if in proprietary or transient formats, may motivate and support learners today, and encourage other language activities, regardless of how well those applications are likely to work in fifty years (Nathan 2006a). The major funders of language documentation have not encouraged or resourced research on and documentation of revitalization methodologies and processes.12 Although both the Endangered Languages Documentation Project (ELDP)13 and the Dokumentation Bedrochter Sprachen (DoBeS) or ‘Documentation of Endangered Languages’ claim to take revitalization seriously, they have funded almost no projects which explicitly aim to document the course of revitalization, and the other major documentation funder, the USA Documenting Endangered Languages program (National Science Foundation/National Endowment for the Humanities) rules out support of revitalization.14 5

Teaching and learning

We have argued that pedagogically useful materials are weakly fostered – or are even discouraged – by some tenets and practices of language documentation. Learning from documentation materials has been compared to ‘learning English from Shakespeare’ (Hermes et al. 2011), and Hinton (2011: 316) complains that ‘people trying to use documentation for language learning must become researchers’. Although the situation might be improved if more language documenters were trained as language teachers, to ask this of them can be asking too much 10 11 12

13 14

Gayarragi, Winangali can be downloaded from http://lah.soas.ac.uk/projects/gw. See www.cicipu.org and www.jewsofiraq.com/ respectively (both accessed 22 March 2012). Although there has been some work by individuals: for example, Crystal Richardson, Karuk Tribe member (northern California), filmed her participation in master–apprentice programme sessions. For ELDP see www.hrelp.org/grants (accessed 22 March 2012). See www.nsf.gov/pubs/2011/nsf11554/nsf11554.htm (accessed 22 March 2012).

Re-imagining documentary linguistics

53

of an already highly burdened and multifaceted task. Rice (2011: 191) speaks of such cross-overs as a confusion of ‘community-based research’ with ‘community service’. Crippen and Robinson (2013) raise the spectre that documenters might do more harm than good through unskilled teaching. However, assuming that documenters need no pedagogical training is as erroneous as proposing that they need no training in syntax, typology, or ethics. Documentation is regularly the source of revitalization materials and will be even more frequently as languages cease to be transmitted in families. Documentation needs to be pedagogically as well as linguistically informed for downstream use. Therefore, documenters should at least have access to basic training in language pedagogy so they know something about teaching and learning materials, and can work effectively with teachers. Wampanoag revitalization expert Jessie Little Doe Baird (Baird 2012), when asked which of all the linguists’ skills are most useful, replied ‘pedagogy’. Neither should documenters underestimate the revitalization task. Unfortunately, many documentation project proposals claim to contribute to revitalization without presenting even basic research into a community’s aspirations for its language, the infrastructure and personnel available for teaching, who the potential learners are (and their motivations, skills, and so forth), or any demonstration of the documenter’s ability to develop usable materials or teach a language. Language documentation and language teaching/learning can be mutually informing, building a more scientific and effective basis for each (Nathan and Fang 2009). Ideally, documentation will be appropriate to the intended pedagogical approach; different materials would be needed for a master–apprentice approach, a language nest, or school-based learning. Or an indigenous pedagogy may provide the framework.15 Documentation can gain more from engagement with revitalization than just recordings of revitalization activities. Darryl Baldwin, a Miami community member active in revitalization of his language, stated that ‘we learned as much from the reclamation [process] as from the language itself’ (Baldwin 2012). Baldwin sees his community’s revitalization goal not as producing a generation of fluent speakers but of changing attitudes to facilitate a future fluency, which will be different from their ancestors’. In other words, the language may not be authentic (in some sense), but the process of learning indisputably is. In this situation, documenting the longitudinal process (as Baldwin is doing) may ultimately be more linguistically enlightening than describing an ephemeral ‘language’. Lewis and Garrett (2012) found that ‘the analytical burden of description for language learning is often higher than for 15

Punana Leo teachers spoke of ‘an indigenous Hawaiian pedagogy’. See footnote 6.

54

David Nathan and Meili Fang

theoretical or typological research’ (our emphasis).16 Similarly, Shaw (2012) found that ‘curriculum development forced us to re-prioritize research in certain domains’ in order to understand particular structures. 6

Ethics and communities

Increasingly, we talk about ‘giving back’ to communities. Rice (2011: 198ff.) characterizes giving back as not taking, or stopping taking. A fairer criterion for giving back is reciprocity or relative value. Rice asks ‘what it might mean for such work [i.e. research] to be of value to a community’ and describes examples, ranging from leaving copies of research produced to ‘working with people to meet their goals of developing curriculum materials’ (Rice: 2000). Sometimes giving back consists of not much more than converting audio files to MP3 format and distributing them on CDs, or burning a few videos onto DVDs. There is a great difference between this simple media copying work and the sustained effort required to learn how to create effective learning materials. Simply exploiting access to specialized equipment is a very weak form of reciprocity, especially when on the other side of the equation is the value of languages to their speakers, which for some is as much as life itself: ‘Without the language, we are warm bodies without a spirit.’17 Giving back could be thought of as ‘digital repatriation’, rather like the return of cultural property by museums.18 For example, community members’ voices could frame, contextualize, and present the documentary materials. This comparison reminds us that such voices are, paradoxically, largely silent in much of today’s documentary practice. 7

Conclusion

Bird and Simons (2003) argue that documentary methods make linguistics accountable. Transcriptions and claims based on them can be checked against primary data if that data is explicit and accessible. However, accountability need not be restricted to formal linguistics. Documentation’s contributions to revitalization can be accountable as well, perhaps to meeting community aspirations, or to defined learning outcomes. As Rice (2011: 188) notes, ‘accountability with respect to speakers, communities, and, when language transmission is diminished, would-be speakers, has become a high priority as well’. Application 16

17 18

Lewis and Garrett (2012) spoke of the challenge of explaining to learners about language functions that would normally appear in different parts of a standard grammar, such as adverbs and prepositions (both of which are needed to express direction and location). Elder Mary Lou Fox, quoted in Czaykowska-Higgins (2009: 15). See http://digitalreturn.wsu.edu/workshop/ (accessed 22 March 2012).

Re-imagining documentary linguistics

55

criteria for grants under the DoBeS programme require that funded documentation ‘can be used for language maintenance and revitalization by the speech community’, but, as we have indicated throughout this chapter, documentation does not yet have the means to be accountable to that requirement. Taken together, the observations in this chapter offer several challenges to conventional documentation theory and practice: 1. Documentation should factor the needs of revitalization into its methods and outputs. 2. Revitalization contexts and prospects need to be researched first and fed into the formulation of documentation proposals. 3. Revitalization needs to proceed in parallel with documentation. Documenters need feedback from their contribution to teaching and learning activities to evaluate its effectiveness. 4. A holistic documentation will embrace revitalization outputs (such as learner-created texts, teachers’ plans, methods, evaluations) as part of the documentation corpus. 5. There should be vigorous documentation of revitalization processes. In 100 years’ time, after language loss has taken its predicted course, we will be judged derelict in our duty if we have not documented language maintenance and revitalization efforts, both successful and unsuccessful. 6. Documentation and description are not two sides of the same (single) coin. Documentation and revitalization are equally as interdependent. Description, typology, and theory ought not to have priority over documentation’s methods and outputs. Modern language documenters took up the baton of language endangerment nearly twenty years ago but have not yet reached a consensus about documentation’s role in language transmission. Much of what could be documented in support of languages has been overlooked in favour of documenting disappearing vocabulary and linguistic structures; this chapter has presented options for documenting the life, as well as the death, of languages.

5

Language documentation and community interests John Henderson

1

Introduction

The language documentation literature has called for richer, more diverse documentation as a more useful representation of a language – a multifunctional, long-term documentation (Himmelman 2008) – and there is a significant view that both the process and the product should be sensitive to the social and cultural context (Hill 2006, Himmelman 2006). One aspect of this is the representation of context in meta-documentation – information about the primary data – examining how the research itself and its context can best be described in order to maximize the value of the data (Good 2010). Speakers and communities can also have a role in determining the content of the meta-documentation in order to serve their present and future community interests. Facilitating such community meta-documentation is both an ethical and a practical issue for researchers and archives. It is also an opportunity to investigate and document the linguistic culture of the community, which can offer additional benefits both for future community use and for scholarly research, where it can contribute to knowledge of how endangered language communities understand and respond to their linguistic situations. This chapter examines how community interests can be represented in meta-documentation and analyses a project involving legacy materials, Laves’ 1931 fieldnotes on Noongar, an endangered language of the south-western corner of Australia. The themes that are identified are relevant both to other cases involving legacy materials and to current language documentation projects.

The protocol project was conducted under a grant from the Australian Institute of Aboriginal and Torres Strait Islander Studies (AIATSIS). I thank the members of the Noongar community reference group and my Noongar co-researchers: Hannah McGlade, Kim Scott, and Denise Smith-Ali. In addition to the Cambridge International Conference on Language Endangerment, some aspects of this chapter were covered in a presentation at SOAS, London, in June 2006. I thank Peter Austin and other participants on both occasions for their comments.

56

Language documentation and community interests

57

Meta-documentation offers opportunities for speakers and their communities. Their participation in the design of the language documentation process and in determining the nature of the products is widely accepted as an important principle in language documentation (Himmelman 2006, Rice 2006), and it has given rise to practical and ethical questions for researchers and archives (P. K. Austin 2003, Dwyer 2006, Penfield et al. 2008, Himmelman 2008). The question now arises as to what role speakers and communities will have in determining how a project is documented, and specifically how this can serve their interests. This includes the question of whether researchers can combine this community meta-documentation with their own purposes, given that the community may have different aims and different ways to achieve them. Viewing these questions as specific goals of language documentation should lead to a more systematic approach that can enhance outcomes for both communities and researchers. While rights of various kinds are a central issue in language documentation, there is also value in considering what the broader range of community interests is, and to what degree these can and should be recognized in community meta-documentation. Negotiations over the content can be an opportunity for the community to consider wider and longer-term language issues beyond the immediate project, and may be a focus for ideological clarification (Kroskrity 2009). A basic question is the function of community meta-documentation for a language documentation archive and future users. It might primarily be seen merely as allowing a community to express their ideas, or for future research on community perspectives, but not as central to the future uses envisaged by the archive or researcher. However, if there is an ethical responsibility to ensure that the interests of the consultant and/or community are at least taken into account in the use of language documentation materials, then the community metadocumentation will need to be made central to the management of the materials in the archive by linking it to the archive’s procedures for managing access. This may be as simple as asking potential users to read the community metadocumentation before requesting access, or it may involve some additional requirement in the archive’s access protocol for the specific materials. Either way may be a challenge to more traditional models of archive management and to their resource limitations. Regardless of whether the community interests expressed in the metadocumentation are formally incorporated into an archive’s protocols, if community meta-documentation nominates goals or forms of consultation, acknowledgement, or other processes that the community values, then it can be described as a community protocol for the materials. The following sections analyse the context and content of one instance of this type of community meta-documentation, the community protocol developed for the Laves 1931 Noongar materials (Scott et al. 2006).

58

John Henderson

2

The Laves 1931 Noongar materials

Gerhardt Laves (1906–93) was an American graduate student who conducted field research on a number of Australian indigenous languages from 1929 to 1931, and produced a substantial body of handwritten fieldnotes (Nash 1993, Bowern 2003). His research included roughly two months’ work on varieties in the south-west that are now recognized as dialects of Noongar, almost all from a variety he labelled Kurinj or Kurrinj, spoken in an area extending east from Albany. Laves was probably the first researcher to apply modern linguistic analysis to Australian languages, and his work is of relatively good quality. His Noongar fieldnotes consist of 10 notebooks with 1,570 (sides of) pages in total and contain mostly texts with varying degrees of interlinear glossing and cultural notes, including genealogies of his Noongar consultants’ families. There are 192 texts of varying length. Nearly all are attributed to 13 named speakers, 6 of whom provided most of the material. There are also 2,453 loose slips which contain vocabulary and analytical notes apparently derived from the notebooks. Laves abandoned his research without much in the way of published results, or even unpublished analysis (Nash 1993). The notes resurfaced in the mid1980s and were deposited in the archival collection of what is now the Australian Institute for Aboriginal and Torres Strait Islander Studies (AIATSIS), a government agency. Research since then has made some use of Laves’ work on Noongar (Dench n.d.), but little of this has resulted in publication. In 2003, AIATSIS started digitizing the pages to colour images in order to make them more accessible, potentially online, and then also commissioned a project to keyboard the materials to increase readability and to make them available in a more useable digital form (Henderson 2008). Noongar language has been severely affected by colonization and displacement by English. It is in very limited everyday use today, although the last three decades have seen a number of revitalization activities in various parts of the region. School programmes have been a particular focus, and a small range of resources has been developed to support those programmes. The Noongar area is large, close to 200,000 square kilometres, and the language community is diffuse, both geographically and socially. The Noongar community may have as many as 30,000 members (SWALSC n.d.). It is difficult to determine with any precision how many people know and/or use the language and to what degree, but, to judge from the reports of community members and researchers, there are no speakers who communicate over the full range of everyday language functions using Noongar. Most reports suggest that, with the exception of a few individuals, only a small proportion of the resources of the language are in everyday use. Description of the language is still relatively limited, and both the historical sources and some non-technical modern sources vary enormously

Language documentation and community interests

59

in quality and are frequently inconsistent in their written representation. In this context, Laves’ notes are a significant resource for the language and culture. 3

The Laves’ Noongar Protocol

Recent community involvement with the Laves materials arose from the AIATSIS proposals for digitization. After the initial approach by AIATSIS to community members, the main consultations were conducted by a research team consisting of the author and three Noongar co-researchers: Hannah McGlade, Kim Scott, and Denise Smith-Ali. A community protocol document was developed in discussions with and among community members over a three-year period. The major topics addressed by the Laves Protocol are: r the audience and goals of the Protocol r the significance of the materials to community members today r copyright and the rights of families today r the need for senior community members to be consulted about the use of the separately authored parts of the materials r the family connections between Laves’ 1931 consultants and senior community members today r the historical notes on Laves’ consultants in 1931, focusing on memories in the community today r the handling of culturally sensitive content r a ‘biography’ of the project, especially the details of the community consultations r the relationship between the language, language labels, people, and country r the history and nature of the materials, including issues of identifying individual authors and interpreting the content r plans and suggestions for future community use of the materials. The content of these topics can be analysed in terms of a number of interrelated themes, discussed in the following sections. Some of these are familiar from the literature on language documentation and linguistic field research, generally as issues to be aware of in planning a project, but the focus here is on their place in community meta-documentation. 3.1

Interests addressed

The fact that Laves’ notebooks mostly identify by name the Noongar consultant who provided each text in 1931 has allowed the possibility of identifying their descendants today. The Protocol recognizes a range of different interests but focuses on the interests of the specific descendant families, their senior members, and the communities around them. The Noongar conception of family is the major form of social organization in the Noongar area: families have an

60

John Henderson

average size of 200–300 (but may consist of a number of smaller family groupings), with members distributed across overlapping areas within the region (Birdsall 1994). As might be expected, relationships between the different families are an important aspect of the social context. The central interest of descendants is in their respective parts of the materials as an inheritance and in the future uses they envisage. The Protocol lists senior representatives of the descendant families to be contacted. It deals less specifically with the various kinds of interest that other Noongar people, including those from other areas, and (outside) researchers might have. Both these groups are explicitly listed as intended audiences, and it specifies a way for them to pursue their interests by contacting the senior contact people as an initial step. Opportunities for researchers in areas such as lexicography are implied by the wish-list of future uses, but the Protocol otherwise leaves this open, with all such details to be negotiated with the senior members of the families. In the general case, there may be value in discussing how future research interests might be negotiated, for example in quid pro quo agreements where the differing interests of both parties can be served. The question of interests also needs to be considered in relation to the social and political context. An obvious aspect of the context of the Protocol project was the socially complex claim by Noongar groups under Native Title law which was under way at the time (Barcham 2008). Competing interests among claimants in other parts of Australia have sometimes had a negative effect on language documentation and language maintenance (Henderson 2002), but whatever Native Title issues might have been in play during the Protocol project, they did not stop the negotiation process or prevent agreement on the content of the Protocol. The Native Title claim is in fact not mentioned in the document, and it remains to be seen whether this will have any significance in the longer term. 3.2

Validity

An important requirement for the effectiveness of a community protocol is that it should be seen by all the audiences to which it is addressed to be a valid expression, although realistically it may well not find validity in the eyes of all parties without at least some disagreement. Validity contributes to the cogency of whatever appeal the protocol makes to others, and to its informational value, but importantly it can also help to protect the reputation and interests of the community members who are involved in the language documentation project or in developing the protocol. Involvement can create a position of social gain for them, but it can also be a social risk. A protocol of this type needs to show that it properly represents the views of all those individuals and institutions who will be accepted by the widest

Language documentation and community interests

61

range of people as having both a valid interest and an authority to speak. The potential difficulties of identifying what constitutes the relevant community and of finding a consensus view are well recognized (e.g. O’Meara and Good 2010), and in some cases a researcher or archive may only be able to resolve this with a fairly narrow group. Determining who has a valid interest will depend both on cultural principles and on social and political factors, but to extend the lesson drawn by Wilkins (1992), it cannot rely just on an outside researcher’s assumptions. The Laves Protocol attempts to establish validity by showing that the negotiations and representation took place in ways that meet community norms. It identifies the community members who were involved, and on what basis. It also details the cycle of consultations that took place, starting from the question: who should be consulted? From initially working with elders who were connected to the Noongar members of the research team, progressively wider groups of people were consulted in order to form a community reference group to work towards a consensus on the materials. Despite the possibly bureaucratic flavour of that description, this type of process of consulting across the community to identify the relevant elders and others can be an important process in Aboriginal negotiation. Another aspect of validity in the Laves Protocol is in indicating that proper consideration was given to issues of cultural sensitivity, especially those which might restrict some material according to gender, age, or family membership. It details the arrangements under which men with appropriate authority from the different families were asked to assess the materials, given that men provided most of Laves’ information. 3.3

Authority

Validity will typically require that some accepted authority is represented. The primary authority in this case is with the closest senior descendants in the families of Laves’ consultants. However, identifying the appropriate authority was not always a simple matter, given the history of enormous social and cultural disruption since the arrival of Europeans (SWALSC, Host, and Owen, 2009). In some cases the family connections back to Laves’ consultants are fairly clear, but in other cases even Laves’ genealogical diagrams do not help. If a direct descendant could not be found, then a close family connection was identified. However, for a couple of Laves’ consultants (who were responsible for a very small proportion of the materials), the consultation process was not able to identify any family connections. The Protocol has nothing specific to say about these materials, although the implication is that if descendants are able to be identified later, then they should have authority over them.

62

John Henderson

There are also other kinds and degrees of authority. For the Noongar families, senior people are assisted by family members or others in various ways for the purposes of the Protocol. These range from informally looking out for their interests through to something more like a delegation of authority. These kinds of role cannot always be spelled out in detail because they might be left relatively open or might change over a project, but it can nonetheless be valuable to identify in a protocol all the main people involved and their roles. The term ‘senior contact person’ is used to group together various authority roles as contact points for their respective families, rather than identifying a single owner of each part of the materials. As noted in the previous section, authority also plays a role in relation to issues of the cultural sensitivity of the content of the materials. Notably, the arrangements in the Protocol do not operate through the authority of community organizations or institutions that are active in various ways in the region. This approach was informed by the Noongar members of the research team and was vindicated by the fact that a wide range of people have participated without suggesting that their participation should be through an organization. Ethical guidelines for research, such as AIATSIS (www.aiatsis .gov.au/research/ethics/), often recommend operating through community organizations that can assist the community and individuals to protect their interests; however, alternative processes can achieve the same goals. A different kind of authority can come from expertise or professional involvement in a relevant field, particularly where the people with primary authority might not have expertise in that field. Some community members who were actively involved in teaching Noongar and developing educational resources were also invited into the reference group on this basis. 3.4

Recognition

Meta-documentation can be a form of recognition for speakers and communities, just as the language documentation process and its products can also be, even if such recognition can also give rise to disputes within a community. Minority groups with an endangered language and a history of social and cultural disruption often place particular importance on the public acknowledgement of their existence, identity, situation, or rights, and the positive value of their language and culture that is made explicit, or just implied. The Laves Protocol gives general recognition to the overall Noongar community, the Kurin area families, and to specific descendants. It gives recognition to the specific individuals who have authority to undertake any consultations and who have continuing roles in the plans and recommendations for future activities. It also recognizes the importance of Noongar language and culture to the community today, and the value consequently placed on the Laves materials. An act of

Language documentation and community interests

63

recognition at the end of the Protocol project was a formal presentation or ‘return’ of copies of the different Laves materials to the senior members of the respective families. 3.5

Persuasion

A central function of a community protocol is to try to influence future uses, and perhaps the language as a whole. This might be achieved through copyright if it is held within the community, but otherwise any influence can only be by persuasion. The issues of copyright were not considered in any depth in the Laves Protocol project (although this would presumably need to be examined if any of the texts were to be published in substantially the same form). There is recognition that Laves had copyright in the material (which presumably passed to his descendants on his death). The Noongar contributors presumably also had copyright, as joint authors, in the texts component and possibly other content, as opposed to Laves’ analysis, and this copyright has also presumably passed to their descendants. However, a formal legal opinion was not obtained, and the Protocol does not explicitly assert any claims based on copyright, a fact which may reflect a reluctance to use legal mechanisms in Noongar cultural matters. Instead, the Protocol attempts to persuade the range of interested parties, although of course attempts to persuade are not necessarily successful, and it may be a matter of how closely the Protocol’s goals can be achieved. The archive is an important audience because, if it accepts the claims of a community protocol, it could support the protocol’s aims through its access procedures. A central issue then is how the implementation of a protocol’s goals is distributed between the community, archive, and researchers. The Laves Protocol does not attempt to frame full access conditions for the archive; nor does it specifically require community permission for access or use: it says only that ‘the senior contact people should be consulted at all stages by anyone wanting to use the field notes’. A separate request from the reference group asked that these people should also be consulted by AIATSIS in relation to access. In fact, AIATSIS rejected the Protocol as a basis for access management, primarily because they interpreted it as claiming a legal right to control access that was inconsistent with their legal responsibilities.1 They also rejected a direct role for the various named senior contact people, instead offering an informal advisory role for one or two contact people. This satisfies the general goal of the Protocol for community consultation, even if it falls short in the details. It is perhaps not surprising that the archive would want more of the 1

AIATSIS has specifically requested that ‘the Protocol not be presented as an ideal way of raising concern over the protection of cultural interests’ with that institution.

64

John Henderson

responsibility to rest with a very limited number of community members, and for them to manage the consultation with the senior contact people in the various families, and therefore presumably reduce the burden on the archive. At the time of writing, AIATSIS has not catalogued the Protocol document and does not refer to it in the catalogue entries for the Laves materials. This misses one kind of opportunity to bring the broader goals of the Protocol to the attention of potential users; however, AIATSIS staff have provided a copy of the Protocol to people who make a specific inquiry about access to the Noongar materials. The Protocol is also attempting to persuade future Noongar communities, whether in fifty years or tomorrow (as Hill (2006) points out, the situation in a language community can change quickly). The wider Noongar community today is a key audience, that is, members of the descendant groups who were not involved in developing the Protocol, as well as members of other Noongar families. Since the Noongar community is relatively large (for an Australian Indigenous group) and diffuse, it cannot be assumed that everyone in this wider community is represented by the Protocol or even aware of it. 3.6

Flexibility

If a protocol is to remain relevant over the longer term, it has to allow for flexibility of various kinds. The Laves Protocol allows for more community members developing an interest in the materials by specifically directing them to the nominated senior contact people. And because key people may pass away, change their interests, or experience a change in their circumstances, and organizations may close down, the Protocol attempts to provide multiple paths to the families. In addition to listing the names of senior contact people today, it contains statements that detail their family connections to Laves’ consultants. Some community members initially felt as if they were being required to justify their family connections, but another participant pointed out that, if publicly available, it could help Noongar people who have become disconnected from their family history to see how they might relate to the consultants’ families and the materials. Interests, ideas, and values may also change over time, and it is possible that at some future point the families might decide that they no longer need to be consulted, but might nonetheless still want users to be informed of the relationship between the community and the materials. Maintaining a community protocol over time would presumably be at some cost to an archive, but if it does not take an active role in maintaining community connections with individual materials it can risk criticism or suspicion from the community. This occurred, in a minor way, in the case of the Laves materials. Researchers and some individual community members had been aware of the existence of the materials since the late 1980s, and some had paper copies. However, when the digitization project was initiated, some other community

Language documentation and community interests

65

members expressed surprise that this resource had been sitting in the archive for over fifteen years before anyone informed them. Clearly, there can be different expectations about the responsibilities of archives. 3.7

Language identity

Noongar was a large complex network of dialect variation before the arrival of Europeans, and despite the negative impact on the language since then, dialect variation remains. An abundance of dialect names, and their associated territories, have been reported in various sources (Thieberger 1993/1996); however, in many cases the function of these names, and the validity of the regions attributed to them, is open to question. In scholarly work, G. N. O’Grady, Voegelin, and Voegelin (1966) recognize thirteen dialects corresponding roughly to Tindale’s (1974) ‘tribes’. However, Morphy (reported in Thieberger 1993/1996) recognizes just three major dialects on the basis of linguistic differences, while Dench (1994) distinguishes six regions for lexical forms. For Noongar people today there is generally recognition of different dialects of Noongar associated with different social groupings, primarily different families from different parts of the region, although there are some different views on the exact distinctions and labels. Despite that, there is clearly a sense within the community that traditional groupings are being continued. There is recognition of a common language under the term Noongar (with minor variation in both the pronunciation and spelling of this label). This use of the term appears to be an innovation: variants of the term ‘Noongar’ occur in Laves’ notes and in other earlier sources with the meaning ‘man’ or ‘person’, but not as a language or group label. There does not appear to have been any other name for the overall language. Laves’ material predominantly covers an area in the south-east of the Noongar region, from Albany to Esperance and Israelite Bay. He identifies the dialects/groups in this area as Kurinj (or Kurrinj) and Minong. These terms are confirmed to some degree in other sources, and as used by some Noongar people, although the descriptions do not always agree in detail. Some members of the reference group describe themselves as Wilomin Noongar, and others identify as Wanjurri and/or Kurin. Since dialect and group labels can have a special significance both within the community and without, the Protocol provides some discussion but cannot resolve all the issues. Part of what gives the materials their special significance is that the language is linguistically distinct in some ways from other Noongar areas; however, the question of any linguistic differences within the Laves region did not come out as a significant factor in the project. The reference group gave approval for a common wordlist, extracted from all the texts, with no requirement to distinguish the different speakers and their specific areas, and therefore no distinction between potentially distinct varieties.

66

John Henderson

The Noongar situation generally fits a common Australian pattern of language affiliation (Merlan 1981, Sutton 1997), where a group collectively owns a language of which individuals inherit ownership through their particular family line. Language ownership has a particular significance for languages like Noongar which are not in substantial use today. There are many people who can claim Noongar as their language but who are not able to speak it to a substantial degree. Ownership involves rights which may cover control of the acquisition and use of the language by non-owners and a right to be consulted on matters involving the language. In the Laves case, the families decided in the Protocol process to allow access only to family members, but in most cases they have indicated that this would probably be a temporary arrangement to allow the community a head start. 3.8

Language change and document reliability

Language documentation can raise the question of how members of a language community perceive linguistic variation, especially differences between documentation products and their perception of their own language. In the Laves case, the materials were recognized as ‘old language’, but the Protocol does not explore the implied changes over time, and there is little detail on how the old language should be reconciled with modern knowledge. There was no suggestion that people should abandon anything that differed from the old language, rather that it is an additional area of the language to learn about: ‘good to see [the old language] written’ and ‘have the old language alongside the words . . . today’. With legacy documentation of this type, perceptions of language change depend on the perceived validity of the documentation. There was little discussion by community members of this aspect of Laves’ record: the working assumption seemed to be that it is a generally accurate representation of the speech of the consultants, if sometimes hard to interpret. 3.9

Documentation/description versus content

Some researchers see language documentation primarily as records of language use which constitute data for analysis. Speakers and communities often see them primarily in terms of the story or social act, and this can make texts and similar content fundamentally different from records of elicitation or dictionaries and grammatical descriptions – the language in abstract. The Laves reference group made such distinctions in focusing on the texts over the note slips (on which the Protocol says little) and in agreeing that the selected wordlist and preliminary grammatical analysis could be extracted without distinguishing the individual contributions of each of Laves’ consultants. By contrast, they decided that each

Language documentation and community interests

67

text should be the property only of the respective descendant families of the individuals who provided that text to Laves, and that the texts should not be distributed generally to all families. This might seem a natural distinction from some cultural and legal perspectives, but at least in some parts of Australia individuals have seen words as property, just as a particular text might be. The Protocol does not discuss whether there may have been any form of wider community ownership of the traditional stories that are represented in the Noongar texts. In some cases, similarities were identified with stories known today, but the texts as Laves recorded them were seen as the clear property of their authors. This division of rights in the texts component according to the respective authors, and therefore the effective division of the materials for consultations on access and use, has received a negative response informally from some researchers. For linguistic and textual analysis there is clearly an advantage in analysing the materials as a whole, since that would allow the aggregation of information across all the texts (and probably result in a better analysis of each text) and a better understanding of Laves’ documentation practices. The main concerns expressed are practical ones, that it increases the time and effort required to access the data and that such a complicated process is less likely to be successful. 3.10

Future use: revitalization

An important part of the Protocol process has been the opportunity to develop general plans and suggestions for future use of the materials, and more broadly for the future of the language. This resulted in some immediate products that were developed from the materials during the project (a grammatical description (Wykman 2005) and a selective wordlist). Co-researcher Kim Scott has since worked extensively with a number of families on the Wirlomin Noongar Language and Stories Project, which has included community workshops that were an opportunity both to learn more of the language and develop a number of the texts for publication. It is expected that once a text has been through this workshop process, there will no longer be a need to restrict access to that part of the materials to family members only. Other projects from the Laves materials are still planned, but these will depend on the availability of people, skills, and resources. 4

Conclusion

If we accept that consultants and communities should be involved in the design and execution of documentation projects on their language, then the

68

John Henderson

meta-documentation should also be produced with their participation. This chapter has identified the kinds of interest represented in meta-documentation for one endangered language community. Analysis of the Laves Protocol reveals themes that are relevant to community meta-documentation in broader contexts.

6

American Indian Sign Language documentary linguistic fieldwork and digital archive Jeffrey E. Davis

1

Introduction

Historically, the vast geographic expanse and extreme linguistic and cultural diversity of North America contributed to Native American groups speaking mutually unintelligible languages. As contact between American Indian communities and nations evolved, there was frequent contact between speakers of numerous mutually unintelligible languages. It has been posited that indigenous sign language emerged in order to mediate this contact and language divide. In fact, during the nineteenth and the early part of the twentieth centuries, the use of sign language among American Indians was so common that it was considered a lingua franca. Based on historical and contemporary evidence, linguists and anthropologists have reported that sign language was used in varying degrees across the major cultural areas of native North America – a type of sign lingua franca (Taylor 1981, 1996, McKay-Cody 1997, Mithun 1999, Campbell 2000, Davis 2010, 2011, Davis and McKay-Cody 2010). This chapter highlights the documentation and description of American indigenous sign language varieties, narratives, and histories through the use of modern technology and development of sustainable resources for indigenous sign and spoken languages to be studied and revitalized.1 The research project featured here centres on the development of a digital corpus of documentary materials based on the author’s extensive archival research and ethnographic fieldwork focusing on the current sociolinguistic status and linguistic properties of indigenous sign language among Native American communities. 2

Degree of language endangerment

American Indian Sign Language (AISL) varieties are distinct from American Sign Language (ASL) used in deaf communities of the USA and Canada.2 The 1

2

In some instances, the term ‘sign’ (or ‘signed’) is used as a modifier – especially in parallel constructions with ‘spoken’. For example, ‘sign and spoken languages represent different linguistic modalities’. American Indian Sign Language (AISL) is used as an umbrella term because the corpus includes indigenous sign language documentary materials from various American Indian linguistic groups

69

70

Jeffrey E. Davis

transmission of AISL has dramatically declined from its widespread use as a lingua franca in previous times, due in part to its replacement by English, and ASL in some cases. Owing to this replacement, today there is an extreme urgency to document, preserve, and revitalize AISL and other indigenous languages now primarily used by elders and by American Indians who are deaf. The fact that indigenous languages have survived and continue to be learned and used is remarkable, especially considering the pressures for linguistic and cultural assimilation historically imposed on Native groups – e.g. to acquire and use the dominant spoken or sign languages of the larger society or community. In former times, education policies prohibited the use of native spoken and sign languages, which were further discouraged from being used in residentialschool settings. The rapid decline of Native languages in past years has been due to many historical, social, cultural, and educational factors. These matters are tightly intertwined and rendered more complex considering the variety of sign and spoken languages used among American cultural groups. Documentary linguistic work is critical to advance our knowledge of the cognitive, cultural, and linguistic underpinnings of indigenous types of sign language.3 3

Terminology issues

A variety of different terms is used in the literature to refer to the same native or indigenous groups of the Americas, and one can find different labelling conventions according to a specific community or discipline. The author is using ethno-linguistic terminology generally recognized by scholars of American Indian languages and cultures. Geographic designations – such as the ‘Great Plains’ – are based on numerous cultural, linguistic, and historical factors. The preference for the term ‘American Indian’ is reflected in the name of the National Museum of the American Indian, Smithsonian Institution, in Washington, DC. ‘Native American’ (USA) or ‘First Nations’ (Canada) are generally considered politically correct terms; however, members of these groups generally call themselves ‘Indians’ (Karttunen 1994, Davis and McKay-Cody 2010). Furthermore, American Indian/Alaskan Native is the convention recommended by the National Congress of American Indians (www.ncai.org/). The shorter

3

and cultural areas. Although at times broadly categorized this way, it is essential to note that different varieties and types of indigenous sign language have been identified among Native American groups. One of the chief aims of my comparative linguistic research is to assess the degree of historical and genetic relatedness among dialects and varieties of AISL. A consequence of intensive language and cultural contact has been a shift towards English as the dominant or primary language of most individuals from American Indian backgrounds. Furthermore, deaf members of Native communities in the USA and Canada generally attend schools for the deaf and are predominantly learning ASL instead of traditional varieties of Indian Sign Language (Davis and McKay-Cody 2010).

American Indian Sign Language

71

terms American Indian, Indian, or Native are typically preferred by members of these cultural groups. This is comparable to the self-designations commonly found among other cultural groups such as African American (Black) and Hearing Impaired (Deaf). Additionally, in this chapter a distinction is made between ‘Native’ and ‘native’. Upper-case ‘Native’ is the shorter form of Native American Indian, and refers to the genetic heritage and indigenous cultural background of a person or peoples, while the lower-case form ‘native’ refers to acculturation or language acquisition from birth through early childhood. When capitalized, ‘Native’ is used as the shorter form of Native American Indian, or to substitute for longer designations such as American Indian/Alaskan Native. Once the longer convention is properly framed, the shorter form may be used, e.g. ‘Indian’ or ‘Native’. These words are preferable to the terms ‘indigenous’ or ‘aboriginal’. The terms ‘nation’ and ‘culture’ are generally preferred over ‘tribe’, which refers to one village or clan. The justification for sovereign Native American groups to be considered nations has been maintained into the twenty-first century by leaders and members of these communities.4 Naturally, it is considered best practice to ask the individuals themselves for their naming preferences and to respect the traditional practices of native groups. Tribal affiliation and cultural-linguistic group are acknowledged whenever possible. Increasingly, linguists and anthropologists are recognizing that most indigenous languages have more than one name; and likewise, major world languages typically have multiple names. Classifying languages as discrete and isolatable units with clearly defined boundaries is not that black-and-white or cut-and-dried. Linguists must take into account the complex dynamics of the language as it is learned and used across different communities, regions, and nations. Again, this is made somewhat more complex by the multilingual/ multicultural and multimodal nature of language use characteristic of many indigenous communities. Thus, the issue of naming is a recurring theme in the study of Native American cultures and languages, further complicated for a variety of reasons (e.g. ingroup/out-group), but mainly due to each language having several names. This is also true for Indian sign language. For example, members of the groups sometimes refer to their native sign language in these ways: ‘Assiniboine Sign Language’, ‘Blackfeet Sign Language’, ‘Cheyenne Sign Language’, etc. The use of this nomenclature is not meant to positively identify a single sign language variety or dialect – just as we know that ‘English’ actually encompasses many varieties, dialects, and vernaculars. The problem with these labels is that 4

When used as a common noun or adjective without specific reference to a group or language, the terms indigenous, nation, native, and sign language are not capitalized.

72

Jeffrey E. Davis

they imply these sign varieties are discrete and isolatable, rather than dialects, vernaculars, and members of the same sign language family.5

4

Linguistic environment and geographic spread

Even the world’s leading scholars of North American Indian languages (most notably, Goddard 1996, Mithun 1999, and Campbell 2000) do not know exactly how many Native American languages there are all together, nor how many there have been that have now vanished. Estimates of the number of surviving or extinct Native American languages have ranged from as few as 400 prior to the arrival of Europeans to more than 2,500 (reported in Campbell 2000: 3). Moreover, it has been well documented in the research literature that highly conventionalized and linguistically enriched ways of signing emerged as a common means of communication among various American Indian communities. The use of sign language among native groups was so prevalent and widespread in previous times that it served as a lingua franca (Taylor 1981, 1996, Mithun 1999, Campbell 2000, Davis 2010). Linguistic and ethnographic documentation from both historical and contemporary sources indicates that signing was used for a variety of discourse purposes across the major American Indian cultural areas – the South-east, Gulf Coast, South-west, Great Plains, Plateau/Great Basin, North-east, Subarctic, and Mesoamerican geographic areas. The best-documented and described variety is Plains Indian Sign Language (PISL), probably due to the central role it served historically as a widespread lingua franca (West 1960, Taylor 1981, Davis 2010). Taylor (1981) identified and described three major Indian lingua francas: Mobilian (a variety of Choctaw-Chickasaw) of the South-east, Chinook of the North-west, and Plains Sign Language of the Great Plains expanse.6 The historical and sociolinguistic evidence suggests that the Plains Indian sign lingua franca had already emerged prior to European contact and developed through nativization and creolization processes – that is, it was acquired natively and expanded lexically and grammatically. In previous times, the lack of a single dominant language group in the Great Plains cultural area 5

6

The author strives to take a well-informed and balanced approach to describing and classifying native languages. Self-designations of native groups are used whenever possible, although naming preferences vary between individuals and communities. Depending on the reference cited, individual preference indicated, or consensus within the community, the use of certain terms may vary. For this reason, or to avoid redundancy, the author alternates the use or shortens the length of some terms. In addition to the three major Indian lingua francas – Mobilian, Chinook, and Plains Sign Language (Taylor 1981) – several European and Indian spoken languages may have historically served as lingua francas to varying degrees (Mithun 1999).

American Indian Sign Language

73

may have been the reason for the adoption of Plains Sign Language over any particular spoken language (Taylor 1981).7

Plains Indian Sign Language (PISL) Traditionally, PISL served various social and discourse functions both within and between native communities of the Plains cultural areas. The Great Plains cultural area was centrally located on the North American continent and spanned over 1.5 million square miles (4.3 million square kilometres), an area similar to that of the European Union’s 27 member states combined. Although considered endangered, and the extant number of PISL varieties and users is unknown, the language has not vanished. It is still used within some Native groups in traditional storytelling, rituals, legends, prayers, and conversational narratives, and is still being learned by some members of the clan, tribe, or nation, among hearing and deaf alike. Plains cultural and linguistic groups have been identified as the most adept signers; at the same time, it has been documented that other indigenous signing communities bordered or were in contact with the Plains Indian cultural groups (e.g. Plateau/Great Basin and South-western Indian communities). Although the PISL variety served a central role as the historic sign lingua franca of the Great Plains and cultural areas bordering this geographic region, we find other American Indian sign language varieties beyond the Great Plains. The author’s fieldwork and documentary linguistics work have involved other varieties of American Indigenous sign language (e.g. Navajo/Din´e). What is more, distinct varieties of indigenous sign language have been identified among Native groups such as the Inuit-Nunavut (Arctic Territory of Canada); the Keresan Pueblo and Navajo/Din´e (South-western USA); and the Maya of western Guatemala and the Mexican states of Yucat´an, Chiapas, and Oaxaca (Mesoamerican cultural area). See Schuit (2012) and Fox Tree (2009) for further descriptions of indigenous sign language. Traditionally, indigenous sign language served various social and discourse functions within and between numerous American Indian communities of the Great Plains and other cultural groups bordering this area. However, the transmission and spread of the Indian sign and spoken languages have dramatically waned from its widespread use in previous times. In the literature, varieties of American indigenous sign language are sometimes collectively referred to 7

Several prominent scholars hypothesized that sign communication was used among North American Indians prior to European contact; and they distinguished the emergence of a sign language lingua franca from the pidgins, trade languages, and mixed systems used by some native groups (Goddard 1979, 1996, Taylor 1981, 1996, Mithun 1999, Campbell 2000).

74

Jeffrey E. Davis

as American Indian Sign Language (Davis 2011). Although at times broadly categorized or labelled this way, it should be reiterated that distinct varieties of Native American indigenous sign language have been identified in several areas (as listed at the end of the preceding paragraph). Further research is under way to determine the linguistic properties and historical or genetic relatedness among the varieties and types of American Indigenous sign language identified thus far. 5

Towards a typology of village and indigenous sign languages

In contrast to industrialized societies, where sign language is used primarily by members of the larger deaf community, in some indigenous villages, signing is used by both deaf and hearing community members (Kendon 1988, R. Johnson 1994, Nyst 2007, Marsaja 2008, Zeshan 2008, Nonaka 2009). Likewise, indigenous sign language has traditionally been used among North American Indians as an alternative to spoken language even when deaf people were not present. It also has been learned as a first language by some deaf members of these indigenous communities (Davis and Supalla 1995, Davis 2010, 2011, Davis and McKay-Cody 2010). Not surprisingly, indigenous or village sign language has been linguistically enriched and expanded as deaf members of the community acquired it as a primary language. Indeed, it has been well documented that deaf tribal members play a vital role in the development and transmission of indigenous sign language (Davis 2010, 2011). As AISL expanded into a lingua franca for international purposes, it was used by many hearing Indians; however, it continued to serve as the primary language for deaf Indians, their families, and other members of these Native American communities. Corpus of American Indian Sign Language (AISL) One of my chief research objectives is to assess the historical and genetic relatedness among dialects and distinct varieties of AISL. Thus far, my research has focused on the linguistic documentation of the complex of Plains Indian sign language varieties to determine the number of native signers and their degree of sign language proficiency. In the process, I have been applying lexicostatistical analyses to determine historical relatedness (language contact) or genetic relationship (same language family). As stated earlier, AISL is used as an umbrella term for the project’s digital corpus, and research is ongoing to determine the historical or genetic relatedness among the documented varieties and types of American Indigenous sign language. Specifically, the project’s AISL corpus encompasses documentary linguistic material from two main sources: 1) historical documentary linguistic/legacy

American Indian Sign Language

75

materials from archival sources, which I have identified, digitized, and collected in collaboration with the Smithsonian Institution (2001–11); and 2) documentation and description of indigenous sign languages of different American Indian communities based on my postdoctoral fieldwork (1990–94) and my most recent fieldwork (2009–12), concentrating on the linguistic status and geographic spread of the PISL variety. These contemporary and historical documentary materials form the basis of the sign language digital corpus maintained at the University of Tennessee. The AISL corpus is predicated on the documentation and linguistic description of sign language among American Indian groups from contemporary fieldwork and historical sources, including my extensive ethnographic fieldwork and archival research. This establishes that indigenous sign language has been used within at least one dozen distinct American Indian language families. Certainly, signing may have been used by more distinct linguistic groups than these, but the use of sign language has been documented in at least these American Indian groups (see Davis 2010, 2011). The project maintains that this historical case of a sign lingua franca was unparalleled, spanning a large geographic area and once used among members of more than forty linguistic and cultural groups of Native North America (Goddard 1979, Taylor 1996, Mithun 1999, Campbell 2000, Davis 2010). 6

Research aims

The research reported here is grounded in multidisciplinary approaches and the emergent discipline of documentary linguistics, which uses digital technologies for the purpose of documenting, describing, and revitalizing endangered languages. The project aims to illuminate the linguistic nature and structure of indigenous sign languages, and to highlight the work of sign language field linguists and members of Indian signing communities for the purpose of language documentation, description, and revitalization. It also seeks to raise awareness about this important, yet sometimes overlooked, part of American Indian cultural and linguistic heritage. For more than two decades, the author has been conducting ethnographic fieldwork and sociolinguistic research in the quest to illuminate several questions: What are the types and varieties of sign language being used in American Indian communities? Do these documented cases of indigenous sign language constitute one language family or typology? Do these cases represent distinct languages without a common genetic ancestor? Or do we find evidence for unrelated languages, each encompassing a variety of dialects? What evidence do we find for genetic relatedness (cognates) or historical relatedness due to language contact and lexical borrowing? What happens when the sign language of the larger American Indian community is acquired as a first language by deaf and hearing members of these sign

76

Jeffrey E. Davis

communities? To shed further light on such questions, I have been conducting comparative linguistic studies of the AISL corpus collected from both contemporary fieldwork and archival sources. Although additional fieldwork and further research are under way, some of my preliminary findings are reported in this chapter. With support from the National Science Foundation’s Documenting Endangered Languages (NSF-DEL) Program, our project’s team of sign language linguists, anthropologists, interpreters, students, and members of American Indian signing communities have been participating in fieldwork in various parts of the USA and Canada to document and examine the sociolinguistic underpinnings, current status, and spread of AISL. Thus far, we have identified several dialects or varieties of sign language used among American Indian communities.8 7

Current fieldwork

Although American indigenous sign language is considered highly endangered, it has not vanished. Since 2009, we have identified and filmed more than twentyfive native signers from among the American Indian nations, communities, and linguistic groups listed in Table 6.1. Reportedly, there are still hundreds of Native signers using AISL in traditional storytelling, rituals, and conversational narratives (Marvin Weatherwax (Blackfeet scholar), pers. comm. 2010). Thus, we are finding that indigenous sign language continues to be used today by some members of American Indian groups and Canadian First Nations and to serve a role as an intertribal and international lingua franca.9 As in previous times, the PISL variety continues to be the best-documented and -described variety of American Indian sign language (cf. West 1960, Taylor 1996, Davis 2010). As reflected in Table 6.1, we find that sign language is still 8

9

The author would like to acknowledge grant support from the NSF-DEL Program, Division of Behavioral and Cognitive Sciences (NSF-BCS 0853665, NSF-BCS 1027735, NSF-BCS 1110211, and NSF-BCS 1160604), awarded to him as Principal Investigator since 2009 to support fieldwork, data collection, and documentary linguistic research of the project. I, project Principal Investigator and author of this chapter, alternate between ‘I’ and ‘our’, ‘me’ and ‘we’ when referring to aspects of our NSF-funded fieldwork. The work of the DEL project reported here involved collaboration with other scholars, linguistic students, and AISL community stakeholders. I am most grateful to my collaborators for sharing their sign languages, insights, and intuitions. Graduates of the interdisciplinary linguistics programme at the University of Tennessee have provided research assistance in preparing the material featured here and disseminated through the project’s website. I am particularly grateful to Justin Jornd and Cody Klecka, who has been my main research assistant since 2010. I take responsibility for all interpretations presented herein and sincerely hope that much more has been gained than lost in the translation of these rare documentary linguistic materials and findings into the content of this chapter and the research project website.

American Indian Sign Language

77

Table 6.1. Documentary linguistic fieldwork: Native signers identified and filmed (2009–12) Language family

Cultural and geographic area

Algonquian Blackfeet (Aamssk´aa´ pipikani) Blackfoot (Aap´atohsipik´ani = Piik´ani = Piegan)

Northern Great Plains Blackfeet Reservation, northern Montana Blackfoot Nation of Alberta, Canada and Montana Blackfoot Nation of Alberta, Canada* N. Cheyenne reservation of south-eastern Montana

Blood (K´a´ınaa = Kainai) Northern Cheyenne (Tse’tsehestahese) Siouan Crow (Apsaalooke) Mandan-Hidasta (Moennitarri) Assiniboine (A’aniinen)

Nakoda and Lakh¸o´ ta (Tetonwan)

*

Great Plains Crow Reservation, Montana Ft Berthold Reservation, North Dakota Ft Belknap Reservation, Montana and Indian Reserves in Saskatchewan and Alberta Ft Belknap and Ft Peck Reservations in Montana, and Indian Reserves in Saskatchewan and Alberta

The K´a´ınaa (Blood), Aap´atohsipik´ani (Northern Piegan), and Aamssk´aa´ pipikani (Southern Piegan or Montana Blackfeet), and Siksik´a (Alberta, Canada Blackfoot) are the four nations that comprise the Blackfoot Nation (aka Blackfoot Confederacy or Niits´ıtapi).

used within some native groups for a variety of discourse purposes, including traditional storytelling, rituals, legends, prayers, jokes, games, conversations, and personal narratives by both hearing and deaf American Indians. Although classified as an endangered language, known and used primarily by deaf and elderly members of some Native communities, we find that it remains resilient and is still comprehended and used by dozens of American Indians and potentially hundreds by some accounts (West 1960, Taylor 1996, Davis 2010). To summarize, PISL is a highly endangered language, having been kept alive over the past several generations primarily by (1) tribal elders, who use it mainly with others of the same or older generations, or with deaf tribal members, and by (2) deaf tribal members, who have found it a fluent means of communication within their Native communities. 8

Project highlights

Prior to this project, indigenous sign language had generally been considered a primitive, emblematic, or dying language. Earlier documentation of AISL had been carried out by anthropologists in the mid 1900s but involved few women and did not include Native American signers who were deaf. Whereas

78

Jeffrey E. Davis

previous fieldwork had concentrated on the alternative or ad hoc role of signing among hearing members of Native American communities, our current fieldwork focuses on both alternative and primary patterns of acquisition and use among deaf and hearing community members. The project is providing new information on various aspects of the language and its evolution – i.e. sign and spoken language contact, multilingualism and code-switching, pidginization and creolization, the conveyance of human language in sign and spoken modalities, and the interrelatedness and distinctiveness of speech, sign, and gesture. Additionally, this research informs theories of language acquisition, multilingualism, contact, and change. Most significantly, the project brings together American Indian signers from different communities and nations to share their knowledge of Native histories, cultures, and languages. It should be mentioned that no formal survey of AISL signers has been carried out since the mid-1950s. Furthermore, this project is the first of its kind in over fifty years to focus on the linguistic status and spread of indigenous sign language and the first to include women and deaf participants as well as signers from different age groups. The project involves young scholars and key personnel with a long history of working in American Indian communities and backgrounds in sign language, linguistics, anthropology, and language pedagogy or related disciplines. The multilingual/multicultural and multimodal contexts central to this fieldwork contribute to our understanding of numerous linguistic variables and outcomes: historical, traditional, or ceremonial (formal or frozen); vernaculars (believed to be independent by their speakers/signers); standard forms (published lexicons and grammars); dialects (regional or social differences); pidgins (hybrid varieties without native speakers); creoles (pidgins that have acquired native speakers). The students and scholars involved in the project are highly trained and prepared to address the vicissitudes inherent in carrying out documentary linguistic work encompassing multicultural and multilingual domains. In addition, our Native collaborators/sign language consultants are educators, researchers, interpreters/translators, community leaders, and gatekeepers – thus, the project is well guided at all levels. We are respecting the wishes of the local Native American signing communities involved and are developing excellent rapport with Native sign language participants and stakeholders. The work of the project is grounded in principles of ethnographic fieldwork, linguistic description, and language assessment. In the process, we are assessing sign language proficiency among participants as well as collecting self-reported information for metadata documentation. The project is compiling extensive metadata for each participant identified, from novice learners to the most proficient signers. In brief, the current documentary linguistics project involves Native individuals and communities in language documentation (e.g. storytelling narratives,

American Indian Sign Language

79

discourse genres, and oral histories of elders) and linguistic description (lexical and grammatical), and is using modern technologies to develop sustainable resources for AISL varieties to be preserved, maintained, and studied. All products of our research and fieldwork adhere to the highest standards advanced by the National Science Foundation (NSF) and are published by professional organizations – e.g. E-MELD, the Linguistic Society of America, and the American Anthropological Association, among others.10 Preliminary research findings For now, the project can offer the following preliminary linguistic findings: 1) AISL is a full-fledged language that can be analysed at various linguistic levels – e.g. phonemic, morphemic, syntactic, semantic, and pragmatic. For example, we find that AISL phonology is shaped by perceptual, physical, and modality properties common among sign languages. Derivational and inflectional morphological processes are evident, and AISL has bound (affixes) and free morphemes (content words and function words). The data collected and analysed thus far shows AISL to have basic SOV word order (generally considered the most common among the world’s languages) and, correspondingly, head-initial directionality. While seemingly predominant, this is not the only word order type evident in AISL. Syntactic analysis of further grammatical markers is ongoing. The morphosyntactic processes identified thus far in AISL appear highly productive, generating distinct lexical categories (nouns, verbs, adjectives, adverbs, etc.), compounds, polysemous forms, and a variety of predicates composed of indicating, depicting, and pointing signs (common among sign languages). Over 1,000 core AISL lexical signs have been identified. Tense is indicated by signs comparable to adverbs – e.g. today, tomorrow, yesterday, since, etc. When occurring, time indicators are at the beginning or ending of phrases or sentences. Other lexical and grammatical features identified include question formation, negation, pronouns, and possession. There are many examples of the rich use of metaphor, including metonyms, hyponyms, and hypernyms. 2) AISL appears typologically similar to other sign languages, which are characterized by certain spatial-grammatical features, verb inflections, and classifier-like constructions. Our research follows two main comparative approaches: 1) AISL compared with other sign languages; and 2) AISL compared with spoken American Indian languages in the same environment. While further synchronic and diachronic 10

The project is well positioned to train students; we have already worked with students from some of the top programmes around the US. All products of our research and fieldwork will adhere to the highest standards published (e.g. P. Austin and Sallabank 2011).

80

Jeffrey E. Davis

studies are currently under way, we find that AISL is a language distinct from ASL – although deaf Native signers may code-switch between their first sign language and their second sign language (ASL). We are also comparing the linguistic properties of AISL with its ambient indigenous spoken languages, which are typically polysynthetic (Yamamoto and Zepeda 2004). In furthering these analyses, we seek to examine and describe the multiple linguistic levels and domains of AISL – syntax (describing its grammar), lexicon (developing an inventory of lexical signs), and comparative analyses – to determine dialect differences and discourse types. Additional fieldwork, linguistic analyses, and descriptions are under way. 3) Degree of genetic relatedness between varieties of AISL; and, assessments of historical AISL – ASL contact. Following methods established in previous lexical similarity studies (lexicostatistical analyses) of sign languages, we have considered two main types of historical relatedness – genetic and lexical borrowing/language contact. Our comparisons included more than 1,000 American Indian lexical signs documented over a 200-year time span (1800s–2000s) and collected from Native signers from at least 8 language phyla representing more than one dozen different language groups from the South-west, Great Plains, Plateau/Great Basin American Indian cultural areas (see Figure 6.1). These studies resulted in an 80–90 per cent range of lexical similarity between historical and contemporary AISL varieties – suggesting that these are dialects of the same language. In contrast, our AISL and ASL comparisons have consistently resulted in lexical similarity in the 50 per cent range. These preliminary findings suggest that AISL and ASL are not genetically related; however, language contact and lexical borrowing have probably occurred (cf. Davis 2007, 2010). Meanwhile, the quest continues to discover more about indigenous sign languages and to investigate patterns of creolization, lexical borrowing, language change, divergence, and convergence. Methodologies In brief, to advance further AISL documentation and revitalization, and to raise awareness about indigenous sign language, we are applying emergent technologies to create captions, voice-over, slow motion and explanations. For these purposes we are using documentary linguistic methods and new technologies for linguistic transcription and analysis such as ELAN (EUDICO Linguistic Annotator), adhering to E-MELD best practices, working closely with Native American signers, and training sign language linguistic students. The project’s website, maintained at the University of Tennessee, features film clips and photographs of the project’s fieldwork, along with AISL legacy materials the author previously collected in collaboration with the

American Indian Sign Language

81

A R

Greenland

United States

C

B TEA U

A

PLA

R

C

Canada

T

C

I

A

United States

I

GREAT BASIN

N

NOR

S SO

T SOU

U

TH

W

E TH

AS

T

T

NO

I

U

L

CA L I F O R N IA

T

S

P

RTH

WEST COAST

C

HE

A

S

ES

T Mexico

Figure 6.1 Map of cultural areas of Native North America. Reproduced from Ubelakar 2006: ix, by permission of the Smithsonian Institution, National Anthropological Archives

Smithsonian’s National Anthropological Archives. This website includes a research blog to encourage the participation of others, thus offering engagement and dissemination to the broadest possible audiences.11 To ensure greater access, searchability, and preservation, we aim to submit these documentary materials to the world’s most reputable language archives, e.g. the Endangered Languages Archival Repository (ELAR); the Rosetta Project (Stanford University); and the Smithsonian. The AISL corpus of this 11

Research website: http://pislresearch.com/ (accessed 24 January 2013).

82

Jeffrey E. Davis

project is based on the documentation and description of sign language among American Indian groups from historical and contemporary sources, including the author’s extensive ethnographic fieldwork and archival research. 8

Summary and conclusions

This chapter highlights the significance of field linguists and anthropologists with strong sign language backgrounds collaborating with Native American communities in the documentation and linguistic description of indigenous sign language varieties, narratives, and histories. The author’s project is attempting to accomplish this vis-`a-vis modern technology and by developing sustainable resources for Native sign and spoken languages to be preserved, maintained, and studied. For these purposes, the project has been conducting the first fieldwork in over fifty years to focus on the linguistic status and underpinnings of AISL, classified as a highly endangered language variety. Although the extant number of AISL signers is unknown at this time, we believe that as many as 100 or more North American Indians may still know and use it to varying degrees of proficiency (Davis 2011). Since 2009, the AISL documentation project has identified and filmed more than twenty-five native signers, including women and deaf tribal members from among Algonquian and Siouan language groups (see Table 6.1). Continued documentation of Indigenous sign language and comparative linguistic work to describe various dialects in separate American Indian communities, tribes, and families is under way. The documentary linguistic work of this project contributes to indigenous language revitalization and enhances our knowledge of the cognitive, cultural, and linguistic underpinnings of indigenous and village sign-language typologies. Although AISL is an endangered language variety, we anticipate that additional AISL signers will be identified, and such efforts are buoyed by a resurgence of interest in the revitalization of Native languages over the past several years.

7

Purism in language documentation and description Michael Riessler and Elena Karvovskaya

1

Introduction

Purism has been described as a barrier to endangered language revitalization, but paradoxically purists are often also key revitalizers. Language choices and inner-group language conflicts are sociopsychological phenomena and thus outside the core of descriptive linguistic research. The evolution of linguistic structure is nevertheless clearly related to the social function of language. The present chapter discusses purism at the interface between language sociology and sociolinguistics, specifically from the perspective of documentary linguistics. Based on data about the use of exclusive focus particles in the seriously endangered Kildin Saami language, we show how puristic attitudes can affect the actual language performance of recorded speakers and potentially give rise to language variation and change. We also discuss the question of how this kind of synchronic variation can be accounted for in the documentation and description of an endangered language. 2

Kildin Saami

2.1

General situation

Kildin Saami belongs to the Peninsula group of the East-Saami languages (Sammallahti 1998: 26–34). Originally, the language was spoken all over the central inland parts and the central coastal regions of the Kola peninsula. Four dialects still exist: the K¯ıllt dialect, the Koarrdegk dialect, the Lujavv’r dialect, and the Arsjogk dialect. Today, more or less compact Kildin Saami settlements in or close to their original villages are found only in Lovozero (Kildin Saami

Tove Bull, Rogier Blokland, Kristina Kotcheva and an anonymous reviewer provided most valuable comments and suggestions on earlier drafts of this chapter. Special thanks are also due to Bruce Mor´en-Duollj´a, Ciprian Gerstenberger, Elisabeth Scheller, and Trond Trosterud for discussing issues on purism and ‘language police’. Needless to say all remaining errors are our own.

83

84

Michael Riessler and Elena Karvovskaya

Barents Sea SKOLT

(MURMANSK) Kola

KILDIN

Lujavv’r (OLENEGORSK)

TER

Krasnosˇcˇel’e

Kola Peninsula

Kanevka

AKKALA Sosnovka

(APATITY)

N

(UMBA)

White Sea Figure 7.1 The traditional Saami settlement area on the Kola peninsula and adjacent parts to Fennoscandia

Lujavv’r) and a few other places on the Kola peninsula (see Figure 7.1). As a result of the forced resettlement of considerable parts of the Kola Saami population to Lovozero, this town is nowadays usually regarded as the ‘Saami capital’ of Russia. Kildin Saami is critically endangered as a result of heavy language shift to Russian during the last century. According to Scheller (2011: 88, 89, 91), today the language only has about 100 active and 600 passive speakers. For more detailed information on the current situation of Kildin Saami see Scheller (2011, 2013) and further references mentioned therein. The recent linguistic history of Kildin Saami in contact with Russian and other languages is dealt with in Riessler (2007, 2009) and Blokland and Riessler (2011). 2.2

Documentation and description

Kildin Saami has been written since the end of the nineteenth century. The Finnish linguist Arvid Genetz was the first to document and describe Kildin and other Kola Saami languages using contemporary linguistic methods. In 1891, Genetz published a descriptive dictionary of Kola Saami based on his

Purism in language documentation and description

85

Bible translation from 1878 and a few other collected texts. Even the first short comparative grammatical description of all Kola Saami languages by Hal´asz (1883) is based on Genetz’s text collection. The first Kildin Saami grammar was ˙ written by Endjukovskij (1937) with the aim of supporting the training of Saami teachers. Consequently, the author applies a moderately prescriptive approach and uses the Latin-based orthographic standard of that time, but not a linguistic transcription. Itkonen’s (1958) comparative Kola Saami dialect dictionary is still the most comprehensive source for the vocabulary of Kildin Saami, and a comprehensive descriptive grammar of Kildin Saami was published by Kert (1971). Although several theoretical descriptions and text collections have since been published, lexical and grammatical descriptions are incomplete, and there is not much data available to reflect current language use. For the time being, no systematic language-planning activities exist for Kildin Saami, but two dictionaries (Afanas’eva et al. 1985, Kert 1986), as well as a number of teaching materials for children, have been produced since the 1980s. The Kola Saami Documentation Project (KSDP), a Dokumentation Bedrohter Sprachen (DoBeS) or ‘Documentation of Endangered Languages’ project in which the current authors are participating, is currently creating a multimedia documentation of Kildin and the other Kola Saami languages of Russia,1 partly in collaboration with the Saami Language Technology Project (Giellatekno).2 3

Purism

3.1

Purism in the sociology of language

We understand purism as an excessive insistence on traditional rules or structures resulting from a static view on one’s own culture. Rather than accepting the fact that cultural variation and change are inevitable in society, purists express evaluative attitudes towards an alleged original state of culture as more pure than, and hence superior to, the actual existing heterogeneity, which they experience as inferior or corrupted owing to innovation. Linguistic purism, which seems to be typical for several large and small language communities (see, for example, Thomas 1991 and Brunstad 2003 for examples from larger communities), results in negative attitudes towards innovative variants in one’s own language. In a situation of language attrition and loss, with bilingual speakers shifting to the majority language, purism essentially results in prescriptivism 1

2

The project was funded by the Volkswagen Foundation (grant II/80999–1). The archive can be accessed online at http://corpus1.mpi.nl/ds/imdi browser?openpath=MPI363060%23. Although the regular funding period for the project was extended in 2010, work on building the archive continues by former project collaborators. See http://giellatekno.uit.no/

86

Michael Riessler and Elena Karvovskaya

entailing the removal of, and protection against, innovative elements, which are very often the result of recent contact-induced changes. Linguistic prescriptivism resulting from purist attitudes is sometimes described as a crucial barrier which has to be overcome in order to revitalize and maintain an endangered language (see, for example, Dorian (1994), Crystal (2000), or, specifically on Kildin Saami, Scheller 2011: 99–100). It has been shown in many cases how purist attitudes among the group of traditional ‘established’ speakers affect language use in the group of ‘revitalized’ speakers, who are often younger and L2 learners of the language in question. From this perspective, purism specifically implies the discouraging effect which permanent criticism (or even correction) has on passive or not fully proficient native speakers in diminishing their desire to continue learning and using the endangered language actively. As a result, purism can lead to an intergenerational struggle about whether the young generation’s innovative language variant is to be preferred or not to ‘death in beauty’ (Trosterud 2003: 213 (our translation)). Besides the purely psychological effects of raising language learners’ barriers to actively use and revitalize the endangered language, purism can also be characterized as a social phenomenon connected to inner-group identity struggles. Purists claim a special social role as ‘language owners’ by (consciously or not) monitoring the native language and keeping it pure through policing language use inside the community. Therefore, purists have sometimes been characterized as ‘language police’ by researchers working in Saami societies (see, for example, Scheller 2009, Braut 2010: 9, Johansen 2010). There are other important perspectives to consider with regard to purism. First of all, in the context of endangered language communities’ struggle towards revitalization, purism often goes hand in hand with a process of native gathering for revival of the ‘own-language tradition’. The purist thus functions as an agitator for preferring the native language to the majority language (Trosterud 2003: 210). Therefore, it is not paradoxical at all that purists typically are also language activists and key persons in the revitalization of endangered languages. In the specific case of Kildin Saami, the purists are also dedicated teachers, authors of teaching materials, language collectors, and linguists. Furthermore, they are among the most valuable resource persons and consultants for the outside linguist who is documenting and describing this critically endangered language. Whereas purist struggles inside the community have in fact shown negative psychological and social effects on the ongoing revitalization attempts for Kildin Saami (Scheller 2011, 2013), in the present chapter we wish to deal with purism and prescriptivism from the perspective of descriptive and documentary linguistics while considering it as a variational linguistic phenomenon.

Purism in language documentation and description

3.2

87

Purism in documentary and descriptive linguistics

Dorian believes that purism can (and should) be channelled into forms which are useful rather than harmful (1994: 480–1). We shall describe such potentially useful forms of purism and prescriptivism from the perspective of documentary and descriptive linguistics at the interface between theoretical research aiming at an adequate linguistic representation of the recorded data and applied research on documentation useful for the language community under investigation. 3.2.1 Documentary linguistics Documentary linguistics is a newly evolved sub-field of linguistics first of all concerned with language documentation as a comprehensive, multifaceted and multipurpose record of the linguistic practices characteristic of the investigated speech community (Himmelmann 2006: 1). Among the most important purposes of documentary linguistics is the creation of data pools for further research on and for endangered languages, for both theoretical (for example, in linguistic anthropology and linguistic typology) and applied research (for example, on language revitalization and language technology). Furthermore, documentary linguistics aims at creating and refining methodologies and technical standards for building, presenting, and evaluating endangered language corpora. In consequence, documentary linguistics is primarily an applied discipline, and language documentation is but one (important) method used in documentary linguistics. However, documentary linguistics as a discipline on its own might be less well established then is sometimes assumed (see, for instance, Peter Austin’s contribution to the present volume), because ‘documentary linguistics’ and ‘language documentation’ are also often used as mutually interchangeable names for a sub-field of theoretical linguistics aiming at better description and analyses of the world’s linguistic diversity. In fact, the theoretical linguistic bias in mainstream endangered language documentation projects is clearly reflected in the focus on deep morphosyntactic corpus-annotations and thorough linguistic descriptions based on these corpora. Rather than interpreting the field as a mere technical framework for language documentation or simply the (newly improved) empirical side of theoretical linguistic research on underdocumented languages, documentary linguistics primarily includes aspects of applied research. Besides theories, methodologies, and techniques overlapping with documentation science and language technology, these applied aspects also include practical work in the field of revitalization, undertaken in collaboration with the investigated speech community. Recent calls for project grant applications3 and numerous recent conferences or workshops organized by different institutions dealing with 3

See, among others, DoBeS (http://www.mpi.nl/dobes/) or HRELP (www.hrelp.org/).

88

Michael Riessler and Elena Karvovskaya

documentary linguistics explicitly include technical, methodological, and theoretical issues connected to the revitalization of endangered languages. If this collaborative approach to documentary linguistics is more than just windowdressing, language documentation must first focus on obtaining results that are useful for the investigated community. It is precisely in this area that documentary linguistics interfaces with applied research on endangered language planning and revitalization.

3.2.2

The completeness and linguistic adequateness of documentation and description Documentary linguistic research on underdescribed languages includes analytic work with annotations, ideally accompanied by grammatical sketches which can, in turn, serve as metadata for understanding the created corpus (see, for example, Mosel 2006b). This is where documentary linguistics interfaces with descriptive linguistics. Any descriptive grammar should be written as accurately and thoroughly as possible to make the work useful for the broadest possible (linguistic and non-linguistic) audience (cf. Noonan 2006). Consequently, a truly comprehensive and descriptively adequate grammar of an endangered language also acknowledges the attested synchronic variation: “[T]he goal of a descriptive grammar should always be fidelity to community usage in all of its messiness and inconsistency (Noonan 2008, cit. Genetti 2011: 9).” However, thorough grammatical description is a time-consuming task which cannot be fulfilled completely by the documentary linguist. Writing a descriptive grammar is a project which can only be based on, but only rarely be included in, language documentation. The documentary linguist nevertheless needs to be sensitive to variation in the data in order to make the documentation as complete and useful as possible. Such variation might pertain to dialects or other, more sociolinguistic, variants such as age, gender, and social class. When working with an endangered language, the degree of attrition is also important. On the one hand, certain morphological forms or syntactic constructions might no longer be used by the recorded speakers, or, on the other hand, these speakers might produce forms or constructions that they have reconstructed themselves or even invented in order to fill apparent gaps. In the present paper we consider specifically linguistic variation resulting from such a compensatory strategy. We connect this phenomenon to purism if it results from the conscious re-creation of an alleged ‘traditional’ and ‘pure’ language with the aim of eliminating the apparent ‘defects’ caused by attrition in a language shift situation. There are of course several other possible motivations for filling in gaps in a linguistic system under attrition. Typically, the variation is not conscious, but

Purism in language documentation and description

89

results from speakers making mistakes, misremembering, or having interference from another language. However, since the purist variants we have documented have sometimes been systematic, we believe that they can potentially give rise to language change. Consequently, attested purist variation should not only be accounted for by sociolinguistic description but also be included in comprehensive language documentation. Purist speakers often (although not always) remember archaic variants of the language. Such data is obviously relevant for historical linguistics and for the description of language change, as documented archaisms can help interpret different kinds of recorded synchronic variation. Interestingly, in our elicitation sessions on different parts of grammar currently undergoing language change (for example, focus particles, possessive inflection, or reflexive pronouns), we have commonly encountered purist variants which are not archaic but, rather, innovative and which are spontaneously invented rather than remembered – and these are not always nonce-forms. Rather than excluding alleged non-authentic (i.e. invented in non-spontaneous speech) forms from our work on language documentation, we have decided to include this kind of variation since it is in line with the general philosophy that underpins documentary linguistics, namely preparing a complete record of as many attested linguistic practices as possible. A large corpus of nonelicited texts, ideally comprising both annotated spontaneous speech and – as in our case – written texts, will help evaluate and interpret different kinds of purist variation and thus contribute to the later preparation of an adequate (socio)linguistic description of the language in question.

3.2.3 Documentary linguistics, language planning, and revitalization Language planning means the conscious interference in the development of a language in order to further advance its use in new domains. Documentary linguistics basically means building corpora for endangered languages. Since useful language documentation must additionally include a grammatical sketch, a wordlist, or other metadata that explain the textual annotations in the corpus of recordings, the most basic prerequisites for endangered language planning (i.e. corpora, dictionaries, and grammatical descriptions) are already produced by the documentary linguist. In our view, the need for thorough linguistic analyses is still overemphasized in those language documentation projects aiming for thorough and multi-tier morphosyntactic analyses in corpus annotation (frequently carried out by researchers manually). While such annotations can be quite precise, they require a lot of time and leave much room for inconsistencies and mistakes. We consider that, for the most part, a simple orthographic transcription combined with a translation into the main lingua franca and a sketch grammar seem sufficient in order to render a language documentation

90

Michael Riessler and Elena Karvovskaya

useful – in other words, readable and accessible for further theoretical analyses or practical applications. This is especially true for languages such as Kildin Saami for which an established orthography and basic phonological and morphological descriptions are already available. Accordingly, rather than providing our corpus with thorough phonological and morphosyntactic analyses and English glosses, we have chosen to simply transcribe our recordings in the established Cyrillic writing system and to provide Russian translations. We also consider that, for purposes of documentation, a learner’s grammar in the lingua franca including phonological, morphological, and syntactic rules and paradigms can sometimes be more useful than a comprehensive descriptive grammar in English, since it is easier to produce and it can also be directly used for teaching. Note that grammatical descriptions and other teaching materials for endangered languages are, by definition, prescriptive and require that variation be reduced (at least to a certain extent) in order to make these materials useful to learners (see Hinton and Hale 2001: 15). This is where our approach differs from that of theoretical linguistics, where such grammatical sketches might seem less adequate. We believe that, when aiming simultaneously at documenting and planning an endangered language, a moderately prescriptive approach to writing a sketch grammar is useful. 4

Case study – Variation in focus particles

The following case study presents an example of ‘synchronic change’, i.e. language variation and, presumably, ongoing change due to language attrition. It also illustrates a specific instance of where the purist attitudes of our consultants have affected their actual language performance during elicitation sessions. The term ‘focus particles’ refers to function words such as only, even, and also (see K¨onig 1991). Those words have been attracting a lot of attention in theoretical linguistics due to their complex meanings and syntactic behaviour. Because of their interaction with information structure, focus particles provide an instrument to explore interfaces between syntax-phonology, syntaxinformation structure, and syntax-semantics. It is assumed that all languages have at least one exclusive focus particle with the meaning ‘only’ (K¨onig 1991: 98). However, in her study of Kildin (and Ter) Saami exclusive focus particles, Karvovskaya (2010) noticed a) variation across different speakers (e.g. dialect and degree of language loss), b) variation across different text types and genres (e.g. spontaneous speech, elicitations, prose, and teaching materials), c) inconsistencies in the use of focus particles by one and the same speaker and within one and the same text type, and d) inconsistent ad hoc replacement by other words. c) and d), in particular, challenge the assumption that Kildin Saami possesses a lexical item equivalent to English ‘only’. Karvovskaya (2010) concludes that the different Kildin Saami varieties do indeed possess at least one exclusive focus particle and that the inconsistencies observed in the elicited

Purism in language documentation and description

91

data are caused by purist attitudes which have resulted in speakers repairing structures they deem to be ‘incomplete’ or ‘defective’. The following sections illustrate these findings with reference to a few selected examples from a case study on variation and change in focus particles. Whereas Subsections 4.1 to 4.2 summarize our analyses of the documented data from the perspective of descriptive and variational linguistics, in the short Subsection 4.3, we discuss a moderately prescriptive explanation which could be used in sketch grammar. For detailed information on focus particles in Kildin Saami and more examples, see Karvovskaya (2010, 2012).

4.1

Linguistic documentation and description of attested focus particles

It would appear that Kildin Saami exhibits two equivalents to the English focus particle ‘only’. Both lexemes are borrowings from Russian: lyˇse < Russian liˇs’ and tol’ke ( tol’ko  tolke  toke) < Russian tol’ko (Itkonen 1958: 196, 604). Since no native Saami lexical item with this meaning is attested in any of the four closely related Kola Saami languages, it is unlikely that the Russian words have only recently replaced inherited Saami lexemes. Conversely, lyˇse and tol’ke must belong to a layer of early Russian loanwords.4 The words are borrowed presumably from northern dialects of Russian, because their borrowing dates back at least to the nineteenth century: tol’ke is recorded by Europaeus (Itkonen 1931), and lyˇse features in Genetz (1891). It is therefore most likely that the loanwords were already fully adopted by the time that language shift gained momentum in the wake of the Russification (later Sovietization) that took place during the twentieth century. One additional fact which points to long-established integration in the language system are the cognates t˚a´lkˇ and la´ˇsa in the closely related Skolt Saami language, which is now predominantly spoken in Finland by speakers who have no knowledge of Russian whatsoever. Interestingly, we find several other instances of exclusive (and other) focus particles being borrowed. Furthermore, the equivalents for ‘only’ are loanwords in practically all Saami languages (see Table 7.15 ). Although tol’ke and lyˇse may well have been borrowed at the same time, there is no evidence that both forms have ever been used in free variation (and competition). Instead, data obtained from speakers in different geographic areas suggests that the occurrence of lyˇse versus tol’ke was originally due to dialectal variation. There are speakers of different dialects who use one or other of these forms exclusively Karvovskaya (2010). 4 5

Riessler (2009) presents a detailed study of loanword layers in Kildin Saami. The North and Inari Saami lexemes are inherited but go back to a Baltic loanword (with a different meaning) in common proto-Saami, reconstructed as *tuˇssˇe by Sammallahti (1998: 240). The North Germanic source words for the particle in Southern, Lule, and Northern Saami are bara (Swedish) / bare, bærre (Norwegian).

92

Michael Riessler and Elena Karvovskaya

Table 7.1 Cognate sets of borrowed exclusive focus particles in different Saami languages South

Lule

North

Inari

d˚asˇsˇa˚ bære

duˇssˇe beare

tuˇssˇe

barre

Skolt

t˚a´lkˇ la´ˇsa

Kildin

tolke lyˇse

Ter

tol’ke lyˇse

Source < < <