Field Linguistics

Field Linguistics
This page intentionally left blank

Field Linguistics
A Beginner’s Guide
Terry Crowley
Edited and prepared for publication by
Nick Thieberger
1
3
Great Clarendon Street, Oxford ox2 6dp
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide in
Oxford New York
Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto
With oYces in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam
Oxford is a registered trade mark of Oxford University Press
in the UK and in certain other countries
Published in the United States
by Oxford University Press Inc., New York
ß The Estate of the Late Professor Terry Crowley 2007
The moral rights of the authors have been asserted
Database right Oxford University Press (maker)
First published 2007
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means,
without the prior permission in writing of Oxford University Press,
or as expressly permitted by law, or under terms agreed with the appropriate
reprographics rights organization. Enquiries concerning reproduction
outside the scope of the above should be sent to the Rights Department,
Oxford University Press, at the address above
You must not circulate this book in any other binding or cover
and you must impose the same condition on any acquirer
British Library Cataloguing in Publication Data
Data available
Library of Congress Cataloguing in Publication Data
Data available
Typeset by SPI Publisher Services, Pondicherry, India
Printed in Great Britain
on acid-free paper by
Biddles Ltd, www.biddles.co.uk
ISBN 978-019-928434-4 (hbk.)

ISBN 978-019-921370-2 (pbk.)
1 3 5 7 9 10 8 6 4 2
Contents
Preface ix
Acknowledgements xiii
Editor’s Foreword xiv
Publisher’s Note xiv
1 Field Linguistics: Why Bother? 1

1.1 The threat to diversity 1
1.1.1 Some lessons from Truganini 1
1.1.2 The value of records 5
1.1.3 Diversity still undocumented 8
1.2 The ‘Big Questions’ 11
1.2.1 Armchairs and dirty feet 11
1.2.2 Fieldwork at home and fieldwork in The Field 13
1.3 What kind of linguistics? 16
1.4 Anthropological and linguistic fieldwork 18
2 Ethical Issues 23
2.1 Linguists and speakers 23
2.1.1 Ethical guidelines 23
2.1.2 Avoiding harm 25
2.1.3 Informed consent 27
2.1.4 Voluntary participation 28
2.1.5 Money and fieldwork 30
2.1.6 Ethical delinquency 32
2.2 Linguists and communities 33
2.2.1 Grammars 34
2.2.2 Dictionaries 38
2.2.3 Reading materials 44
2.2.4 Technical advice 48
2.2.5 Public awareness 48
2.2.6 Terminology development 50
2.2.7 Recognizing our limitations 50
2.3 Linguists and other linguists 52
vi contents
3 Getting Started 57
3.1 The fieldworker 57
3.2 Choosing a language 58
3.3 Background work 62
3.3.1 Reading the literature 62
3.3.2 The linguist and the lingua franca 64
3.4 Planning your fieldtrip 65
3.4.1 Fieldwork duration 66
3.4.2 Planning for contingencies 68
3.5 Funding 69
3.6 Permits 70
3.7 Equipment and supplies 73
3.7.1 Recording gear 73
3.7.2 Other needs 76
3.8 Getting about in the field 78
3.9 First contact 80
3.9.1 Arriving with an invitation 80
3.9.2 Arriving without an invitation 81
3.9.3 Local reactions to the fieldworker 82
4 Gathering your Data 85

4.1 Choosing language-helpers 85
4.1.1 ‘Informants’ 85
4.1.2 Language-helpers at home 86
4.1.3 Language-helpers in the field 88
4.2 Phonological and grammatical elicitation 92
4.2.1 Language of elicitation 93
4.2.2 Working environment 94
4.2.3 Initial elicitation 95
4.2.4 Eliciting grammatical patterns 97
4.2.5 Taking notes 101
4.2.6 Native-speaker judgements 102
4.3 Gathering words 104
4.3.1 Direct elicitation 105
4.3.2 Beyond elicitation 106
4.3.3 Spontaneous speech 108
4.3.4 Sound substitutions 110
contents vii
4.3.5 When is the dictionary done? 111

4.4 Keeping track of data 111
4.4.1 Daily records 112
4.4.2 Filing systems 113
4.4.3 Computer storage 116
4.5 Archiving 117
5 Beyond Elicitation 121

5.1 The textual corpus 121
5.1.1 Early corpora 121
5.1.2 The recording of speech 122
5.1.3 What to record 124
5.1.4 Reasons for recording 127
5.1.5 How large a corpus? 129
5.1.6 Different genres 130
5.2 Getting it right 131
5.2.1 Evolving analyses 132
5.2.2 Native-speaker error 132
5.2.3 Fieldworker error 134
5.2.4 The problem of tunnel vision 136
5.3 Text transcription 137
5.4 Participant observation 141
5.4.1 Extracting and observing language 141
5.4.2 Overhearing in the field 147
5.4.3 Beyond linguistics 149
6 Problems and Pitfalls 151

6.1 Analysis in the field 151
6.2 Learning to speak the language 154
6.2.1 Reasons for trying 155
6.2.2 Reasons for success and failure 156
6.2.3 Individual differences 159
6.3 Going troppo 161
6.4 Getting caught up 165
6.4.1 Becoming part of the scene 165
6.4.2 Unexpected entanglements 166
6.5 Linguists behaving badly 172
viii contents
7 Salvage Fieldwork 177

7.1 Getting started 177
7.1.1 Responding to the enthusiastic 177
7.1.2 Persuading the reluctant 179
7.2 Working with a limited number of speakers 181
7.2.1 Diminishing speaker populations 181
7.2.2 Diminishing structures 182
7.3 Legitimate linguistic authority 185
7.4 Saving the language? 187
References 191
Index 197
Preface
A huge number of the world’s languages remain poorly described, or even

completely undescribed. Many may have disappeared altogether by the
end of the twenty-Wrst century, and only a small number of people are
doing anything about this. Even among linguists—who we might expect
to be among the most concerned—there are surprisingly many who are
doing surprisingly little.
This book deals with one aspect of the issue of linguistic diversity in the
world: the task of linguistic documentation arising out of original re-
search in the Weld. Although many linguistics programmes include a Weld
methods component, I have often been asked by academics, ‘When you
do Weldwork, how do you actually start?’ This suggests that researchers
are possibly sometimes still being thrown into the Weld at the deep end,
much as I suppose I was back in the early 1970s when I was sent on my
Wrst Weldtrip. I went with an abundance of enthusiasm, tempered
with some trepidation, but not a whole lot of practical training. There
therefore has to be a substantial need for a book such as this.
I have chosen to concentrate in this volume on only a subset of Weld-
work activities. Clearly, somebody who is involved in recording a corpus
of some variety of non-standard spoken English is just as much engaged
in Weldwork as somebody who journeys forth to the Kamchatka Penin-
sula to document a previously undescribed language. While some of the
issues are the same for all kinds of linguistic Weld research, the linguist
who is heading for Kamchatka is likely to be faced with quite diVerent
sorts of issues from the linguist who is headed for the Bronx. The latter
kind of Weld linguist is extremely well served with information about how
to assemble a corpus (e.g. Milroy 1987), while the Kamchatka-bound
linguist is much less well served.
There are, admittedly, other guides to linguistic Weldwork, but
none seems to me to be completely satisfactory. Some were written in a
diVerent technological era when reel-to-reel tapes and typewriters were
our main tools. Some are far too dry, concentrating on how to take
notes and how to make satisfactory voice recordings, while ignoring the
interrelationships between the linguist and language speakers as
people. Some concentrate exclusively on problems arising from personal
x preface
relationships in the Weld while sidestepping crucial issues of data-gather-

ing. Some completely ignore the fact that Weldwork involves an array of
ethical issues which need to be addressed, and treat Weldwork instead
entirely as an exercise in advanced linguistic analysis. Some seem better
designed for the experienced Weldworker who wants to look back on
Weldwork, rather than being geared towards the neophyte who has yet
to test the waters.
This volume is geared towards people who already have some
background in linguistics and who intend to apply their training to the
recording, analysis, and description of previously undescribed, or poorly
described, languages. I anticipate that my major audience will be students
enrolled in undergraduate courses in Field Methods (or any number of
other courses with vaguely similar-sounding names), as well as graduate
students—or even linguists occupying academic positions in universities—
who are contemplating the possibility of their Wrst foray into The Field.
However, I hope that this volume is appreciated also by anthropo-
logical Weldworkers, and those from kindred disciplines such as psych-
ology, education, and sociology. There has long been a tradition of
discussing, debating, and dissecting the process of Weldwork in the
discipline of anthropology, and there are plenty of published sources
dealing with a wide range of issues. However, this volume will oVer a
very diVerent perspective.
I hold to the belief that real ethnographic insights cannot come only via
questionnaires mediated by a hired translator. Even if a scholar is able to
conduct interviews in the Weld in person through the medium of a lingua
franca, he or she is still likely to be missing something that can only come
through the local language. An ethnographer needs to be able to partici-
pate as well as to observe, and that means at the very least being able
to understand something of what is being said in the immediate surround-
ings. While there is no pill that I can oVer to instantaneously give you
a command of a new language, I hope that this volume will at least make
you understand that learning something of the language of your chosen
Weld site is not only possible, but desirable, even essential.
I plan to avoid dryness by basing this account of linguistic Weldwork
substantially on my own continuous career of Weldwork dating back to
the early 1970s, as well as on what I know of the Weld experiences of some
of my colleagues in linguistics. We all learn by our mistakes, and I have
plenty of my own to share with you. I have also been observing the
Location of languages recorded by Terry Crowley
xii preface
failures of others for as long as I have been committing errors of my own,

and some of these also provide excellent lessons for us all. (I promise to be
discreet by not mentioning names.)
For those who do not know me, I should probably establish my
credentials at the outset. I have produced—or am currently working
on—published accounts ranging from very brief grammatical sketches
to detailed grammatical accounts and accompanying dictionaries and
collections of texts for eighteen diVerent languages: Yaygir and Bandja-
lang (northern New South Wales); Uradhi, Mpakwithi, and Cape York
Creole (far north Queensland); Tasmanian (actually a number of lan-
guages); Bislama (Vanuatu); Sye and Ura (Erromango, in Vanuatu);
Paamese (Paama, in Vanuatu); Nāti, Naman, Neve’ei, Tape, Nese, and
Avava (Malakula, in Vanuatu); Mwotlap (Banks Islands, in Vanuatu);
and Gela (in Solomon Islands). In addition to English, I speak Bislama
(and closely related Tok Pisin), Paamese, and French (but in reality, only
with somebody who can’t speak much English). In the past I have also
been able to get by suYciently well in Bahasa Indonesian, German, and
Dutch to convince my examiners to give me surprisingly good passing
grades. I have also taught linguistics at university level in both Papua
New Guinea and Vanuatu for a total of twelve years, prior to taking up
my current teaching position in New Zealand. I therefore consider myself,
linguistically speaking, to have been around.
Terry Crowley
Hamilton (New Zealand)
April 2004
Acknowledgements
While I would like to claim the original title of this volume as my own
creation, I must be honest and express my intellectual debt in this respect
to Don Laycock. Don was a Weld linguist of considerable standing,
having worked on a number of non-Austronesian languages in Papua
New Guinea. He had long hankered after an excuse to write a book with
the title The Cunning Linguist which was this book’s working title. It was
Don who originally encouraged me to pursue my interest in linguistics
when I was only 15 years old and it is hard to Wnd a way to thank him for
this. Regrettably, Don died in 1988 and was never able to write his book.
I dedicate this volume to his memory.
While I have based this volume substantially on my own experiences in
the Weld, as well as my observations of other people in the Weld, I have also
beneWted from anecdotes, information, and advice from a number of
people. I would like to oVer my gratitude to the following for their
contributions: Lyle Campbell, Michael Goldsmith, John Lynch, Bob
Tonkinson, and Nick Thieberger.
Terry Crowley
Editor’s Foreword
Terry Crowley’s death in early 2005 left behind a number of works in

progress, including the present book. I received the manuscript when it
had already been read by two anonymous reviewers, and Terry had
responded but not yet included their comments in the draft. He had
sent me the manuscript for my input, especially on new technologies
and Weldwork, and I have added that here. I have also added some
references to further reading, taken account of the reviewers’ comments,
and updated some sections. Thanks to Eva Fenwick for work on the
index, and to Margaret Florey and Marianne Mithun for their comments
on an earlier draft.
Nicholas Thieberger
Publisher’s Note
The publishers wish to express their gratitude to Dr Nicholas Thieberger

for his work on Wnalizing the manuscript of this book which Professor
Crowley had completed before his death on 15 January 2005.
1
Field Linguistics: Why Bother?
1.1 The threat to diversity
No linguist can fail to be aware that during the twentieth century

a substantial number of small languages ceased to be spoken. Some of
these were described before their last words were spoken; others were not.
A substantial number of other speech traditions are now similarly
destined for linguistic oblivion. We can hope social, economic, and
political practices will be able to adapt such that the world’s surviving
linguistic diversity can survive. But if it cannot, then hopefully those
languages that do disappear in the future will at least be recorded for
posterity.
1.1.1 Some lessons from Truganini
Many of my readers have possibly never have heard of Truganini. Some

of you may have heard of her, but you may know little of her life. But
I want to begin this volume by asking, ‘What could Truganini have taught
us?’
So, who was Truganini then? In one sense, she was nobody terribly
important at all. She was a woman who died in her sixties in 1876 in what
was then something close to the end of the world, in the small British
colonial settlement of Hobart, Tasmania. She was not formally educated
at a time when there were many uneducated women. She never held any
public oYce; she did not become a famous leader; and she never accu-
mulated even a modicum of wealth. Although she had married, her
husband died many years before her, and they produced no heirs. She
wrote no books, and she is not credited with having expressed any great
wisdom. Yet Truganini’s life, and her death, was a great tragedy for us all.
2 field linguistics: why bother?
For Truganini was the last person in the world to die who had grown
up speaking one of the several Aboriginal languages of Tasmania. She
was born shortly after the Wrst European settlement at what is now
Hobart in 1803. Her birthplace was some distance away from this
settlement in what is still today pretty much a rural backwater. During
the childhood that she spent with her people, while there were plenty of
opportunities for contacts with European sealers and sawyers, she would
have spoken with her parents and other close relatives in the language
that today we call South-Eastern Tasmanian. Unfortunately, we have no
way of knowing what Truganini’s people called their own language, as
nobody at the time ever seems to have thought to ask anybody this
question.
Truganini was an observer of, and active participant in, events which
can only be described as one of the great tragedies of colonial history in
the nineteenth century (Ryan 1996). Truganini’s own local people, as well
as Tasmanian Aborigines all over the island, were disinherited of their
lands, rivers, and seas and the bounteous resources that they provided
them, as European settlers—some might prefer to see me use the term
‘invaders’—rapidly fanned out from the new settlement of Hobart to
establish farms, homesteads, and rural townships.
The Aboriginal population dropped very quickly, for a variety of
reasons. European settlers brought with them a number of diseases to
which they had for generations—if not centuries—enjoyed a degree of
immunity. While among Europeans these diseases were not necessarily
fatal, they produced high mortality rates among Aborigines when they
were exposed to them for the Wrst time. Aboriginal people’s immunity was
perhaps compromised anyway by lack of adequate nutrition brought
about by restricted access to traditional hunting grounds when
their land had been ‘settled’ by Europeans for pastoral purposes. To
compound problems for the Aboriginal population, birth rates dropped
as Aboriginal women were infected with the venereal diseases that were
introduced by Europeans.
These factors alone would possibly have been enough ultimately to
bring about a demographic catastrophe for Truganini’s people. However,
their problems were made worse by the brutal policy of the colonial
government, which deliberately set about capturing every living
Tasmanian Aborigine and relocating them to a previously unoccupied
oVshore island to which none of those people ‘belonged’. This policy of
field linguistics: why bother? 3
ethnic separation—perhaps we should call this early apartheid or ethnic

cleansing—and the establishment of the island ‘homeland’ was triggered
by a period of active resistance by Aboriginal people along the expanding
frontier of European settlement in which uncertain numbers of
Europeans and Aborigines were killed in a series of violent skirmishes.
Once relocated, the surviving Aboriginal remnant population—which
included Truganini—continued to decline in numbers. Their new home
had an unhealthy climate, and the health care that was made available
there was wholly inadequate. Eventually, this remote island settlement had
to be abandoned and a new settlement was established for those who had
survived at a healthier site closer to Hobart. However, the population
had by this time passed the stage where it could be self-sustaining.
The remaining members of this small community were past child-bearing
age, and they died, one by one. The last to go was Truganini.
Before 1803, everybody in Tasmania spoke a Tasmanian Aboriginal
language. After 1876, nobody did. Most of this tragedy played out during
Truganini’s lifetime.
There are many more tragedies that I could have reported along the
way in the life of Truganini (Rae-Ellis 1981, 1988). One of these tragedies
is the emergence of the myth that Truganini was the ‘last Tasmanian’, and
with her passing, her race became ‘extinct’. While she had no children of
her own, many other Aboriginal women at the time did in fact bear
children, albeit to European fathers working in the sealing trade.
Descendants of those children number several thousands in Tasmania
today, and they proudly identify with their Aboriginal forebears. They
now call themselves the Palawa people, after the documented word in one
of the dozen or so original Palawa languages for ‘(Aboriginal) person’.
Those children who originally grew up with their Aboriginal mothers and
their sealer European fathers grew up speaking English. Some did learn
a few words and set expressions in their mothers’ languages, and some of
these were passed down the generations to the present (Crowley 1993).
However, those children did not learn how to construct new sentences
in any of their mothers’ languages, and they were able to carry on
conversations only in English.
So, the Palawa population of Tasmania today is an English-speaking
community, as it has been for generations. However, Palawa people have
a proud association with their traditional past as well as a very powerful
association with their tragic colonial history. They feel an overpowering
sense of sadness over the loss of their ancestral languages. This sense of
loss is all the harder to bear because so little of those languages was ever
recorded.
Although today’s Palawa people can trace their ancestry directly back
to the Aboriginal women of the nineteenth century, many have so much
European blood that physical appearance is no longer a reliable guide to
ethnic identiWcation. In such situations, having a language of your own
that is clearly diVerent from that of the English-speaking mainstream in
today’s overwhelmingly Anglo-Celtic Tasmania would be a powerful
symbol of ethnic distinctness. The Welsh, for example, cannot be physic-
ally distinguished from their English neighbours, but anybody who has
been to Wales and seen the street signs in Welsh and watched TV broad-
casts in Welsh—and heard Welsh spoken in the local store—will be well
aware of how important it is for Welsh people that the language
should continue to be spoken.
But this is not what we Wnd today among the Palawa people. While
something of the original Tasmanian languages was recorded in the
nineteenth century at a time when some of their last speakers were still
alive, the records were extremely fragmentary. For one thing, Truganini’s
language was only one of possibly as many as a dozen separate languages,
and some of the surviving records have mixed words up from diVerent
people as if there was just a single language.
In any case, words were written without the beneWt of any kind of
phonetic training. Observers simply recorded words using the spelling
system of English. This is poorly enough designed for representing the
sounds of English, and it is completely inadequate clearly to indicate the
pronunciations of words in completely unknown languages. Finally, what
was recorded of the Tasmanian languages was for the most part just
words, with no recognition of the fact that in order properly to document
a language, it is necessary also to show how these words are used in a wide
range of diVerent kinds of sentences. This means that we have almost no
idea about how the grammars of these various languages were organized,
and we will never be in a position to know this.
This, then, is the linguistic tragedy for today’s Palawa people. In the
face of repeated claims that they no longer exist as a distinct people,
attempts are now being made to use Palawa words that were recorded in
the nineteenth century to create a new language. They call this
Palawa Kani, using the documented word kani meaning ‘talk’. Of course,
while a historically legitimate vocabulary can be reconstructed to some

extent, with so little of the original grammar ever having been recorded,
there are real limitations on what can be achieved with this kind of
linguistic reconstitution, which nevertheless clearly has importance for
Palawa people today.
1.1.2 The value of records
It didn’t have to be this way, of course. If only a selection of diVerent

kinds of sentences had been accurately recorded by those nineteenth-
century observers, we would be able to say so much more about the
original grammatical patterns of these languages. There are places in
the world where the original languages have become extinct, but enough
information was left behind in written form that people today have been
able to work out in a fair amount of detail what the original language was
like. Today’s Cornish language enthusiasts, for example, are able to make
use of extensive written records that were kept before the last native
speaker of the language died in 1777. If Truganini and her friends and
relatives in that dwindling community had been given the education and
the motivation, they could have written to each other in their languages,
much as speakers of Cornish did in the 1700s. Alternatively, if a person
with real linguistic insight had been able to sit down with Truganini to
record her language in writing for posterity before she died, today’s
Palawa people could have been so much richer, for so little eVort.
But the tragedy of Truganini’s passing has become a tragedy not just
for today’s Palawa people. It is also a tragedy for humanity at large. This
means that it is a tragedy for you, and for me, because with the loss of any
language without documentation there is a whole range of questions that
can never be answered.
For one thing, the records of the Tasmanian languages are so poor that
it is impossible to say what their closest linguistic relatives might be. We
might expect that they would be related to the Aboriginal languages of
the Australian mainland. So might they have been, but the speakers of
those languages had been cut oV from the mainland by rising sea levels
that came with the end of the last ice age about ten thousand years ago,
the languages therefore had so long to evolve independently that any
original similarities could easily have been obscured beyond recognition,
even if we had excellent records. After all, the island of New Guinea has
been cut oV from mainland Australia for about the same length of time
and the languages there show no convincing evidence of any kind of
relationship with any of the Australian languages, or sometimes, even
with each other.
However, having poor records has not stopped people from trying to
make all manner of claims about historical connections between the
original Tasmanians and other peoples. Some such claims are demon-
strably wrong. For example, the claim that the Moriori of the Chatham
Islands in New Zealand were a Tasmanoid people who were driven to
extinction with the later arrival of the Maori people (King 2003: 56–7) can
easily be shown to be wrong. Nineteenth-century documentation of the
now-extinct Moriori language indicates very clearly that it was a close
relative of Maori.
Others have claimed that the Tasmanian languages are related to the
little-known languages of the Andaman Islanders as part of Greenberg’s
(1971) Indo-PaciWc Hypothesis. No convincing proof of any kind of
relationship between Tasmanian languages and any other language
grouping has been, or can be, presented. However, the fact that such
claims have been made at all seems to be encouraged—rather than
discouraged—by the poor records of the Tasmanian languages.
But linguists ask questions about more than just linguistic history. The
prime motivation in modern linguistics is to answer the question: What is
possible in human language and what is not? If someone were to set out to
design a language for Klingons—a rather violent non-human species
whose name will be immediately recognizable to Star Trek aWcionados
—what features would they ascribe to it? It would obviously need to have
plenty of vocabulary to express Wghting and honour, such as qaD
‘challenge’, bIj ‘punishment’, vaj ‘warrior’, and vID ‘be belligerent’
(Okrand 1992: 181–5). But its grammar would probably also contain
features that are never found at all in human languages, or which, if
they are found, are very uncommon. So, while human languages very
commonly have Subject Verb Object (SVO), Subject Object Verb (SOV),
and Verb Subject Object (VSO) word orders, it is much less common for
languages to have word orders in which objects come before subjects.
The rarest word order in the world is Object Subject Verb (OSV), which
is found in just a handful of languages. So, OVS would be a good
candidate for Klingon syntax, if we wanted to make Klingon look as
exotic as possible. Thus:
puq legh yaS

child sees oYcer
‘The oYcer sees the child.’ (Okrand 1992: 59)
So, what sorts of features can be present in human languages and what
cannot? If I were able to talk to Truganini—and assuming that she might
have been willing to share this kind of information with me—what kinds
of questions might I have asked her? Of course, I would have wanted to
know what the basic constituent order in her sentences was. Statistically,
it is likely to have been SVO or SOV, but it could logically have been any
of four other possibilities (VSO, VOS, OVS, OSV). There is one Wnal
possibility: it could also have been a language with free constituent order,
as long as there were suYcient mechanisms elsewhere in the language
for distinguishing between performers and undergoers of actions. If
Truganini’s language were of this ‘non-conWgurational’ type, we would
therefore expect it to have had case markers of some kind: suYxes
or preWxes, or postpositions or prepositions.
But I would want to Wnd out so much more than just this. Among many
other things, I would want to know if her language had a system of noun
classes or not. If it did, I would want to know how many classes it had.
I would want to know the details of the semantic content of each of these
noun classes. I would want to know if the language marked a distinction
between singular and plural nouns and, if it did, whether it made the same
noun class distinctions with plural nouns as it did with singular nouns.
There are lots of languages where the number of noun classes is the
same in the singular and plural. For instance, the deWnite article in
Spanish in the singular is el with masculine nouns and la with feminine
nouns, while with masculine plural nouns it is los and with feminine plural
nouns it is las. There are also lots of languages in which more noun classes
are marked with singular nouns than in the plural. For instance, there are
separate deWnite articles in French for masculine singular nouns (le) and
feminine singular nouns (la), but there is only a single article in the plural
for nouns of both genders (les).
Predictions have been made that there are no languages in which the
reverse is true, where there are more noun class distinctions marked with
plural nouns than with singular nouns (Greenberg 1966: 95). So, if
Truganini’s language turned out to have noun classes, one of the things
I would want to check is how its singular nouns behaved in relation to its
plural nouns. If Truganini’s language turned out to be the one and only
language on earth which went against the prediction, she would have been
able to teach us a valuable lesson, and Greenberg’s earlier generalization
would have to be revised.
Regrettably, the main lesson that Truganini can teach linguists
now is the folly of allowing a language to disappear without properly
documenting it Wrst.
1.1.3 Diversity still undocumented
Tragic as the circumstances are surrounding the loss of Truganini’s

language, the situation that I have described is far from unique. Hers was
one of perhaps a dozen distinct languages that were originally spoken in
Tasmania, and we are left with only very fragmentary records of all of them.
On the Australian mainland, none of the languages of Victoria has been
spoken for many generations. Although some of these languages were
rather better recorded thanks to the work of one linguist (Hercus 1969),
none has been described in any great detail, and some have disappeared
with only very fragmentary records. The last Xuent speakers of a handful of
languages in New South Wales were recorded in some detail by linguists
who conducted Weldwork in the 1970s (Crowley 1978, Eades 1979,
Donaldson 1980, Williams 1980), but most of the remaining languages
disappeared with rather less complete records, or even records that can
only be described as extremely fragmentary. It was estimated in 2001 that of
the 250 or so languages that were originally spoken in Australia, only about
17 could be considered strong (McConvell and Thieberger 2001: 61).
Descriptions similar to what I have just recounted for parts of
Australia can be repeated for many other parts of the world that have
been subject to European colonial expansion in the last few centuries.
While indigenous languages survive in some parts of the Americas—most
notably in Central America and Amazonia—there are other parts of
North and South America where there has been widespread language
loss. In the United States, many of the original languages have disap-
peared (Nettle and Romaine 2000: 5), and this pattern is mirrored in
much of Canada (Nettle and Romaine 2000: 8). Of course, the indigenous
languages of the Caribbean have long been completely extinct, given that
this was the Wrst part of the New World to be settled after the arrival of
Christopher Columbus (Crystal 2000: 24).
Tragic as these kinds of stories are, a volume such as this cannot devote
too much time to crying over linguistic spilt milk. What has happened has
happened and we cannot turn back the clock. What we need to do now is
try to stop any more milk from being spilt. And there is quite a lot of milk
in containers that are now teetering.
Of the remaining ninety languages in Australia which still have
speakers, only twenty are currently being passed on to today’s generation
of children (McConvell and Thieberger 2001: 17), and there is no
guarantee that they will all be in a position to pass on these languages
in due course to their own children. There are substantial numbers of
languages in North America that are probably now in their Wnal gener-
ations of speakers. There are pressures in many other parts of the world
from large languages on small languages which also have uncertain
futures. One recent estimate has suggested that as few as 10 per cent of
the world’s 6,000 or so languages that are spoken today will continue to
be spoken by the end of this century. While this is one of the more
alarming among a wide range of predictions, if this turns out to be
correct, then Truganini’s tragedy stands to be played out at the rate of
four or Wve languages a month, every year, for the whole century. Even if
we experience a slower rate of extinction, then Truganini’s tragedy may
play out at only two or three languages a month (Krauss 1992: 6–7).
Let’s think about this. We are not dealing with just idle statistics here,
as real people and real communities are also involved. In the 1970s,
I visited parts of northern New South Wales and far north Queensland
to record people who were regarded at the time as the last speakers of the
Bandjalang, Yaygir, Mpakwithi, and Uradhi languages. Since then, the
people whose speech became the basis for published accounts of those
languages (Crowley 1978, 1979, 1981, 1983) have all died. Since then,
these written accounts have been used by subsequent generations in
a variety of ways, including language learning lessons, and as support
for applications for land rights for which it is necessary in Australian law
to demonstrate an ancestral connection with the land in question.
While it is certainly more urgent for dying languages to be documented
than it is to document those which continue to be passed on to succeeding
generations of children, linguistic Weldwork does not have beneWts just for
speakers of dying languages. The bulk of the world’s 6,000 or so languages
are in fact still being passed on to children, though the pressures on many of
these are such that we cannot expect that all will survive in the longer term.
The task of fully documenting even a single language is enormous. In

fact, no language—not even English—has ever been truly fully described.
(If it had, presumably we would be able to engage in person-to-computer
chat sessions about Chomsky’s politics, the Sapir–Whorf Hypothesis,
and the likelihood that it will rain this afternoon.) Even if we were to
lower our expectations somewhat, there is still a huge task of linguistic
documentation awaiting us. For the sake of argument, let us assume that
a language can be said to be reasonably ‘well described’ if there is a
published grammar which runs to 200–300 pages of published text,
since this approximates to what is expected of a doctoral dissertation.
(Mind you, so limited would such an account be that a computer which
had been programmed with this grammar would still probably make
for a very strange conversationalist.) Of course, for a language to be
considered to be truly ‘well described’, there should also be a published
dictionary of roughly similar length which would be a reasonable
reXection of the everyday vocabulary.
Anything less than this should qualify a language as being less than
‘well described’. So, a language for which there is only a published
grammar, or only a published dictionary, cannot yet be considered to
be adequately described. A language with a grammatical description and
dictionary that each run to between 100 and 200 pages could perhaps be
considered to be only moderately well described, while a description of
less than this could be considered as representing only a sketchy account.
While the world’s major languages no doubt meet these fairly basic
criteria for being well described, the descriptive coverage of the world’s
many small languages—which are, of course, also those which are most
vulnerable to language shift—is much more patchy. I would like to be
able to report how many of the total number of the world’s languages
remain undescribed or only sketchily described, but I do not have these
kinds of Wgures. Nor do I have any Wgures for how many of the world’s
languages can be regarded as moderately described.
However, let us look at one country with which I am quite familiar.
Vanuatu is the world’s most diverse nation in terms of the number of
languages per head of population. Its 200,000 people speak at least 100
quite distinct languages. About twenty of these are in fact moribund and
are destined to disappear with the passing of the current older generation,
which leaves about eighty languages that are being actively passed on
to younger generations. Of these, only two qualify as well described
according to the criteria that I have just presented. The vast bulk of
these languages—well over eighty in all—are either almost completely
undescribed or only sketchily known (Lynch and Crowley 2001: 19).
I do not know how safe it would be to extrapolate from this situation
to other parts of the world, but I do know that there are huge numbers
of languages in the world that are only poorly described. Many of
the world’s linguistic hotspots—the rest of Melanesia, including
Papua New Guinea, Indonesia, India, the Caucasus, equatorial Africa,
Amazonia—are covered to extents that are probably not hugely diVerent
from what we currently Wnd in Vanuatu. That is, for the many parts of
the world, there is likely to be a scattering of well-described languages
and a somewhat larger number of moderately-described languages, while
a very large number of languages remain either completely unknown or
only very sketchily known.
However, there are some rays of hope. In Australia in the 1960s, the
situation would have been very similar to what I have just described
for Vanuatu, but since the 1970s there have been huge strides in the
documentation of those languages which are still spoken. Some of those
languages that have only recently been documented have in fact in the
meantime lost their Wnal speakers. In these cases, Truganini’s tragedy has
not been repeated, at least not in its entirety.
1.2 The ‘Big Questions’
Linguistics could not be linguistics without linguistic descriptions. If we

are going to respond to the Big Questions of our science—how diVerent
can languages be? and how similar must diVerent languages be?—our
answers can only be as accurate as the extent of the raw data that we have
looked at. Linguistics therefore depends crucially on the availability of
a broad collection of detailed and accurate descriptions of the world’s
languages.
1.2.1 Armchairs and dirty feet
It could be said that there are two kinds of linguists in the world. Let me
caricature each type with the labels ‘armchair linguists’ and ‘dirty-feet
linguists’.
Armchair linguists often publish quite extensively without necessarily

putting in any eVort to conduct original research on any language, except
perhaps their own native language, which surprisingly often seems to be
English. Some linguists argue that it is not really necessary to consider
data from a wide range of languages when talking about ‘language
universals’, as it is far more useful to analyse data from a restricted
number of languages obtained from great ‘derivational depth’. This
corresponds to the tradition of the ‘deep–narrow’ approach to the study
of linguistic diversity (Comrie 1981: 1–5). It is not diYcult to conjure up
an image of a linguist in this tradition relaxing in his or her comfy chair,
conjuring up grammatical sentences, along with a set of contrasting
ungrammatical strings of words which can be marked with asterisks, for
inclusion in his or her next paper in Linguistic Inquiry.
The other kind of linguist is someone who gets actively involved in
the study of previously undescribed languages (or, sometimes, in the
re-description of previously described languages that they feel to be in
need of description from a new perspective). Dirty-feet linguists are,
of course, not frequently referred to as such, though speaking as an
enthusiastic career-long Weld linguist myself, I can conWrm that it usually
takes several days of soaking and scrubbing to remove the ingrained dirt
from the soles of my feet after a few months spent shoeless in the Weld.
Field linguists typically fall into the ‘broad-surface’ approach to the study
of linguistic universals in that we stress the need for reliable
descriptive observations from as broad a range of languages as possible
in order to make deductions about what is and what is not possible in
human language.
Dirty-feet linguists provide the fodder upon which many armchair
linguists depend. While there is a long tradition of linguistic Weldwork
in North America that dates back to the period of Sapir and BloomWeld
in the 1920s and 1930s, American universities since the Chomskyan
revolution of the 1960s have for the most part been dominated by
armchair formalists. European universities are similarly dominated by
the armchair tradition. In many of these universities, Weldwork-based
dissertations in linguistics are sometimes discouraged as being ‘too
descriptive’, or as being insuYciently informed by theory.
In Australia, however, the Weldwork tradition is alive and well, reXecting
the enduring inXuence of R. M. W. Dixon who was inXuential in
establishing the modern grammar-writing tradition in that country after
his Wrst academic appointment there in 1970. In Australian universities, the

tradition has developed of a linguist going out to the Weld and ‘doing’
a language for his or her doctoral dissertation. That was certainly how
I gained my doctorate: I ‘did’ Bandjalang as part of my undergraduate
training (Crowley 1978), and then I ‘did’ the Paamese language of
Vanuatu for my Ph.D. (Crowley 1982). Fortunately, there are plenty
of universities elsewhere in the world reXecting the same grammar-writing
tradition, despite the seemingly irresistible power of the armchair
formalists in some places.
Armchair linguists and dirty-feet linguists tend to look down on
followers of the other tradition. Dirty-feet linguists typically characterize
armchair formalists as being out of touch with reality, as dealing with
introspective and intuitive data that has not been subject to empirical
veriWcation, and as delving often into theoretical trivia. Armchair
linguists tend to look down on their dirty-feet counterparts because
they feel that they fail to carry out analysis at suYcient levels of deriv-
ational depth, and because they often do not include in their descriptions
any discussions of constructions which reXect any level of theoretical
sophistication.
In reality, of course, this kind of enmity is unnecessary and unhelpful
(Foley 1993, Gil 2001: 125–8). No descriptive Weldworker can carry out
linguistic analysis of a previously undescribed language in a theoretical
vacuum. We therefore need the insights that theoretical linguistics can oVer
in order to guide our search for new data, and to inXuence our analysis of
that data. Similarly, no theoretician can oVer valid generalizations without
access to a wide range of reliable descriptive data. So, new linguistic
descriptions will always need to be written, and theoretical discussions
will forever need to be re-evaluated in the light of these descriptions.
1.2.2 Fieldwork at home and Weldwork in The Field
There is a kind of halfway house in linguistic Weldwork between sitting in

the armchair and getting dirty feet. Those of you who have done
a university Field Methods course have already been there. I have seen
some academic curricula vitae in which linguists have stated that they
have ‘done Weldwork’ on such-and-such a language, though in reality they
may have travelled no further than the outer suburbs of San Francisco or
Manchester.
While this is clearly a kind of Weldwork, it is not really the kind of

Weldwork that I will be primarily dealing with in this volume. At most,
this kind of Weldwork is useful if you are only interested in studying
a particular feature of a language without intending to produce
a coherent overall account. If, on the other hand, you are planning on
producing a comprehensive account of a language, this kind of Weldwork
might represent a legitimate initial foray to prepare yourself for the real
task of going out into the community where the language is spoken.
As part of any preparation for a Weld trip, you should certainly try to
Wnd a speaker of the language in your own area before you leave.
Particularly in a university community, there are likely to be people
from many parts of the world, and there may turn out to be a member
of the language community that you have chosen living nearby. Such
a person can also help you to become familiar with the general situation
in the language area. And of course, such a person may even be able to
help you to get started in the Weld by introducing you to people within his
or her home community. Being introduced by an insider can, after all, be
a huge beneWt.
But should you consider doing all of your Weldwork without leaving
home? Any work that you do with a speaker of a language other than
English with a view to publishing the resulting linguistic analysis can
legitimately be referred to as Weldwork. Fieldwork ‘at home’ like this
has many advantages for the Weld linguist. I have found that regular
short sessions with a speaker of a language where you can work at
home with an ergonomic chair and desk, electric lights, and a computer
nearby on which you can regularly update your analysis as new material
comes in can be an extremely productive way to work on a language.
However, there are real drawbacks associated with this approach. If
you were to write a linguistic description on the basis of information from
just a single speaker and you do not have the advantage of being able to
observe natural interactions between speakers, you run the risk of missing
some elements of the language. One problem is that the context of
elicitation may not be suYciently varied pragmatically to produce a full
range of constructions or vocabulary. There is also a real danger that
your single speaker may end up eVectively ‘Wltering’ out certain kinds of
constructions because he or she is subconsciously evaluating how good
a command of the language you have acquired, and judging what kinds of
structures you are ready to deal with. This means that it is essential in
a good linguistic description for data to come not just from direct elicit-
ation or from what is recorded in narrative texts, but it must come also
from observations of casual utterances between people speaking spon-
taneously around you.
Let me show you what I mean with an actual example. In a description of
one language that was described on the basis of information from a single
speaker ‘at home’, instrumental noun phrases (INST) are described as being
marked by means of a particular preposition according to the pattern:
I washed my clothes inst soap.
With third person singular inanimate noun phrases, the pronoun is

usually expressed as zero. Thus, in order to say ‘I washed my clothes
with it’, you would say the equivalent of something like:
I washed my clothes inst.
So far, so good, in that there is nothing that is inaccurate in what the Wnal
grammar ended up saying.
After the grammar was completed, the dissertation was examined, and
it was deposited for posterity in the library, conversational data gathered
within the community where the language was regularly spoken revealed
that when the object of the preposition is fronted to the beginning of the
clause—as happens often in spontaneous speech for a variety of
reasons—then the instrumental preposition is very often shifted so that
it appears between the verb and its object, resulting in patterns such as the
following:
This is the soap that I washed inst my clothes.
This construction would probably never appear in sentences translated

directly from English, as the more English-like construction below still
represents a grammatically acceptable possibility:
This is the soap that I wash my clothes inst.
In this case, the narrative texts which formed the basis of much of the
grammar did not provide enough variation in the pragmatic contexts for
these kinds of constructions to appear. The grammar, therefore, ended up
being incomplete.
So, Weldwork should ideally be conducted within the community where
the language is used in order to ensure access to a suYciently rich and
varied set of data. A more pure-minded Weldworker might also want to

argue in any case that Weldwork at home in the comfort of your living
room involves insuYcient levels of self-deprivation. There is a part of me
which says that for your grammar to be truly worthy, you must have
suVered at least one bout of malaria—or some other impressive-sounding
tropical ailment—in its writing, or you should have had at least one
toenail ripped oV by your hiking boots, or you should have developed
a nasty boil on an unmentionable part of your body. I, of course, have
suVered all of these misfortunes in the Weld, and many more. And, by
golly, if I had to go through this, then I feel that everybody else should
have to suVer to a similar extent!
Further reading: Hyman (2001).
1.3 What kind of linguistics?
Some descriptive linguists deWne basic linguistic documentation as

reXecting the very essence of the discipline of linguistics, which they
regard as taking precedence over all other areas of linguistic activity:
The ideal apprenticeship [for a linguist] is to undertake Weldwork on some

previously undescribed (or scarcely described) language—recording, transcribing
and analysing texts; observing how people use the language in the daily round;
writing a grammar and phonology; compiling a dictionary; and publishing
a volume of annotated texts. (Dixon 2001)
There are probably as many diVerent ways of describing the phonology,

morphology, and syntax of a language as there are linguists, though
diVerent approaches can often be gathered roughly under a variety of
diVerent theoretical labels. Linguistic descriptions that are rigidly bound
to particular theoretical constructs do not tend successfully to outlast the
theories which spawned them. Grammars that are written, for example,
according to the strict structuralist formulae that were in vogue up to the
early 1960s, or according to the transformational model which became all
the rage immediately after this, are often extremely diYcult for people to
read today. Similarly, descriptions that are expressed exclusively in terms
that derive from some of today’s theoretical traditions are often equally
unreadable to those who hail from other traditions.
The most readable—and arguably, therefore, the most valuable—

accounts of languages seem to be those which are relatively open to
theoretical eclecticism. By this, I mean that a linguistic description should
set out to allow the linguistic data to govern the form of the description,
rather than requiring a single theoretical model rigidly to dictate the shape of
the entire grammar. Some would argue that there is a single theory-neutral
model for grammatical description which can be used for any language, as
implied by Dixon’s (1997: 128–38) use of the expression Basic Linguistic
Theory. While it would perhaps be nice if this were the case, any linguistic
description is in reality going to exhibit certain kinds of theoretical biases,
many of which may be implicit. The most important consideration in
producing a good linguistic description is that the writer’s particular theor-
etical assumptions should be clearly recognizable to the reader, that any
terminological conventions should be clearly explained, and that the gram-
mar should be richly exempliWed with natural data so that readers coming
from other theoretical persuasions are able to make sense of your discussion.
With new technological tools it is possible to create rich contextual data in
a media corpus for which your analysis provides a gateway (e.g. Thieberger
2004) and which allows others to conWrm and extend your analysis.
Linguistics is a hugely diverse discipline, and Weld linguists can
contribute to our knowledge in a wide variety of ways. While most
linguists are primarily interested in matters of phonology, morphology,
and syntax, the ways in which language is used ‘in the daily round’ noted
by Dixon seldom receive more than passing mention. Only a minority
of linguists—better, perhaps, linguistic anthropologists—have shown
themselves to be interested in documenting how language functions as
a ‘mode of social interaction [which] provides the material out of which
a group of people recognize themselves as a community’ (Duranti
1997: 99). Such accounts cannot be based on the traditional direct
‘elicitation’ of language data from ‘informants’; rather, the Weldworker
must become a long-term participant observer, recording extensive
collections of both audio and video data of natural language use between
diVerent individuals within the community.
At the same time, it is clear that Duranti’s (1994) discussion of the
ways in which grammatical constructions in Samoan are used to achieve
socio-political goals depends crucially on a prior analysis of the language
in a traditional descriptive account such as that of Mosel and
Hovdhaugen (1992). Even the title of Duranti’s linguistic anthropological
account—From Grammar to Politics—implies this priority. Similarly,

SchieVelin and Ochs (1986) show how language acquisition by children
is related to socialization into diVerent cultures using the same kinds of
basic linguistic information. Traditional descriptive accounts, then,
should be promoted as an essential stepping stone for bigger—and
some would no doubt argue, better—things.
The goals of Weldwork can sometimes be much more limited than a full
published grammar or dictionary, or a detailed ethnographic account of the
communicative strategies that are used by people in a particular society.
Sometimes, previously published materials may spark the interest of
specialists who return to somebody else’s previous Weld site in order to
analyse some particular features of the language in more detail. For
instance, in some of the languages of Vanuatu, there are phonemically
contrastive linguo-labial consonants in which nasals, stops, and fricatives
are produced with the tip of the tongue touching the upper lips. Ian
Maddieson undertook a Weldtrip from the United States to Vanuatu in the
1980s simply to study these speech sounds in detail in a number of languages
in which they are found. A guide that is speciWcally geared towards instru-
mental phonetics is beyond the scope of this volume, though Ladefoged
(2003) and Maddieson (2001) provide a great deal of very readable material.
Published grammars concentrate for the most part on regular patterns,
resulting in a lack of attention devoted to the study of patterns of variable
data. There is absolutely no reason why the kinds of corpus-based statis-
tical studies that have been carried out extensively on diVerent varieties of
English could not be carried out in other languages as well, e.g. Dorian’s
(2001) work on Gaelic. However, in order to do this, a linguist would need
to pay close attention to a much wider range of sampling and data-
gathering issues than is commonly done in order to ensure that statistically
representative samples of diVerent categories of speakers have been
recorded. While I have also decided to exclude sociolinguistic studies
from the scope of this volume, readers who are interested in an
up-to-date guide to data-gathering of this kind could consult Milroy (1987).
1.4 Anthropological and linguistic fieldwork
The threat to the world’s linguistic diversity to which I have already

referred is not a threat that we have only recently become aware of.
In North America in the late nineteenth century, people realized that

things were not well. People saw that traditional knowledge, belief
systems, and languages were about to be irretrievably lost with the
passing of the last generations of Native Americans who were born before
the push by European Americans to occupy the West.
Part of the response to this realization was the establishment of the
Boasian tradition of anthropological Weldwork which involved for the
Wrst time academics from North American universities going out to stay
in diVerent communities in order to Wnd out how people lived, what they
believed, and how they spoke. Franz Boas (1911: 60) himself regarded
a command of the language as ‘an indispensable means of obtaining
accurate and thorough [ethnographic] knowledge’. Edward Sapir was
drawn into this Weldwork tradition, and while he was in part
a traditional anthropologist, he was also a gifted linguist. He contributed
a number of substantial original descriptions of previously undescribed
languages, while at the same time playing an important role in developing
a methodology for describing new languages.
While anthropology and linguistics began very much as kindred
disciplines at the time of Boas and Sapir, they have diverged quite
substantially in the meantime to the point where these days, the
stereotypical ethnographer and the stereotypical linguist aim to produce
quite diVerent outputs and they often go about their Weldwork in quite
diVerent sorts of ways. Anthropologists often aim for a much more
modest speaking ability in the language, in spite of Malinowski’s (1929:
465) warning that:
There has been no greater source of error in Anthropology than the use of
misunderstood and misinterpreted fragments of native vocabulary by observers
not thoroughly conversant with native tongues and ignorant of the sociological
nature of language.
Margaret Mead wrote of the importance of a knowledge of the local

language in carrying out anthropological Weldwork, yet she apparently
had a far from sophisticated command of Samoan despite the fact
that much of her anthropological reputation derived from her work in
American Samoa. Her work has in recent years been subject to intense
debate, with her command of the language being subject to scrutiny
(Freeman 1983: 286, Holmes 1987: 112). Michael Goldsmsith (pers.
comm.) reports that at a public lecture that Margaret Mead once gave
in Fiji, a Samoan woman stood up and pointed out that many Samoans
did not agree with some of the things that she had written about them.
Rather than engaging in debate about the serious issues being raised,
Mead’s response was to oVer the formulaic Fa’afetai, fa’afetai lava,
fa’afetai tele lava ‘Thank you very very much’, which evidently impressed
the non-Samoans in the audience, but it hardly represented a staggering
display of linguistic virtuosity.
Despite these diVerences between linguistics and anthropology, there
is a very real sense in which no Weld linguist can fail to be part-
anthropologist, and no Weld anthropologist can fail to be part-linguist.
Any linguist who aims to learn to acquire a speaking knowledge of the
language that he or she is studying clearly has to understand at least some
of the cultural norms of the community in order to understand what
people mean by what they are saying. Any anthropologist who cannot, to
some extent at least, overhear what people are saying in their ordinary
conversations with each other is entirely dependent on his or her own
direct questions for an understanding of local cultural practices.
Arguably, the Weld linguist has something of an advantage over the
Weld ethnographer in terms of our Wnal outputs, as we linguists know
pretty well what sorts of things we should set out to discover as soon as we
get to our Weld sites. We know, for example, that whatever language we
are going to study, it must have phonemes. We won’t know how many
phonemes there will be, or what shape they might have, but we can be
absolutely certain that there will be phonemes. We also know that those
phonemes will go together to make up morphemes, which will in turn
make up words, and these will in turn be put together to make up phrases
and sentences. We can be certain about all of this because every language
in the world, by deWnition, has a hierarchically organized grammar.
Obviously, the kinds of meanings that are going to be distinguished by
the morphemes of any given language are something that we will have to
work out from scratch when we get into the Weld, as well as the way in
which the grammar is organized, but we know exactly where to start. We
linguists therefore arrive in the Weld with a rough series of chapter
and section headings already set in our minds: Consonants, Vowels,
Phonotactics, Morphophonemics, Noun Phrases, Pronouns, Possession,
Relative Clauses, Verbs, Adverbs, Simple Sentences, Complex Sentences,
and so on. Obviously, we have to do some Wne-tuning, perhaps
abandoning some of the headings that we arrived with and introducing
new headings in order to accommodate particular features of the

language that we are dealing with. But the fact remains that we have
some kind of broad framework upon which we can hang the various facts
that we are about to discover. There are even published guides to tell us
what sorts of possibilities we can expect to encounter along the way
(Shopen 1985a, 1985b, 1985c, Payne 1997, Whaley 1997).
By way of contrast, the task of the anthropologist is much more
open-ended. While there are reportedly somewhere between 11 and 141
phonemes in all of the known languages of the world, there is nothing that
is remotely parallel to this unit of analysis in the study of culture. If there
were, then all behaviour within any given cultural group could ultimately
be broken down into a combination of a highly restricted number of
individual ‘behaviouremes’, which could then be combined in set ways
into larger behavioural units. Anthropologists have in the past exhibited
some envy for linguists and their phonemes and they have invoked the
linguistic distinction between etics and emics in their own work. However,
an ethnographer cannot arrive in the Weld with anything similar to the
chapter and section headings of linguists.
2
Ethical Issues
When I went on my Wrst linguistic Weldtrip in the early 1970s, there was no
expectation that I would have to prepare a submission for the ethical
approval for project. How times have changed! It is now de rigueur for
any research proposal that involves human subjects to pass scrutiny from
the appropriate Ethics Committee. Most linguistic research that is aimed
at documenting little-known languages, if it is conducted at all sensitively
and sensibly, should not face any serious challenges from an Ethics
Committee. However, there are some important ethical issues to which
any researcher should give serious thought, including a number of issues
that go beyond what any Ethics Committee will ordinarily be expected
to deal with.
2.1 Linguists and speakers
First and foremost among the ethical obligations of a Weld linguist are
those which involve his or her relationship with the people from whom
the linguistic data will be recorded. It is this relationship which an Ethics
Committee will normally be primarily involved with.
2.1.1 Ethical guidelines
Any responsible research institution these days is certain to have its own
speciWc guidelines for the submission of applications for the ethical
approval of research which involves human subjects, as well as a system
for overseeing these guidelines, including an Ethics Committee which is
charged with overseeing individual research projects involving human
subjects. Moreover, any outside funding agency is likely to make any
24 ethical issues
granting of funds for projects dealing with human subjects conditional

upon the ethical approval of the project by an appropriate committee.
DiVerent national Linguistic Societies have often established sets of
ethical guidelines for the conduct of linguistic research within their own
countries, or by that society’s members while carrying out research in
other countries. You should consult with your own country’s Linguistic
Society—as well as the Linguistic Society in the country in which you
plan to conduct your research—to see if they can provide you with
documents setting out issues that you should aim to address in
your project. These points will typically be geared speciWcally towards
Linguistics as a discipline.
If you are planning on carrying out research in a country other than the
one in which you are based, you may also need to gain ethical approval
from a relevant body in that country. If you have never been to the
country before, it may be diYcult for you to Wnd out what the procedures
are, but a relevant diplomatic mission or a creative search on the internet
should hopefully lead to the kinds of documents that you need.
Generally, the guidelines of such a committee will seek to ensure that:
. information which is provided by an individual is not made public in

a way that is harmful either to himself/herself or to others;
. anybody from whom you will be gathering linguistic data should be
properly informed of the nature of your research project;
. anybody who you approach should be free to agree to participate in
the project without any pressure from the researcher, and that he or
she should be free to withdraw from participation at any time
without having to oVer any explanation for this decision.
These kinds of issues should in most cases be easy enough to address if

you are aiming to produce a straightforward grammatical description of a
language, along with a dictionary or an accompanying collection of texts.
However, there are some possible pitfalls, which I will address in turn.
If you are unlucky enough to receive an initial response from your Ethics
Committee that seems to you to show a particular lack of linguistic aware-
ness, just remember that this may not necessarily be all the fault of the
committee members. It may be, in fact, that your original submission lacked
some essential background information which you incorrectly assumed that
non-linguists would already know. You should respond by carefully
providing the requested additional information and resubmit your proposal.
ethical issues 25
2.1.2 Avoiding harm
You will need to demonstrate to your Ethics Committee that you are
aware of any possible circumstances in which your research may result in
harm either to those who have provided you with information, or to
others. It is almost unimaginable that harm could arise to anybody as a
result of the recording of everyday words for ordinary things. Similarly,
many of the stories that you record as part of your research will
be intended primarily for enjoyment, amusement, or to provide moral
lessons for children. These kinds of stories are also unlikely to result in
harm to anybody. You should be aware, however, that what may appear
to be an innocent children’s tale recorded for posterity could later be
construed by that community as, for example, a claim for land based on
the speaker’s knowledge of the story and that story’s association with
spirits inhabiting certain areas of land.
Because linguistics at many universities is a relatively small discipline,
there may well not be a representative from this discipline on your own
Ethics Committee. You may Wnd that you are facing the task of trying to
get your project approved by a committee that is more used to vetting
proposals from Psychology or Sociology, where people may be soliciting
attitudes about highly sensitive issues such as transsexuality, spouse
abuse, internet pornography, or addiction to illegal drugs. The public
release of personal information obtained in interviews in such circum-
stances clearly does have the potential to harm either the interviewee or
other people. Because our discipline is very often poorly understood by
academics from other disciplines, it may occasionally turn out to be
surprisingly diYcult to convince committee members that the person
who is going to tell you that the word for ‘dog’ in the Eastern Rainforest
language is siliwan is not subjecting himself or herself to unspeakable
potential harm in providing you with this information.
However, you do need to understand that some of the things that
linguists do in the Weld can be damaging. Your description of the
language could, in principle, be based on nothing but recordings of
slanderous lies about other people. After all, we use the same grammar
and vocabulary when we are lying that we use when we are telling the
truth and these kinds of utterances would represent linguistically valid
data. Such recordings would, of course, be quite damaging if anybody
else ever heard them, so you will need to ensure that you avoid recording
any information of this kind that is known to be untruthful and hurtful.
26 ethical issues
Avoiding harm in the stories that you record goes beyond the need to
avoid deliberate untruths. I remember being told one story while carrying
out Weldwork in Vanuatu that was highly entertaining and, I thought, well
worth adding to my collection of stories. As soon as I suggested that we
proceed to record the story on tape, the narrator baulked. The story was
about some people who always did things the opposite way to ordinary
people. So, instead of eating the Xesh of yams, they just ate the skin,
instead of using pigs as exchange at weddings, they married them, and so
on. My presumption was that these were mythical beings of some kind,
but it turns out that in local tradition, these were seen as the ancestors of
a particular lineage on the island, and this lineage included the local
member of parliament at the time. Although everybody on the island
knew these stories, it was felt that to document them in writing risked
insulting this important person, and this might have opened up all
manner of political divisions that were lurking just below the surface.
A visiting linguist should clearly not be seen as provoking this kind of
controversy, so I did not pursue the issue of recording these stories
on tape.
Oral tradition in traditional non-literate societies often also serves the
purpose of providing a link between people and places by establishing
ancestry. While it might appear to the outsider-linguist that a story about
a family that is descended from some particular animal might be little
more than a story, it may well be that this story is implicitly taken by
local people to indicate that the family involved has rights to land in
whatever places are associated with the pig in the story. Such stories may
be entirely uncontroversial locally if there happen to be no disputes or
disagreements over the land in question. However, you may well be
unaware that members of another family assert their link to the same
land via a diVerent story, or perhaps even by the same story, but with
a few particular twists.
I recently quite innocently recorded, transcribed, edited, and distrib-
uted a story for which I was given no indication that it might relate in any
way to a local land dispute. When I returned to my Weld site on my next
visit, I was taken aside by a member of the family on the opposite side of
the dispute who explained that his family felt that their position on a local
land dispute had been misrepresented in the story. Fortunately, the
version that I had sent out was only a draft that was not very widely
distributed. I was able to apologize and I promised to excise the story
ethical issues 27
from any collections of stories that I might distribute more widely in

future. Things could easily have been much worse, and publicizing a story
unwisely could result in the welcome mat being removed by the
community. I don’t personally know of any linguists who have been
‘dis-invited’ from the community in which they were based, but I have
heard of anthropologists to whom this has happened.
2.1.3 Informed consent
Before your work gets under way, you will need to provide useful
information about your project to people from whom you are going to
be gathering data. In some societies, this may turn out to be quite diYcult.
Even explaining what you mean by a grammatical description of their
language, if that is what you are aiming to produce, will often be far
from easy. People will almost certainly not know what you mean by
‘phonology’, ‘morphology’, and ‘syntax’, so any attempt to solicit
informed consent should avoid such notions. And if your primary
interests are, for example, in matters such as split ergativity and its
relationship with animacy in noun phrases, then explaining this in any
kind of meaningful way to a linguistically naive audience is almost
certainly a lost cause from the outset. It is best, therefore, not even to try.
What you will need to do instead is describe your intentions in terms
that people are likely readily to understand, while at the same time not
being misleading. It has been necessary in my experience sometimes to say
to people something like ‘I have come to write your language’. Given that
your project will almost certainly entail gathering lexical data, and
gathering narrative texts in written form as a basis for grammatical
analysis, you can satisfy your participants’ need to be properly informed
by describing your project in these terms.
Since you will also be attempting to ‘learn the language by collecting
words and stories’, you can also tell people that this is what your plans
involve. Sometimes people have asked if I have come ‘to translate the
language’, to which I can truthfully answer that I have—though it has
sometimes been necessary to add that I would not be ‘translating the
Bible’ into the language, as people have sometimes assumed (and hoped).
I am certainly not the only linguist who has been faced with assumptions
about religious motivations. Lyle Campbell (pers. comm.) reports that in
Central America, sometimes the only reason that people would agree to
28 ethical issues
share their linguistic knowledge with him was their belief that he was
a priest, even though he had previously told them he was not.
2.1.4 Voluntary participation
When human subjects are being interviewed within your own community
for research projects in history, sociology, geography, and suchlike, your
Ethics Committee is very likely to insist that participants be provided
with a written consent form that each participant is expected to sign.
While this may be unproblematic with many non-linguistic projects, there
can sometimes be real problems with this when conducting linguistic
Weld research.
For one thing, one excellent way of recording linguistic information is
to note things as they are said in natural spoken contexts rather than in
a formal interview situation. It is quite impractical to insist that
somebody who has just spontaneously used a particularly interesting—
and previously unknown—word or grammatical construction in your
presence be asked to stop speaking so that you can have them sign a
consent form in retrospect before they be allowed to continue speaking so
that you can write down what you have just heard. After all, you are not
interested in the content of what has been said; rather, you are merely
interested in the manner in which that content was expressed.
While it might be appropriate to ask people in highly literate urban
societies to sign informed consent forms when they are participating in
interview-like direct elicitation, there are potential problems associated
with this in the less literate and predominantly rural societies in which
much linguistic Weldwork is carried out. One of the overriding principles
of ethical research is that the researcher should avoid any potential harm
to the subject. However, in a society in which an inability to read and write
may be reasonably common, it may cause considerable embarrassment
when someone is asked to sign a form that he or she cannot read, or when
that person may not even be able to sign their name.
The problems do not end there. Even if an individual can read and
write, it is possible that in a society where literacy is relatively marginal
and where quite diVerent sets of social assumptions are associated with an
ability to read and write, asking people to sign a formal document may
raise more doubts and suspicions about a project than a simple verbal
approval to participate. In some of the rural societies where I have been
ethical issues 29
working in Vanuatu, for example, the only legal documents that most
people have ever had to sign are contracts with overseas logging
companies. These contracts are then not infrequently broken unilaterally
by the companies involved, leading to all manner of nastiness. Associating
linguistic research in the minds of participants with these kinds of disputes
is arguably a very bad idea.
In the absence of signed consent forms, some Ethics Committees may
be happy for formal approval to be given by an appropriate person or
institution on behalf of participating individuals. This may involve a chief,
or perhaps a council of chiefs, or some other local equivalent, where the
chiefs traditionally operate only with the consent of community members
rather than operating dictatorially. On the other hand, if you feel that you
can ask for a signed consent form, then you could include a paragraph
outlining the speaker’s expectations about how the records you make will
be used. For example, you could include a statement that the recordings
will be used for educational and not for commercial purposes. As you will
no doubt be depositing your tapes in a linguistic archive it is a good idea
to clarify who can listen to them so that the archive can keep the data
according to your and your speakers’ wishes.
Ethics Committees may also ask the researcher to get people to signal
their willingness to participate verbally in a recording. Even without
being asked to do this, I have found that when people oVer to record
a story for transcription, their stories often begin with words to the
following eVect, which clearly signal their willing participation:
My name is so-and-so and I want to tell a story about such-and-such.
Even where the speaker does not speciWcally mention his or her own
name, I always record an introductory opening message which says
something like the following in a language that the participant can
understand:
Today’s date is such-and-such and I am in such-and-such a place, with

so-and-so who is going to tell a story in his/her language.
As long as the narrator clearly signals agreement with my introduction with

words such as the following, this can certainly be construed as an expression
of consent which should satisfy the needs of an Ethics Committee:
I want to tell a story about such-and-such.

30 ethical issues
2.1.5 Money and Weldwork
To produce a substantial description of any language is likely to take

several years of your time, but it is also going to involve quite a commit-
ment of time from your language-helpers in the Weld. In traditional rural
societies, life can often be a struggle. People may have to hunt and Wsh for
their meat, they may have to watch over their cattle, they may have to
plant and tend their crops, and they may have to gather Wrewood and
fetch water. This can all be extremely hard work, and in such societies
nobody can aVord the luxury of indolence. I learned this the hard way on
my last Weldtrip to Malakula when the mother of the family that I was
staying with went to visit her sons in town for a few months. Suddenly
I found myself occasionally having to take charge of preparing food and
cooking over an open Wre for working parties of Wfteen hungry men who
were building a house for the family. I certainly got no language
work done on those days.
Any demands that you make on somebody’s time to help you with your
project almost certainly eat into time that he or she would have devoted to
other more life-sustaining activities. You will therefore need to be
somewhat Xexible about the time that you plan to work together to
allow your language-helper to see to his or her own needs from time
to time. There is also a question of compensation for that person’s time.
The budget for your project must include provision for the payment of
language-helpers at a locally appropriate rate. Sometimes, local rates of
pay may seem unrealistically low to somebody who has spent their entire
life in a major western city. However, to insist on oVering rates of pay
similar to what you would pay for Weldwork help in New York when you
are in a tiny rural village somewhere in Cameroon is certainly not going to
Wx any of the fundamental economic inequities that have brought this
about. Moreover, paying people more than the going rate is likely to
cause problems for future workers. Dimmendaal (2001: 59) reports that in
one situation in Ethiopia, a foreign Wlm crew with a large budget paid
local people at high rates, which local people expected to be matched by
linguists who arrived later, even though they were operating with a much
more limited budget.
While paying a wage may often be entirely appropriate and very much
appreciated, you may need to become sensitive to local ways of giving
money to people. In Vanuatu, for example, you never hand cash over to
ethical issues 31
somebody in public, even if it is their money that you are repaying from
a loan. Cash is normally discreetly proVered inside your hand with only
the back of your hand visible. Alternatively, you can hand over cash
to somebody inside an envelope.
Direct payment of cash can of course sometimes cause embarrassment.
I had been working sporadically with one man on the transcription and
translation of texts. Christmas was approaching and I knew that people in
the village were Wnding things tough that year because copra prices had
fallen. I therefore discreetly oVered him a cash payment. He initially
refused, saying that he was happy to help me with my project because
of its importance in the community. I ended up having to reassure him
that the payment was only fair, and in any case, it was not my own
money, but it was money that had been set aside as part of the project.
To signal the oYcial—rather than personal—nature of the transaction,
I asked him to sign my receipt book for me.
While some people may be embarrassed about accepting cash, you
can expect to encounter others who may be rather more mercenary.
I remember one man who was well known in the community I was
working in for latching onto relatives from town who had jobs. Inevitably
there would be a request for money. A visiting linguist, of course, is also
fair game, especially if it is known that he or she has given cash to
somebody else who has been working on the linguistic project. I could
see this man constantly watching me so he could catch me alone. Once,
when he had successfully cornered me, he started talking about Wngers,
and the names in the local language for each of the Wve Wngers. I had
already recorded these words, and although I could guess where this was
leading, I could see no easy way out. Eventually, he asked me straight up
if I would pay him for these Wnger-words. Fortunately, our privacy was
disturbed when somebody else happened along at just the right time and
local mores dictated that talk of money had to cease. If that had not
happened, I might have had little choice but to get myself out of this
awkward situation by buying this man a packet of cigarettes.
Sometimes, it may be necessary to be Xexible about the manner of
compensation for somebody’s time, as direct cash payments may not
always be appropriate. It is best to be sensitive to local practices and if
payment in kind seems more appropriate, then you can by all means oVer
people some kind of gift instead. I had been working with one rather old
lady on Erromango. While her mind was razor sharp, she was suYciently
32 ethical issues
frail that even a trip to the local store to spend cash would have been quite
taxing. It was suggested that it would be more appropriate for me to buy
her a length of cloth, a few packets of sugar, and some blocks of laundry
soap rather than oVer her straight cash, and that is what I did. (Local
tradition dictated that a gift such as mine be reciprocated. This lady
picked up her walking stick and hobbled over to her basket where she
kept her money. She took out a note and handed it to me. While I felt
uncomfortable about accepting this, I was relieved to note that her return
gift was substantially less than the cash value of my original oVering.)
You also need to appreciate that in many rural areas, even with cash in
hand there is often a limit to what is available in local stores. People may
therefore prefer gifts that you have brought in from town where the stores
are better stocked. I have also found that a gift that is bought overseas
rather than in one of the local stores sometimes has a mystique about it
that makes it so much more appreciated. I have tried arguing that a Seiko
watch that is bought in New Zealand is the same Seiko watch that can be
bought in the local capital (where it is sometimes even cheaper), but
people still persist in placing orders for me to buy things in New Zealand.
I therefore often accede to hints—or direct requests—for all manner of
items, as long as the cost is not excessive, and transport is not an issue.
This has included items as varied as liquor, clothes, shoes, watches,
cassette-radios, and even a tent and a pair of binoculars. Of course, this
kind of payment can make for some diYculties in accounting. Your home
institution may require proper receipts for every penny that you have
spent, yet you can hardly give somebody a gift and then ask them to sign
a receipt for it. You may therefore need to acquire some skills in creative
accounting by shifting the cost of such gifts under some other heading.
Further reading: Dimmendaal (2001), Samarin (1967: 20–44), Bouquiaux and

Thomas (1992:36), Vaux and Cooper (1999:13).
2.1.6 Ethical delinquency
While linguistic research in the Weld should ordinarily not present too
many ethical problems in terms of your relationship to the individuals
who are providing your data, some linguistic projects in the past have
been problematic, and this has caused diYculties for others down the line.
For instance, there was once a serious proposal in Papua New Guinea
to gather together a number of monolingual people from structurally
ethical issues 33
diverse languages in what came to be referred to as the Desert

Island Experiment. The intention was to provide these people each
with the same small vocabulary and see how their communicative
needs within this artiWcially created community came to be handled
over a twelve-month period. The ultimate aim was to study the process
of pidginization with a view to Wnding out what sorts of grammatical
structures emerged.
There were, however, some very serious ethical problems with this
project as it was formulated. The monolingual Papua New Guineans
who were needed for the project to succeed would necessarily have been
uneducated, as well as coming from very remote communities. They
would therefore have had little contact with the outside world, and
would almost certainly have had little idea about how scientists
conduct their experiments. Explaining the nature of the project and
gaining people’s voluntary informed consent would therefore have been
extremely diYcult given that these people would not have been speakers
of any lingua franca. In the end, this project was canned—quite rightly
so—after having been condemned by linguists and political authorities in
Papua New Guinea for its exploitative and inherently unethical nature.
Few linguistic research projects of which I am aware have exhibited
anything like this level of human exploitation, and linguists have for the
most part tended to be seen as fairly harmless sorts of drudges who collect
obscure words and grammatical constructions. However, we do sometimes
get confused with anthropologists, and sometimes anthropologists are
confused with commercial artefact collectors or Wlm-makers. In some places
people have felt betrayed by the actions of such commercial activities, and
this may be reXected in a reluctance to participate in linguistic research. It
may therefore be necessary to explain in some detail precisely what sorts
of things you are interested in, and what you are not intending to do.
2.2 Linguists and communities
In addition to any general ethical responsibilities towards the individual

human subjects who participate in your research that may be governed by
the formal requirements of your Ethics Committee, any linguist also
needs to accept that they have a number of broader ethical obligations
vis-à-vis the language community as a whole.
34 ethical issues
Some countries—or parts of countries—have in the past declared

themselves closed to foreign researchers for several years because
researchers in the past had failed to produce copies of materials that
they have produced for members of the community to see and use. This
has included some linguists who have been seen collecting vocabulary and
oral traditions for some years, without publishing their results, or
even distributing informally produced copies for the beneWt of members
of the language community. To people in the community, it looked very
much as if the linguist was taking material away, and not giving anything
back to the community in return. Vanuatu was one country which
declared a research moratorium over an extended period for these sorts
of reasons.
Research opportunities have opened up again in Vanuatu, but the
approval of linguistic research is now subject to a requirement that the
researcher sign a legally binding contract with the Vanuatu Cultural
Centre, which states (among other things) that:
All research projects will include a cultural product of immediate beneWt and use
to the local community. . . . Such projects could include booklets of kastom
[i.e. traditional] information, photo albums of visual records, simple educational
booklets for use in schools . . . programs for the revitalisation of particular kastom
skills in the community, training workshops in cultural documentation, etc.
We researchers should therefore consider ourselves obligated to provide

something for the community in which we have done our work. What we
provide for the community by way of feedback should be something that
will be valued by members of the community, as well as being of some
kind of practical beneWt to people in the community.
2.2.1 Grammars
A researcher needs to be aware that in many situations a grammatical

description by itself will not satisfy the kinds of expectations that mem-
bers of a community may have. This is because most grammars are by and
large not comprehensible to non-specialists, and some grammars are
comprehensible only to small coteries of linguists who adopt very speciWc
theoretical assumptions and terminology. Given that most linguists are
primarily interested in grammatical research, while most communities are
more likely to be interested in other kinds of linguistic research outputs,
there is a real potential for conXict between community and academy
ethical issues 35
in descriptive linguistic research projects. This is still something of an

unresolved issue, and it is one that any intending researcher needs to be
well aware of in order to take the appropriate steps to keep the conXict of
interests to a minimum.
Let me illustrate this point with a personal example. Between 1994 and
1998, I was conducting research on the island of Erromango in Vanuatu.
My primary academic interests were as follows, ranked in the order given:
. the production of a grammar of the actively spoken Sye language;

. the production of a grammar of the moribund Ura language;
. the production of a bilingual Sye–English dictionary;
. the production of a bilingual Ura–English dictionary.
These interests—and their ranking—are governed in part by the

academic law of publish-or-perish. This dictum is increasingly being for-
malized in universities as some kind of ‘research quantum’ that is subject
to external audit, with government funding being dependent on scores
across the university. That is, I—and nowadays also my university—stand
to gain more academic brownie points by analysing the grammatical
system of a language than I would by publishing an alphabetized list of
words and their meanings.
There was also an issue of personal preference here, as I am more
interested in describing a viable language than a moribund one. I could
immerse myself into a social context in which Sye was spoken naturally
around me. I could form friendships where Sye was a daily medium of
exchange and I could learn about a new culture through its language.
With moribund Ura, however, I could not expect to have the same kinds
of interactions. First, the handful of speakers of Ura were dispersed
across several villages rather than being concentrated in a single
community. Second, most speakers of Ura were quite old and no longer
very mobile. This meant that there was not much prospect of ‘hanging
out’ with speakers and this did not seem very appealing to me.
However, judging by what people said to me on Erromango,
community expectations from my research seemed to be as follows (in
the order given):
. the production of a written collection of oral traditions in Sye;

. the production of an Ura–Sye or an Ura–Bislama dictionary;
. the production of a Sye dictionary of some kind.
36 ethical issues
You will see that a grammar of neither Sye nor Ura Wgured at all in
community expectations. While the word ‘dictionary’ is widely known
on Erromango, the word ‘grammar’ is probably only known to a very
small number of people who have tertiary or upper secondary education.
In fact, even those educated people who know the word ‘grammar’ may
think that it is something that applies only to English. I remember
eliciting a series of verbal paradigms from a speaker of Gela, a language
of the Solomon Islands. Eventually, when it dawned on him what I was
doing, his eyes lit up and he said, ‘Oh, I never knew that my language
had a grammar.’
People on Erromango are aware that much oral tradition has been
retained by the older generation, and that the earlier practice of passing
down stories at night around the Wre is disappearing. To ensure that
generations to come have access to these stories, people wanted to see
these stories recorded on tape, as well as in writing. The Vanuatu Cultural
Centre has for many years provided training and recording equipment to
enable local people to do this kind of recording. However, in the tropical
climate of this island, which has no proper tape storage facilities—nor
even any electricity—such material quickly deteriorates. Material has
been deposited in safe storage facilities in the Vanuatu Cultural Centre
archive in the capital, but for years the lack of cataloguing has meant that
once material is deposited, it can easily become inaccessible to all but the
most determined. In any case, the archives are located on another island,
which is an expensive forty-Wve-minute Xight away from Erromango.
Thus, recording material on tape by itself does not ensure the survival
of oral traditions; rather, it simply ‘museumizes’ them. Providing
an accessible written record of oral tradition therefore obviates these
problems.
Erromangans’ interest in an Ura–Sye or Ura–Bislama dictionary is also
perfectly understandable. In the last century, Ura was widely spoken in
the northern part of the island. However, massive depopulation and
resulting demographic realignments in the nineteenth and early twentieth
centuries have resulted in a situation where there are today only about
half a dozen elderly speakers of Ura left. It is clear to everybody that this
is a highly endangered language. The major language of the island today
is Sye. All adults on the island are also able to speak Bislama, the national
language of Vanuatu. Thus, it makes sense that people would be
interested in an Ura–Sye or an Ura–Bislama dictionary.
ethical issues 37
However, in order to Wnd a respectable publisher for a dictionary of

Ura from whom I will earn appropriate ‘research quantum’, it would
need to be an Ura–English dictionary. But an entry that translates the
Ura word lau as ‘heliconia’ is unlikely to be particularly illuminating to
most Erromangans. The simple fact is that most Erromangans have only
six years (or less) of primary education, and I very much doubt that the
English word heliconia is at all widely known. (In fact, how many native
speakers of English know what a heliconia is?) On the other hand, if
I were to gloss Ura lau as ndau in Sye, everybody on the island—of
whatever level of education in English—would know immediately what
it means. Given that Bislama is the national language, translating Ura
lau as Bislama lif laplap would also be immediately and universally
understood.
While an Ura–Sye dictionary would satisfy local needs, I would almost
certainly not be able to Wnd a respectable academic publisher to produce
such a volume. There would be only 1,900 potential buyers in the world,
all living on the island of Erromango, and most of those people could
not aVord to buy such a volume anyway, as they are living a largely
subsistence lifestyle. An Ura–Bislama dictionary would also fail to attract
orders from libraries overseas on which many small specialist publishers
depend for sales.
Finally, I was well aware that Erromangans badly wanted to see me
produce a dictionary of their Sye language. People seemed to compete
with each other to think of new words and expressions with
obscure meanings for me to include in my dictionary. I could not go
anywhere without people asking me if I had recorded the word for things
such as little toe, haemorrhoid, or a particular kind of soil that is only
found in caves (with a high content of bat droppings) which causes the
skin to itch if you get it on yourself. Older people were also keen to ply me
with words that younger people generally do not know, and which they
would like to see recorded for posterity. Some older people also asked me
to ensure that certain words should be recorded only with their ‘correct’
original meanings, rather than the innovative ‘incorrect’ meanings that
are encountered in the speech of younger people.
In the end, I did produce my grammars (Crowley 1998b, 1999:
106–225), and I also produced my Sye–English and Ura–English diction-
aries (Crowley 1999: 10–69, 2000), and I dutifully deposited copies of all
of these materials in the national archive. However, while I was happy to
38 ethical issues
return copies of the dictionaries to the local community—along with a

printed collection of stories, as I will discuss below—I have never been
able to bring myself to give people copies of the grammars that I wrote,
let alone even to show them to people on the island. I am overwhelmingly
conWdent that the sections headed Morphophonemics, Subordinate
Clauses, and Counterassertive Verbal Paradigms—in fact, just about
the whole thing—would be completely unintelligible to just about
everybody.
2.2.2 Dictionaries
Clearly, then, producing a dictionary of a language is part of the solution

to this ethical dilemma. It is particularly disappointing to note, however,
that while sometimes even the most eminent of descriptive linguists may
put considerable eVort into producing published grammars which are
widely distributed among the academic community, they make no
attempt to publish extensive dictionaries or text compilations which
may well be of considerably greater interest to local communities. Part
of our task as Weld linguists, then, should be to put pressure on any
ethically delinquent linguist who produces only a grammatical descrip-
tion of a language and who fails to follow this up with a published
dictionary, or some other body of material that is accessible to, and
appreciated in some way by, the local community.
But doing the right thing by a community is not as straightforward as
simply deciding to write a dictionary. In my experience, the kinds of
dictionaries that have been produced by linguists in the past have, by
and large, not been nearly as accessible as they could be to speakers of the
languages that are represented in them. Some of the faults which typically
lie with these dictionaries include the following:
(i) Dictionaries are typically produced bilingually with deWnitions given

in the linguist’s language. This will typically be a major world
language, which is more than likely going to be English. Many
speakers will not know enough of the major language to be able to
gain anything from the deWnition, especially as many meanings will
need to be deWned using fairly diYcult terminology, as I have already
illustrated with respect to a deWnition of Ura lau as ‘heliconia’.
(ii) Dictionaries often contain a lot of distracting information which
may require a considerable amount of eVort or education—and
ethical issues 39
sometimes even technical training in linguistics—to decipher. For

instance, Keesing’s (1975: 155) dictionary of the Kwaio language of
the Solomon Islands includes a deWnition such as the following (set
out as indicated):
2478 mooli 2.
*mauli *ML
mooli diVerently, a diVerent way, in a wrong direction
I invite you to imagine what this must look like to an uneducated speaker
of Kwaio, who will typically be bilingual only in Kwaio and Pijin, the
national lingua franca. Precisely why it was thought necessary, for
example, to indicate that this was the 2,478th entry in the dictionary is
beyond me.
(iii) Forms are often cited in non-occurring underlying roots rather than
widely used citation forms encountered in natural speech. Thus,
Crowley (1978: 213) cites the verb /wadji/ ‘tell’ in the Bandjalang
language of New South Wales in its underlying form, despite the
fact that it only ever appears in inXected forms such as the following
(some of which involve a systematic lowering in the Wnal vowel of
the root when lengthening is involved): /wadjini/ ‘told’, /wadjiya:/ ‘in
order to tell’, /wadje:fi/ ‘will tell’, /wadje:/ ‘tell’ (imperative).
(iv) Many dictionaries are produced by publishers who need to keep
costs to a minimum, so there is little sophistication in terms of the
creative use of font, illustrations, and layout. Newly literate dic-
tionary users are often faced by typographically fairly uniform
pages with no illustrations, and little use of indentation and
white space. The pages are, quite simply, daunting. There are
enough publications of this type that it would be unfair to refer
to a single author’s work as an example.
(v) Dictionaries are often presented in some version of a phonemic script
rather than a more recognizable spelling system. Crowley (1978) is
guilty of this for Bandjalang (along with many others with regard to
other languages). In the spelling that is preferred by the local
community, /wadje:fi/ ‘will tell’ is written wadhehyn, with /dj/
being represented by dh, vowel length by h, and word-Wnal /fi/ by yn.
(vi) Dictionaries are too expensive and diYcult for individuals to order
in remote rural communities where people often live substantially
40 ethical issues
according to a subsistence lifestyle. Linguistics publications typic-

ally cost many times more than what most people pay for the only
other book in their possession, such as the Bible and hymnals.
While these religious materials are usually easily obtainable
locally, the only way to obtain other published books is to obtain
a bank draft in foreign currency to order overseas, which in most
remote locations is simply not practical.
In Crowley (1992), I attempted to overcome some—though by no

means all—of these problems in my published dictionary of Paamese.
In particular, I put some of the distracting linguistic information, such as
underlying forms of roots, towards the end of an entry, with the entry
itself being cited under the least marked inXected form. Thus, for
example, the word for ‘eye’ was entered as meten, which means ‘his/her/
its eye’, rather than the underlying root mete- which never appears as such
in isolation. The verb meaning ‘fat’ is entered as raho, which means ‘(s)he/
it is/was fat’, rather than under the underlying root tahoo.
However, I also set out to produce a dictionary that would earn me
some academic brownie points, in that I intended it to include information
that academic linguists could make use of. Because I had attempted to
compromise—albeit only partially—in the direction of community inter-
ests, this meant that it was necessarily going to involve a bit more work
on the part of academic linguists to Wnd some of the kinds of information
that they are traditionally accustomed to being given on a plate.
An academic who goes too far in making his/her dictionary accessible
to a local community runs the risk of attracting criticism from his/her
fellow academics, i.e. you start losing brownie points. One reviewer of my
dictionary clearly recognized that it was intended to be more accessible to
the local community than many other dictionaries when he commented,
‘[Crowley] aimed this [work on Paamese] at a diVerent audience’
(Bradshaw 1994: 262). However, the reviewer also went on to say, ‘But
I do regret that the dictionary is of such limited utility to linguists’ (ibid.)
and he commented further that, ‘[T]he traditional audience of . . .
linguists will Wnd this dictionary almost as frustrating to use as non-
linguist speakers of the relevant languages will Wnd the other 120 books
[of PaciWc Linguistics]’ (Bradshaw 1994: 257). The very points that made
the dictionary relatively accessible to speakers of Paamese were, by and
large, the ones that the reviewer then spent a signiWcant proportion of the
ethical issues 41
remainder of the review criticizing. It seems, then, that an ethically

responsive lexicographer cannot easily win.
DiVerent linguistic situations in the world would call for diVerent kinds
of appropriate lexicographical responses. I have for the most part been
discussing situations where a language is actively spoken by a community
that makes little use of a major world language such as English as part
of its normal repertoire with in-group members. In many parts of
Aboriginal Australia, as well as in New Zealand, and also in Canada
and the United States, indigenous people often no longer speak
traditional languages natively, having switched largely to some variety
of English. The lexicographical needs of such people are going to be more
similar to those of the language learner than of the native speaker, so their
dictionaries will necessarily be bilingual. However, they will still need to
be stripped of purely linguistic distractions. Some community-friendly
work of this kind has been produced in recent years by academic linguists,
though much valuable lexicographical information still lies to a
large extent semi-hidden from community view by a strict adherence to
academic conventions.
A dictionary of a moribund language that is intended to be community-
friendly would possibly need to be more prescriptive than a dictionary that
is prepared with native speakers in mind because users would not have the
beneWt of native-speaker intuitions to fall back on in their interpretation of
material in entries. Thus, for languages that have large native-speaker
populations, a certain amount of phonological ambiguity might be
acceptable in a dictionary, with, for example, contrastive vowel length
perhaps not being marked. In the case of Maori in New Zealand, only
about 10 per cent of the total Maori population are native speakers of the
language. This means that the majority of dictionary users are now likely
to be non-native speakers learning the language in a classroom situation.
In this case, the unambiguous marking of vowel length becomes essential.
There is also likely to be a much greater need for illustrative sentences
in such a dictionary than would be required of a dictionary aimed at
native speakers. If a word is identiWed for its word class membership, it
could be assumed that native speakers will be able correctly to use that
word if the classiWcation of word classes is suYciently Wnely tuned.
However, when the users of a dictionary are predominantly non-native
speakers, the presentation of the form in an illustrative sentence (or two)
could be extremely helpful, and perhaps even essential.
42 ethical issues
Western linguists need to ask this fundamental question when working

on other languages: What kinds of dictionaries should be produced for
minority languages? In an ideal world, perhaps we should aim to produce
two diVerent dictionaries for every language, i.e. a linguist-friendly
volume and a separate community-friendly volume. With computer
technology, the same database could be adapted to these diVerent formats
without too much additional work, though the community-friendly
dictionaries would still need both a generous benefactor to Wnance their
production, as well as an academic who is willing to devote some of his or
her precious research time to this task for minimal academic reward. By
depositing the digital database in a suitable repository you will be
providing a useful basis for others to rework the data, perhaps to
construct other kinds of dictionaries, perhaps to use in lists of lexical
comparisons.
Given the likelihood that these goals will probably only be easily
achievable for a relatively small number of languages, we should prob-
ably aim to produce volumes that are going to be reasonably acceptable
to both audiences. Some academic dictionary users will therefore need to
accept that features of future dictionaries might present them with some
level of inconvenience. This might include the citation of inXected forms
of obligatorily bound roots rather than underlying uninXected roots. It
will also mean more work for lexicographers in that it may be necessary
to negotiate with a community regarding which particular inXected
form should be adopted as the citation form in cases where there is no
immediately obvious candidate. Lexicographers will also need to make
more creative use of font and layout choices, as well as illustrations. The
fact that many dictionaries have in the past been produced in IPA
script eVectively renders them of little value to community members.
Lexicographers should show themselves to be more willing to provide
entries in a carefully considered practical orthography that is acceptable
to a community, with a clear guide for linguists as to how this should
be interpreted in phonemic terms.
If a dictionary is to be published bilingually, then the English–
Vernacular section should ideally be much more than a simple reversal
of the Vernacular–English section, so that learners of English can look for
deWnitions of unfamiliar English words in the language that they know
best. This would again require more work from lexicographers, as they
would have to come up with useful vernacular deWnitions of words such
ethical issues 43
as codicil (and many more) which do not appear in the list of translation
equivalents in the Vernacular–English section. Ideally, they would need
to translate all of the deWnitions in, say, the Oxford Advanced Learner’s
Dictionary into the language. In reality, of course, they would need to
decide which additional English words can be most practically included in
the English–Vernacular section beyond the simple reversal.
This is essentially the kind of model that I followed in the production of
a recent dictionary of Bislama (Crowley 1995). Despite the fact that
I consciously adopted practices that I felt would be received positively
by non-linguists who speak both Bislama and English, it is a dictionary
from which I think a considerable amount of useful information can still
be extracted by academic linguists. At the same time, it was hard work to
produce that dictionary for me, and it was a task that I have felt unable
to replicate in my published dictionaries of other smaller vernaculars.
There is one kind of dictionary that would be unambiguously aimed at
speakers of the language rather than members of the academic commu-
nity and that would be a monolingual dictionary, in which words are
deWned in that language itself, rather than simply being given translation
equivalents in a major world language. However, such a dictionary would
normally be such a major task that it would be well beyond the capabilities
of the individual researcher, so this would not represent a practical
solution to the ordinary Weldworker’s ethical pressure to produce useful
and valuable materials for the community. In addition to the size of
the task, there is also a question of economics. The production of a
monolingual dictionary might be Wnancially viable for a language with
a relatively large number of speakers, as long as a substantial proportion
of the population has enough buying power, and as long as the language
in question has a conWrmed place within the education system of the
country. Even so, it is likely that such a task would be impractical
without a considerable amount of funding support, and the amount of
work involved would call for a well-coordinated team of lexicographers
working together for many years.
This is never likely to be a practical proposition for many of the
smallest languages of the world. However, it might still be possible to
meet conXicting demands by adopting a ‘mono-bilingual’ approach,
i.e. by providing bilingual glosses (combined with a comprehensive
English–Vernacular section), as well as monolingual deWnitions. This
would result in a certain amount of typographic clutter and it would be
44 ethical issues
necessary to Wnd ways to clearly distinguish material in diVerent

languages to make such a dictionary usable to its two audiences.
2.2.3 Reading materials
Another obvious contribution that a researcher can make is to produce

some kind of community materials that will feed into community interests
in promoting vernacular literacy. It would therefore represent a valid
contribution alongside your grammar to respond to a request Wrst to
help in the development of a spelling system that is acceptable to the
community, and also to help in the production of reading materials which
make use of that spelling system.
With regard to the development of a writing system, many of us learn
very early on in our linguistic training that the ideal spelling system for a
language is one which matches each distinct phoneme in the language
with a unique orthographic symbol or, when no appropriate single-
symbol spelling is apparent, perhaps making use of a digraph or a
diacritic. In sticking rigidly to this requirement, many of us seem to forget
that a spelling system which falls short of this idea can still be perfectly
serviceable. Witness, for example, the perfectly readable horrors of the
spelling system of the language in which I am writing right now. What is
more important for a spelling system than its strict adherence to the
phonemic principle is systematic usage and community acceptability. It
may well be that a community is going to be perfectly happy to use a
spelling system which fails to mark certain phonemic contrasts.
For example, in the Paamese language of Vanuatu, there is a phonemic
contrast between long and short vowels. This is not a contrast that is
found in just a handful of words either. It has very high functional load,
with lots and lots of minimal pairs. The Paamese have been writing their
language for about a century now using a spelling system that was
originally devised by Presbyterian missionaries who had never heard
of phonemes. In that spelling system, short and long vowels were
represented identically, yet this does not seem to cause any problems for
speakers of the language. They seem to be able to work out which vowels
are to be pronounced long and which are short from the context in which
the word appears. A linguist would be taking a great risk in a situation
such as this in producing a collection of reading materials intended for
community use in which a revision to the spelling system is foisted upon
ethical issues 45
the community, at least without some kind of process of consultation

with and acceptance from the community.
There are also cases of spelling systems which do make all of the
phonemic contrasts but do so in a way which may not appeal to the
linguist for some reason. I can think of another language where there is a
phonemic contrast of vowel length where the tradition in the community
has been to mark the long vowels with macrons (or not to mark length at
all). A linguist working on this language produced a very substantial
dictionary in which long vowels were written instead as double vowels,
i.e. aa, ee, ii, etc. When this dictionary appeared, people in the community
wanted nothing to do with it even though otherwise it was an excellent
piece of work.
So, a linguist needs to learn some humility when it comes to proposing
a spelling system for a language. You may think that a certain
orthographic solution is particularly elegant, but you must be prepared
for the possibility that you may not be able to sell your idea to the
community. For instance, I have been involved in the preparation of
reading materials and a dictionary for the Neve‘ei language of Vanuatu.
In this language, there is a contrast between ordinary bilabial consonants
and labio-velar consonants. This contrast is also found in Paamese, and
the established tradition there is to write the labio-velars with a tilde,
e.g. m̃, ṽ. This is a practice that is fairly common in other languages of
Vanuatu where the same contrast is made.
I therefore suggested to Neve‘ei speakers that they could represent their
labio-velars in the same way, and they initially agreed. I went ahead and
produced a substantial collection of draft reading materials based on
transcriptions of recorded stories, as well as a draft dictionary. I sent
this back to members of the community to look at. On my next visit,
people were asking me what the strange wiggly thing was in the materials
that I had written. It seems that even though they had agreed that this was
a reasonable way to write these sounds, they were not familiar with the
use of this symbol in other languages, and they were not able to work out
how I had used it for theirs.
So, I had to sit down once more with people and come up with a better
solution. It seemed that they were far happier in the end using digraphs,
and opted instead for the spellings mw, vw. Fortunately, computers make
it relatively easy to eVect such changes with the ‘change-all’ command,
and I sent out a completely revised version of the earlier materials.
46 ethical issues
When I went back to the village the next time, there was not a single
comment along the lines of: What’s that strange w in these materials?
That led me to assume that perhaps people were much more satisWed with
the new spellings that I had used for their language.
If a community wants to initiate some kind of initial vernacular
programme for their children, they may legitimately expect you to help
in the production of appropriate reading materials which goes beyond
assistance with the development of a spelling system. You may feel that
this is outside your area of expertise, but you can always seek the advice
of others who do have such expertise if you are not conWdent. If there are
members of the Summer Institute of Linguistics active locally, they will
often respond positively with advice when they can.
This kind of assistance need not represent a major drain on your time
or energy, especially if the community is enthusiastic about the project. In
the village of Vinmavis in Vanuatu, for example, a church-run school
wanted to encourage its children to begin reading in their own language
rather than through the medium of English, which none of the young
children could speak, or even understand. A teacher at another school
had already produced a series of Big Book readers aimed at pre-school
children. The text was handwritten in Bislama, and each line of text had a
page-sized colour illustration that teachers and other community
members had drawn themselves. The decision was taken to translate the
text of these stories into the local Neve‘ei language in association with the
same illustrations. None of the stories was more than half a dozen lines
long, and translating the text of a few dozen stories into Neve‘ei did not
take more than a day when I sat down to work with a couple of the
teachers.
I was then able electronically to scan each of the illustrations, and
I typed in the relevant line of text beneath it in a large clear font. The
pages were printed in colour, spiral-bound, and provided with a clear
plastic cover at low cost. For relatively little eVort, as well as limited cost,
the community now has a collection of vernacular stories which, for the
Wrst time, will allow pre-school teachers to read stories to their children in
their own language. If the books are damaged or lost, I can easily produce
a new printout and send it oV to the community. Having seen these Big
Books, members of the community will see how easy it would be to
produce more of the same, and it will not be diYcult for me to help in
their production if that is what the community wants.
ethical issues 47
Another contribution that you may be called upon to make is the

provision of written forms of any oral tradition that you may have
recorded as the basis for your grammatical study. This is something
that communities may well see as very important, because the older
traditions of passing down stories are often being eroded. This is also
something that you can provide which will not be too diYcult for you
to do.
While this may seem like a relatively straightforward process, there are
still some signiWcant issues that you will need to pay attention to. First,
simply copying your transcriptions, even if presented in a practical
orthography, is unlikely to represent a satisfactory response to a commu-
nity’s expectations. If your transcriptions accurately reXect the way that
somebody has spoken, unless a storyteller has remarkable narrative skills,
then it is likely that they will contain a number of diVerent kinds of
performance errors which people will be most unhappy about seeing
recorded in writing in perpetuity.
This will involve the local equivalent of ums and ahs, which are the
easiest elements of all to edit out. However, there are other performance
issues which you will need to be aware of. People sometimes accidentally
give a character’s name wrongly, or mention an incorrect place where
something happened. While this may seem unimportant to you, it can
sometimes be of great importance locally, as all sorts of other things may
follow from the story that you are unaware of. For example, a story
may provide some kind of justiWcation for one group’s ownership of a
particular area of land, in which case it is obviously going to be of
considerable importance to ensure that the details are recorded correctly
in a written form of a story that is going to be distributed publicly.
Sometimes when telling a story, people skip ahead to a point that has
not yet been mentioned and then they backtrack to the original point in
the story. Sometimes also, the narrator may say something like He said
such-and-such and He did this, without clearly signalling which character
in the story is talking and which character is performing a particular
action. Often, because local people are already familiar to some extent
with the content of the story, this does not cause any confusion. In any
case, there will often be some kind of subtle clues as to what’s going on in
the narrator’s intonation or the facial expression, rather than in the
words. However, when you are converting a written transcription into
a written text, local people may feel that it is necessary to reorganize
48 ethical issues
certain elements of the story as it was told, and to insert additional clues
clearly to indicate who is saying and doing what. This is something that
you can’t always do by yourself, so you will need to have local editorial
help in converting your transcriptions into acceptable written texts.
2.2.4 Technical advice
A linguist’s input into a community need not just take the form of
published outputs. Given the lack of trained linguists in many PaciWc
countries, one legitimate activity that we can engage in is the provision of
technical advice to governments and other agencies in these countries
(Bradley 1998: 52). Such advice could then be taken into account in the
formulation of social policies, whether implemented by government
or non-government organizations. There is a wide range of published
material already available relating to choices which are available for
diVerent countries, and linguists could provide their personal perspectives
on these materials in relation to individual countries on the basis of their
own Weld experience.
It must also be recognized that attention is most likely to be paid to
such advice when it is given in response to a local request rather than
when it is simply oVered by an academic (or, as sometimes happens, when
it is eVectively thrust upon a local audience). Local people all around the
world have had to develop strategies for listening to the advice of overseas
‘experts’ who sometimes spend only a week or two in the country before
depositing their wisdom and then leaving (while drawing upon often quite
hefty consultancy fees). Such advice is not infrequently politely received
and then—sometimes quite justiWably—ignored in large part, or even in
total. EVective action can only come from internally motivated policies,
rather than those imposed from outside (Ostler 1998).
2.2.5 Public awareness
Public awareness of the current linguistic situation in a country is an

essential element in any plan for concerted action regarding language
maintenance. Kaplan and Baldauf (2002: 222–5) discuss prestige
planning as one of four major types of language planning activity,
alongside status planning, corpus planning, and language-in-education
planning. Professional linguists clearly have a legitimate role to play in
making people aware of what has happened in comparable situations
ethical issues 49
in other countries, and what kinds of activities have proved useful—or

less than useful—in those countries.
Education of this kind can be provided through local tertiary
institutions and the various national teacher training institutions, as
well as a range of non-government organizations, and the national and
regional media organizations. Linguists could also make use of their own
social networks in the countries where they have worked to provide local
people with knowledge about what is already happening regarding the
status and use of local languages, as well as reasonable and viable ways of
acting against any negative trends.
Closely related to the issue of public awareness, of course, is the
question of community motivation. It is arguably not enough simply to
make people aware of issues relating to language maintenance, as
language communities need to be encouraged to feel conWdent in their
own ability to stand up against globalizing socio-economic and cultural
pressures which serve to threaten the future viability of minority
languages. The kind of action of this kind that could be practicably
performed by a linguist would be making people aware not only of the
dangers facing a language on the basis of experience elsewhere in
the world, but also making people aware of some of the successes
that communities have enjoyed in maintaining their languages against
external threats.
Much useful work in the area of public awareness is fortunately already
being done, though arguably only a small proportion of those whose
opinions need to be inXuenced have been successfully reached in this
way. Tertiary institutions have a capability to re-educate inXuential
people in policy-making positions, though in my experience the shortage
of such well-educated people, combined with the political instability
of many newly independent countries, means that there is often little
guarantee that a positive-thinking individual will necessarily be in the
same area of responsibility in a year’s time.
The activities to which I have just referred essentially represent a
top-down approach in which the eVects of awareness programmes are
expected to trickle down to the community at large from people placed in
positions of responsibility, most probably in the public sector. However,
public awareness can also be promoted via a bottom-up approach in
which emphasis is placed on raising awareness at the local level. Faraclas
(1996), for example, discusses the promotion of vernacular literacy in
50 ethical issues
the context of a programme of more general language awareness in Papua

New Guinea. However, individuals and non-government organizations
can often provide inXuence only on a small scale at a local level because of
funding limitations or a restricted infrastructure. In any case, it would be
very diYcult for a professional linguist to have any substantial input into
language programmes beyond a very restricted geographical area with
which they are familiar.
2.2.6 Terminology development
It is rather more diYcult for a linguist to provide answers to the perennial

question of how speakers of a PaciWc language might express introduced
concepts without using borrowed vocabulary. Some descriptive linguists
are loath to regard borrowings as necessarily bad, though it must be
acknowledged that there is a widespread implicit tendency even among
linguists to view borrowings prescriptively as suspect.
A linguist can obviously make suggestions to speakers of a particular
language as to how a new meaning might be expressed based on what he
or she knows of how such issues have been resolved in other language
groups, whether as a result of spontaneous linguistic development or
conscious language planning. Amery (pers. comm.) reports that a base-
ten counting system has been created for the Kaurna language of South
Australia—for which there was originally no counting system—out of the
roots of a no longer used series of birth-order names for children.
However, the question of whether a ‘plane’ should be expressed in a
particular language by simply borrowing the word from English (as, say,
plen), by extending the meaning of an existing word (such as Paamese
aman ‘bird’), or by joining together two existing indigenous roots to make
a new word (such as Fijian waqa-vuka ‘boat-Xy’) is going to depend
entirely on local community preferences. Once again, therefore, the
input of a linguist is likely to be restricted, with the Wnal outcome being
decided on the basis of community preferences.
2.2.7 Recognizing our limitations
Beyond these areas of activity, however, it is diYcult in many cases to

imagine what more any individual linguist could possibly be expected
to achieve with regard to language maintenance, apart, perhaps, from the
very limited degree of language revival that has been reported in situations
ethical issues 51
such as that to which I referred earlier for Kaurna. There are certainly
other areas of activity which can be considered as possibly contributing to
language maintenance, such as the incorporation of vernaculars into the
education systems, or the promotion of opportunities for social and
economic advancement without the need for urbanization or overseas
(or internal) migration. However, actions in these areas are clearly
matters which lie outside the purview of professional linguists, and are
for the most part beyond the inXuence of any single individual.
We no longer live in a world where outsiders can come into a recently
independent country and tell people what they must do, how they must
speak, or even what language they must (or must not) use. The ultimate
choices are now up to speakers of local languages themselves. If they
make choices which lead ultimately to the loss of their own languages, our
responsibility as linguists can only extend to making people aware of
what the related issues are, and what are the various options that might be
available along the way.
Such an attitude may suggest that I am being dispassionate about the
possible loss of indigenous languages. However, I know of no linguist
who doesn’t experience considerable angst at the loss—or imminent
loss—of a language. I have been a close witness to more examples of
this phenomenon than I care to contemplate. While I do not Wnd this loss
any easier to accept with each new situation that I encounter, I recognize
that it would be presumptuous of me to try to dictate to people how and
when they should speak their own languages.
It must also be recognized that a linguist’s sense of loss does not
necessarily always match up with what members of a local community
are feeling. In the case of Ura, for example, it would be fair to say that
people are mildly sad at most that the language is moribund, and Lynch
(1998: 270–1) indicates that speakers of other vernaculars in the PaciWc
have even enthusiastically embarked on a path of language shift.
Attempting to preach against language shift in these kinds of situations
would be akin to a situation where a linguist who drinks Coca-Cola tries
to tell other people that Coca-Cola is bad for them and they should refuse
to drink it, restricting themselves to drinking only fresh coconuts.
Another disconcerting trend that faces many academic linguists is the
fact that market forces now dominate in tertiary education in the western
world, with an increasing tendency towards the commodiWcation of
courses—and even research—in contrast to the older emphasis on tertiary
52 ethical issues
education primarily for the sake of learning in its own right. Many
of us feel these days that we are teaching at the University of Bums-
on-Seats.
Given that there are still several thousand languages in the world
awaiting documentation—and a substantial number of these are under
varying degrees of threat—it would take several thousand additional
linguists quite a few years of their academic careers to carry out
this task. It seems unlikely that university administrations and funding
agencies world wide will commit their shrinking resources towards
funding such a large-scale enterprise, when even holding onto existing
teaching positions in academic institutions in linguistics these days is
enough of a challenge.
Traditional linguistic analysis is therefore not nearly as ‘sexy’ as it once
was among vice chancellors and deans as more and more people are being
tempted by the money-earning potential of a qualiWcation in English-
teaching. Linguistics, after having concentrated since Sapir’s time
on describing a wide variety of small and previously undocumented
languages, is in serious danger of being recolonized by specialists in
English (or other major languages), to the exclusion of minor or
threatened languages.
2.3 Linguists and other linguists
The Wnal ethical guidelines that we should keep in mind relate to our
relationships with our colleagues. I am not talking just about standard
issues such as the proper acknowledgement of other people’s work, the
avoidance of plagiarism, the need to be sure that we have accurately
reXected somebody’s views when we discuss their work in print, or the
need to avoid the falsiWcation of our data. Rather, I want to concentrate
speciWcally on Weldwork-related behaviour.
First, given that there are so many undocumented languages out there,
the Weld does not need to be a crowded place. Some linguists have a
tendency occasionally to be a little territorial and do not like to feel that
other researchers are getting too close to their Weldwork area. Rather than
viewing somebody who is working in a language close to where we
are working ourselves as a threat, we should of course welcome the
opportunity for more data to be made public from a nearby language.
ethical issues 53
At the same time, in choosing your own Weldwork site, you need to be
aware of people’s possible sensitivities. This means that you should
consult ahead of time with people who know an area to make sure that
you are not going to end up stepping on somebody else’s toes with your
project. I can think of one case where a linguist had been working on one
language for some time, and there were several small languages spoken
nearby which she might also ultimately have taken the opportunity to
gather data on. However, another linguist then arrived, began gathering
data, and only when his work was well under way did he ask, ‘You don’t
have any problem with me working on these languages, do you?’ That
could be seen as rather like a smoker lighting a cigarette and then saying,
‘You don’t mind if I smoke, do you?’ It would have been so much more
considerate not to have presented the other person with the fait accompli
in the Wrst place.
It particularly makes no sense to plan on describing a language that
somebody else already has plans to work on, or that they have started
work on but have not yet had an opportunity to complete their work. In
such cases, you should graciously shift your attention to another Weld
site, as there are bound to be plenty of other opportunities, unless your
intention is to work on the language from a substantially diVerent
perspective from that of the original researcher.
Any Weld linguist should regard it as a duty to make his or her data
publicly available in some kind of publication. Publication means that
your data can be seen and easily cited by others. This means that a
grammatical description and an accompanying dictionary for the language
you have been working on should be published. Primary data (Weld
recordings, transcripts, texts, and the lexical database that is the source
for the published dictionary) should be deposited with a linguistic archive.
It can be very frustrating to see a draft copy of somebody’s work that
comes with warnings that the work is preliminary and that it should not be
cited, but where the Wnal work never appears. I know of far too many cases
where people have worked on languages for many years—sometimes even
decades—and they have never produced a grammar in a form that can be
publicly cited. Of course, sometimes circumstances may prevent an
individual from publishing a linguistic description. In such cases, linguists
should be prepared to provide copies to other interested people.
Keeping one’s data to oneself is not just selWsh; it is also risky. If your
unpublished work has been deposited in a professionally curated archive,
54 ethical issues
it may be relatively safe. I have heard of far too many cases where the only
copy of a potentially valuable set of data has been lost in cyclones, Wres,
volcanic eruptions, earthquakes, or where data has simply been mislaid.
The last speaker of the Utaha language of Erromango died in 1954.
His language was almost completely undocumented, so before he died,
he apparently wrote down quite a lot of material in some notebooks.
However, nobody now knows where those notebooks are. The linguistic
material that was lost in those notes was completely irreplaceable.
As a Wnal point regarding the storage of data, you may Wnd that your
Ethics Committee has established a generic guideline that primary data
should be destroyed after a given period. When people in a variety of
social sciences are conducting surveys and interviews, it is essential that
the data be available to the researcher while the analysis and write-up is
proceeding. The data must also be available for some time after that so
that any challenges to publications arising out of this data can be met.
However, some committees prefer to see primary data destroyed after
a given period as a way of guaranteeing the privacy of the various
participants.
When faced with this prospect in relation to ethical approval for one of
my linguistic projects, I was able to persuade the relevant committee that
in the discipline of linguistics it would be the destruction of primary data
that would be unethical rather than its non-destruction. Speakers of
languages typically agree to participate in projects of linguistic documen-
tation precisely so that the data that they provide will be available in
perpetuity, especially if their language is one that is likely to disappear
with the passing of the current generation of speakers. We cannot hope to
extract every single piece of useful information from our recordings, as we
cannot always foresee what sorts of questions people will be asking about
language data in the distant future. If, in 200 years’ time, all of the
relatively small number of languages in which there are bilabial trills
have become extinct and I failed to keep my recordings of the
Avava language of Vanuatu in which such sounds are quite common,
my foolishness would rightly be condemned by future generations
of linguists.
I have been talking about the role of academic linguists in the
promotion of indigenous languages so far on the assumption that the
linguists in question are expatriates, and for the most part Europeans.
This is in fact true in large part in many parts of the world. From time to
ethical issues 55
time, one hears arguments about the lack of inXuence of speakers

of indigenous languages in descriptive linguistics, usually in private
conversations, at conferences and seminar presentations, or in email
discussion forums (though relatively rarely in published form). Some
such commentary engenders spirited debate, as it is very often formulated
in terms of white western academics consciously or subconsciously
excluding indigenous linguists from working on their own languages. In
such forums, linguistics is eVectively painted as a neo-colonial discipline
(though it should be acknowledged that there are relatively few areas of
social science research where the same would not be true).
In reply, it can be pointed out that some conference organizers go
to great lengths to attempt to include native-speaker linguists in their
conference programmes. In some cases, the native-speaker linguists who
participate do so on a completely equal basis with other linguists, all
sharing the same kinds of descriptive and theoretical interests (such as
has happened recently with the series of Austronesian Formal Linguistics
Association conferences). However, in order to maximize the participation
of speakers of PaciWc languages in linguistic conferences, it has
usually proved necessary to include specially designed sections of the
conference relating to applied issues (such as happened, for example, at
the Fourth International Conference in Oceanic Linguistics held in Niue
in the PaciWc in 1999), or to organize completely separate conferences
exclusively around applied themes, to which descriptive and theoretical
linguists are either not invited, or in which it is accepted that they will have
minimal interest (such as with the PaciWc Languages: Directions for
the Future conference held in Vanuatu in 1984, the Vernacular Languages
in South PaciWc Education conference held in Vanuatu in 1988, or
UNESCO-sponsored conferences held in Papua New Guinea in 1999
and in Fiji in 2001).
A straightforward insistence that conferences on descriptive or
theoretical (or historical) linguistics must be attended by native speakers
can turn out to be as patronizing as any assumption that native speakers
should not be present, if those people happen to have primarily applied
interests. Those in attendance are likely to be either bored or confounded
by the theoretical presentations, and nobody is likely to beneWt from the
attendance of native speakers.
There is, however, a rather more objectionable assumption which is
implicit in this kind of argument, and that is the idea that there
56 ethical issues
is something inherently more worthy about descriptive, theoretical, and

historical work in linguistics over linguistics in the PaciWc that is applied
to language maintenance issues. This kind of thinking is surely just as
patronizing as that which seeks to exclude speakers of local languages
from less applied forums. Linguistics as a discipline therefore needs to
accord greater recognition to the value of applied—as against purely
theoretical and descriptive—activities in order for the contributions of
indigenous linguists to be fully recognized. In this regard, then, Dixon’s
(2001) view that the only worthy linguist is one who has produced a full
descriptive account of a language must be resisted by emphasizing the
validity of a wide range of diVerent kinds of linguistics.
Keeping in mind that in many parts of the world, there are still
relatively few—if any—speakers of indigenous languages who are
formally qualiWed to practise as linguists, an outsider-linguist should as
far as possible seek to involve any qualiWed insiders within our projects,
and to plan genuine research partnerships leading to jointly authored
publications. I am aware of one case where somebody with a postgraduate
qualiWcation in linguistics who is a native speaker of a moribund language
had an interest in doing descriptive work on that language. However, an
outsider linguist happened along and effectively pulled the rug from be-
neath their feet by independently gathering a lot of data, doing all of the
analysis, and gaining sole authorship of the resulting publication, leaving
the indigenous linguist feeling rather let down. This kind of behaviour is
something that we should all seek to avoid.
Further reading: Dwyer (2006), Rice (2005).

3
Getting Started
3.1 The fieldworker
In its simplest terms, linguistic Weldwork involves a language, somebody

who speaks the language, and you, the Weldworker. Some linguists are
drawn to Weldwork, and I am one of those. Right from when I started my
undergraduate programme in linguistics, I was fascinated by stories of
linguistic discoveries involving previously undocumented languages.
I was lucky enough to be sent on my Wrst Weldtrip at the end of my
second year of undergraduate study. Linguistically speaking, it wasn’t
a very productive trip. I had been sent to Tasmania to record what little
was still remembered of the original Aboriginal languages of the island,
which certainly was not much. But it was nevertheless exciting to try to
relate what I had recorded with what had already been documented. I was
hooked, and since that time, I have probably carried out some kind of
Weldwork every year.
Other linguists seem to regard a stint of Weldwork as something like
a rite of passage. It is something they endure—and possibly, to an extent
at least, even enjoy—early in their career. When they have served their
time, they then retire to the comfort of their oYce for the remainder of
their tenure. Yet other linguists, while recognizing the importance of
Weldwork, express puzzlement as to why sane academics would willingly
submit themselves to all manner of discomfort, and possibly even danger,
by disappearing into the bush for months on end to study a language.
Of course, every linguist has something to oVer the discipline. If you are
one of those who shudders at the very thought of a day-long hike across
a tropical island with a heavy pack on your back, or hanging over the rail
of an ocean-going vessel while you feed the marlin, then we will all be
better oV if you stay where you are. It is almost impossible to generalize
58 getting started
about what sort of person will make a good Weldworker and what sort of
person will not. While there are certainly some linguists I have met who
I would never want to inXict on any community, people with a wide
diversity of personality types have succeeded at Weldwork in the past. I
have known successful Weldworkers who are extroverted and introverted;
boorish and boring; technophile and technophobic; young and old; male
and female; single and married; religious and non-religious; heterosexual,
homosexual, bisexual, and asexual.
What you do need is solid training in all of the main areas of descriptive
linguistics: articulatory phonetics, phonology, morphology, and syntax.
A knowledge of cross-cultural communication and lexicography is also
helpful, though many people seem to be able to pick up what is needed
once they get going. But in addition to all of this, you need to have
a burning desire to do Weldwork.
3.2 Choosing a language
You, as a potential linguistic Weldworker, will need to make your own

decision about which language family you are going to work in. A wide
variety of considerations are likely to come into play when you make this
kind of decision. You may have an interest in particular kinds of
languages (tone languages, languages with bilabial trills, languages with
free word order, ergative languages, languages with serial verbs, object-
incorporating languages) or languages of a particular geographical area
(Papuan languages, South-East Asian languages, West African languages,
Amerindian languages) or languages from a particular genetic grouping
(Australian languages, Austronesian languages, Mon-Khmer languages).
Since successful Weldwork will in many cases call for an advanced
command of a language other than English—French in many parts
of Africa, Portuguese in Brazil, Spanish in Central and most of
South America, Bahasa Indonesia in Indonesia, Melanesian Pidgin in
Papua New Guinea, Solomon Islands, and Vanuatu, Mandarin in China,
Hindi in many parts of India, Russian in Russia, and so on—you may
also be guided in your ultimate destination by your current language
capabilities. If you have an excellent command of French but you know
no Spanish, it probably makes more sense to look for a Weld site in
francophone Africa rather than Mexico.
getting started 59
Unless personal factors come into play in determining which particular

language community you are going to try to work in—maybe you have
close friends from a particular place, maybe you are married to somebody
who speaks a particular language, maybe there is a fantastic surf beach
that you want to be based near—then you are going to have to make some
decisions about your preferred speciWc destination. You may be lucky
enough to have taken a specialist linguistics course in the languages of
a particular area or a particular language family which will give you some
of the background information that you need to help you to zero in on
a suitable Weld site. Alternatively, you may be attracted to a Weld site as
a result of contact with a language during a linguistic Weld methods course
in your undergraduate years.
If you haven’t made any decisions about where to go, then the Wrst
thing you need to do is some background reading. These days, there are
increasing numbers of survey volumes produced by major publishing
houses that provide excellent overviews of particular groupings of lan-
guages. Surveys of this type typically include up-to-date and detailed lists
of additional sources. Curzon Press, for example, has survey volumes in
print—or at least planned—in its Curzon Language Family Series on the
Oceanic languages, the Austronesian languages of Asia and Madagascar,
the Khoesan languages, the Manchu-Tungusic languages, the Sino-
Tibetan languages, the Indo-Aryan languages, the Iranian languages,
the languages of the Caucasus, the Bantu languages, and the Mongolic
languages. Cambridge University Press also has survey volumes in its
Cambridge Language Surveys series on Australian languages, Dravidian
languages, Indo-Aryan languages, North American languages, and Pap-
uan languages. Routledge have produced—or are planning—a number of
volumes in their Routledge Language Family Series, including surveys of
Bantu languages, Indo-Aryan languages, Mongolic languages, and
Oceanic languages. With just these publishers, a substantial proportion
of the world’s language groupings are represented, and a little careful
searching in library catalogues or on the internet should fairly quickly
enable you to Wll any gaps.
These volumes will typically provide quite a deal of information about
the language family as a whole. It is important to get some idea of the
kinds of features that are widely distributed in the languages of the family
that you are planning to work in. This can save you from having to
reinvent the wheel if other people have successfully tackled a similar
60 getting started
problem to one that might arise in your own data. For instance, if you are
planning on working on an Australian language, it helps to know from
the outset that nouns in many of these languages inXect according to an
ergative-absolutive pattern, while pronouns are more likely to inXect
according to a nominative-accusative pattern. If your chosen language
comes from the Oceanic subgroup of the Austronesian family, your
preparatory reading will indicate that you can expect to encounter
a particularly complex possessive system on nouns. It will not be possible
to say exactly how the system will work in any individual language,
but you can expect to Wnd some kind of overt grammatical distinction
between what we refer to as alienable possession, i.e. the possession
of things over which we have some kind of choice, and inalienable
possession, which refers to the possession of things where no choice is
involved (such as body parts and kin terms).
Choosing a particular language within a larger grouping can be subject
to fairly random factors, which often depend on issues of personal
preference. For instance, when I Wrst thought about doing Weldwork in
Vanuatu as a graduate student, I initially had no particular language in
mind. At the time, there was a large number of almost completely
undescribed languages, and because there were few researchers active in
the area at the time, there was relatively little risk of me stepping on other
people’s toes. The Weld was therefore fairly wide open.
I had a personal preference for a language with some morphosyntactic
complexity. That meant that while I could avoid the small number of
morphologically fairly simple Polynesian languages that are spoken in
the country, there were still many dozens of other relatively complex
languages left to choose from. I decided that if I were going to do a good
job, it would probably be best if I were to choose a location where
I stood a reasonable chance of acquiring some kind of a speaking
ability in the language. I thought that if I were to work on a large island
with many languages, Bislama—the national lingua franca—would
possibly become too much of a ready-made crutch, so I decided to
restrict my search to small islands where there was just a single language.
This reduced the number of choices quite substantially. Of the various
options that this left me with, I ended up choosing the Paamese
language.
Other budding Weld linguists have come to me over the years for advice
about what language they might work on in the same country. People
getting started 61
have all sorts of reasons for wanting to go to particular kinds of places.

People who are planning on doing Weldwork with young children in tow
have expressed a need for a location which is not too far from a hospital.
Others have wanted to be close to an airstrip or an anchorage for ease of
access, and from where it would be easy to get out if there was an urgent
medical need. Some have wanted to go to a location where there might be
other Europeans nearby in case they needed advice about things from
their own cultural perspective. Yet others have wanted to go somewhere
without any other Europeans, presumably to avoid the distraction of
familiarity. Some people feel the need to be near a telephone, or near
roads, or near a regular supply of electricity so they can charge batteries
on equipment that they may have. If you know that you cannot cope
without a regular supply of chocolate biscuits, or you have some other
special needs that can only be satisWed in a reasonably well-equipped
store, you will need to make sure that you choose a Weld site that has
relatively easy access to a store. Of course, some personal preferences are
a bit harder to fulWl than others. I know of one budding Weld linguist, for
example, who wanted to go to a tropical Weld location where there
were no spiders!
While individuals vary enormously in the kinds of situations that they
can be expected to adapt to, single males may be in a better position in
many cases to go to relatively remote locations, while people in families or
single females might Wnd it easier to work somewhere a little less remote.
But, there can be no hard and fast rules here. I know of single females who
have found things perfectly satisfactory in very remote locations, while I
know of single males who have struggled to cope with life in places far less
remote. Personality obviously plays a major role in how well you will be
suited to what kind of Weld site.
From the outset, you should be prepared to accept that you may not
necessarily be the person you think you are, and that you may have a fair
amount of learning to do wherever you end up. I would also oVer the
general advice that once you have made up your mind about where you
want to go, you should make sure that you are fully committed to going
to that place. Being only half-heartedly committed to a Weld site is not
a good way to start. There are enough things that can make Weldwork
more diYcult than you originally anticipated so you need to start from
a reasonable level of enthusiasm.
62 getting started
3.3 Background work
Let us imagine that you have now settled upon the language that
you want to work on. In addition to your general background reading,
you will now need to do as much reading as possible on that particular
language, or its immediate neighbours, as well as any additional reading
on the history of the area, and of the culture of the people. For purely
practical matters, it is worth getting hold of a copy of the relevant local
travel guidebook. These often include remarkably useful details about
transport and accommodation possibilities, as well as pointing out some
of the pitfalls to avoid in local travel.
3.3.1 Reading the literature
It may be that little—or even nothing—has ever been written on the

language and its neighbours, but often there will be at least something.
When you read these materials, you will need to be aware that they may
not be totally reliable. Any linguistic sources written before the 1940s or
1950s in many parts of the world are likely to have been written by people
without any formal training in linguistics, which may mean that
phonemic contrasts have been missed, and particular grammatical
patterns may have been misrepresented. Even sources written more
recently than this can be suspect if the author is an interested amateur
rather than a trained linguist. At the same time, there are plenty of
examples of early sources written by amateurs which have turned out to
represent excellent sources of linguistic data, and you may be lucky
enough to be faced with inspired early sources.
If there are any older published materials written in the language in
question, you may also want to scour these for useful background infor-
mation. For many small languages in parts of the world where people’s
earliest contact with the outside world came with Christian missionaries,
you may be able to Wnd some old translations of parts of the Bible, or
other kinds of religious literature. I would oVer a word of caution about
relying too heavily on such materials, particularly in the initial stages
of your Weldwork. In my experience, many such translations exhibit
grammatical and stylistic oddities that do not correspond closely to
normal spoken usage. While local people may have in the meantime
come to regard these oddities as indicative of high ecclesiastical style,
getting started 63
they often began simply as misanalysed grammatical constructions. If you

want to Wnd out how people speak in ordinary day-to-day situations, it
would probably not be wise for you to arrive with a knowledge of
constructions that remind people more of something like the King
James version of their language.
If these kinds of sources date from the nineteenth century or earlier, or
even from the early twentieth century, there is every likelihood that those
writing the language may have had limited understanding of phonetics. It
must also be remembered that in those times, the concept of the phoneme
had not been clearly or widely enunciated, so even where a writer shows
evidence of good phonetic insight, the resulting written forms may well be
phonemically underdiVerentiated or overdiVerentiated. For example, in
the Erromangan language of Vanuatu, missionaries wrote and translated
materials into the language between the 1860s and the 1920s. Although
there is a phonemic contrast in this language between /k/ and /ª/, the velar
fricative was either written as k, or it was simply left out.
Even published sources compiled by trained linguists sometimes need
to be treated with caution. You can probably assume that an extensive
grammar or dictionary has been based on a substantial amount of contact
with speakers of the language and that it is more likely to be reliable
because of this. A good grammar should say something about the nature
and duration of the Weldwork, and the extent of the recorded textual
corpus, so it is worth checking to see if the author oVers any words along
these lines which you can take as reassuring. I know full well that some of
the ‘conclusions’ that I have reached early in my Weldwork have turned
out to be wrong, and I am immensely grateful that I did not immediately
rush into print. Some linguists, however, have not always been quite
that fortunate, or wise.
The shorter the sketch of a language, the greater the possibility that
certain facts have been either oversimpliWed or even incorrectly analysed.
Despite what I have just said, I have in the past published short sketches
of languages and I have since discovered that some of what I reported is
mistaken. For example, I published a twelve-page sketch of the Neve‘ei
language of Vanuatu (Crowley 2002) which was based exclusively on
elicited data from a single speaker. In that sketch, I wrote that the
language had only singular and plural pronouns. It turned out, in fact,
that there is a separate series of dual pronouns which I had missed
because these forms only appear in pragmatically fairly restricted sorts
64 getting started
of contexts. I am now faced with the need in any subsequent publications

relating to this language to point out my initial error.
Of course, when the only published source for a language is a short
wordlist which contains little or no grammatical information, there are all
sorts of possibilities for forms to sneak past the compiler incorrectly
glossed, phonemically underdiVerentiated or overdiVerentiated, or gram-
matically misanalysed. I know of one such wordlist in which the word
labim appears as a grammatically complex item, consisting of the root
labi- ‘excrement’ followed by the possessive suYx -m ‘your’. It turned out,
in fact, that the word labim cannot be further analysed grammatically,
and that it means ‘area where people go to defecate’ rather than ‘your
excrement’.
Further reading: Austin and Crowley (1995).
3.3.2 The linguist and the lingua franca
Part of your pre-departure preparation may require some language learn-

ing on your part. There are many parts of the world where people in rural
areas speak little or no English. Some other intermediary language may
be needed in order for you to be able to speak comfortably with local
people, and this will almost certainly be the language in which you
conduct your initial elicitation, and possibly even all of your elicitation.
Since you are going to be making some fairly Wne judgements in this
language, your command of it will need to be fairly conWdent.
If you are keen to go to an area where English is not the main local
lingua franca, you will need to put substantial eVort into learning that
language. In some cases, there may be few opportunities to learn this
language before you go into the Weld. While it will generally be possible to
Wnd ways of learning major world languages such as French or Spanish,
acquiring a knowledge of Swahili prior to going on Weldwork in East
Africa may be more diYcult. In some cases, then, you may have to defer
this part of your preparation until you arrive in the country itself.
When I Wrst made my decision to do Weldwork in Vanuatu, I knew that
I would be operating largely through the medium of Bislama, which is the
name by which Melanesian Pidgin is known locally. One linguist had
produced a set of language lessons with accompanying tapes for Tok
Pisin, the variety of the same language that is spoken in nearby
Papua New Guinea, so I began working through those. When I Wrst
getting started 65
arrived in Vanuatu, I planned to spend a month learning how to ‘convert’

my knowledge of Tok Pisin into Bislama. When I Wnally left for my Weld site
of Paama, while I was certainly not conWdent in Bislama, I felt that
I at least had suYcient command of the language to explain the purpose
of my visit.
You have to be prepared for the possibility that things might not be
quite as easy as you think. I also decided that since I was going to
Vanuatu, it might be a good idea to try to resuscitate my knowledge of
high-school French by spending four days in New Caledonia on the way.
As a country lad who still liked my meat well done at the time, if not
lightly burnt, I discovered the hard way that my progress in French had
been much slower than I would have liked. I ordered steak tartare in
a restaurant and was utterly conWdent that the English translation of raw
steak on the back of the menu could not possibly be right. I, however, was
wrong, and they were right: it was clearly going to take more than four
days to beef up my French. And even though I had put such a lot of eVort
into my Bislama, something like twenty years later I was reminiscing with
somebody about my early days on Paama. He commented that people at
the time found it quite diYcult to talk to me because I ‘couldn’t speak
Bislama’. At the time, I thought I was speaking Bislama, but local people
clearly had a somewhat diVerent perception!
3.4 Planning your fieldtrip
If you are embarking on your Wrst Weldtrip, this is likely to be one of the
greatest adventures of your life so far. But you are going to have to
approach this in a very diVerent way from how you might go about
planning for a three-month backpacking holiday in Europe. When you
are going on holiday, you don’t have to have an itinerary and you can
decide on a daily basis what cathedral you are going to visit, whether to
go swimming or hiking, or whether to try out some new restaurants. You
have no overall goal other than to have a great time.
A Weldtrip is fundamentally diVerent in that you have a very diVerent
goal at the end: the documentation of a language. If you are a graduate
student, the dissertation that you produce will probably be about 100,000
words long and it will be the most substantial piece of work you have
produced so far. If you go on in academia, you may eventually conclude
66 getting started
that writing your dissertation was the most demanding thing you ever did.
The consequences of failure may be quite severe if your chosen career
path depends crucially on successfully completing the dissertation and
this will always be lurking in the back of your mind.
What this means is that your Weldtrip needs to be planned. Of course,
you should not expect to be able to plan things down to the minutest
detail, as one of the things that you will hopefully learn out of your
Weldwork experiences is to go with the Xow and to adapt to changing
circumstances. However, you should have some kind of overall plan in
mind. What I want to do now is point to some of the major issues that you
will need to think through before you leave for the Weld.
3.4.1 Fieldwork duration
Perhaps the Wrst decision you will need to make is how long you are
going to need in the Weld. Of course, there is no single answer to
a question such as this, as linguistic projects can vary enormously in
terms of their goals. However, if your intention is to write a dissertation-
length grammar of a language, you should probably aim for a total of
between nine and twelve months in the Weld. You should probably also
aim for this period to be spread between two distinct periods of
Weldwork with an intermediate period of analysis and preliminary
write-up at home.
I would recommend that your Wrst visit be a longer one and that your
second visit be somewhat shorter. If you are planning on twelve months
in the Weld, this means that your Wrst visit may be of nine months’
duration while your second visit would be of three months. The point
of returning home after a substantial initial period in the Weld is so that
you can properly organize your data and so that you can come up with
a provisional analysis of what you have already done. It is likely that
during this writing up, you will become aware of certain gaps in your
data which you were not aware of when you were in the Weld. This is
because you will be in a position to bounce ideas oV your colleagues, or
your supervisor, and you will have access to a library. You will also have
more time to sit and think without interruption. All of this would be
diYcult—or impossible—if you had stayed in the Weld. Then, when you
return to the Weld for your Wnal period of data-gathering, you can
concentrate on checking the accuracy of material about which you may
getting started 67
have developed some doubts during your preliminary write-up, and you
can Wll the gaps that you have noted.
Having an initial trip with a follow-up trip should be considered
essential planning. I have examined one doctoral dissertation which was
based on only a single period of Weldwork and the Wnal work that was
submitted was replete with statements such as ‘Further work is needed to
investigate this phenomenon’ or ‘It is not possible to decide between
these two solutions without access to further data.’ Such words are
anathema to Ph.D. examiners, as well as to pre-publication reviewers of
book manuscripts.
Exactly how you should break up your period in the Weld is going to
depend to some extent on your personal circumstances. If you have
personal commitments that allow for only an initial Weldtrip of six
months, then by all means plan for two equal visits of six months each.
You should also be prepared to make changes in your plans as your
Weldwork progresses. My original plans for Weldwork on Paama in
Vanuatu were for an initial nine-month Weldtrip with a shorter
three-month follow-up trip for checking and Wlling gaps. However,
while I had done Weldwork before in Australia, my Weldtrips were always
much shorter and more frequent, and I also had access to private spaces
where I could work comfortably with electric lighting. On Paama, I had
no table. (People eat on mats on the Xoor.) There was also no electricity.
(People use kerosene lamps and torches at night.) There was also little
opportunity for privacy. (People would typically want to come and chat if
they saw me working by myself.)
While I was accumulating lots of textual data and I had done a lot of
direct elicitation, I started to feel that I was Xoundering after about six
months and I had little idea of the directions in which further elicitation
should go. My lexical data was a mess because I had no opportunity to
organize things properly. I therefore made the decision to jump on a ship
back to the capital and then to return to Australia to do some serious
writing up. I remember that my return three months early annoyed the
head of my department at the time, but I placated him with a promise to
spend six months on my second Weldtrip instead of the three months that
was originally planned. That’s what I did, and when I went back with
properly organized lexical material and some clearly set-out provisional
analysis, I was back in the swing of things for much more productive
68 getting started
elicitation. My second trip ended up being so much more worthwhile and

so much more enjoyable than the Wrst.
3.4.2 Planning for contingencies
You are going to have to plan for a variety of contingencies in the

Weld. Some level of self-suYciency may be necessary in health matters,
depending on where you are going. Malaria is an unpleasant disease
which can certainly leave you debilitated for a few weeks, and it can
sometimes be worse than simply debilitating. However, it is largely
preventable. You need to visit your doctor before you leave for
a prescription for malarial prophylaxis. What you will Wnd is that there
is a range of options and you should discuss these with your doctor. Some
treatments have nasty side eVects with small numbers of people. It would
be a good idea to establish which medication is most suitable for you, as it
would not be a good idea if you were to take a medication-induced
psychotic turn in the Weld. This happened to me once, so I threw the
tablets down the toilet . . . and then came down with malaria as a result.
Obviously, malaria is not the only potential health problem. Depending
on where you are going, vaccinations for a number of diVerent communic-
able diseases may be required, or at least recommended. You should
therefore inform your doctor where you are going and talk things over
with him or her well before you leave. There are parts of the world
where HIV/AIDS is a major public health issue, so a high level of personal
responsibility can also reduce the risk of coming down with health problems.
Even minor health problems need to be catered for when you are in the
Weld. I never leave for the Weld without a supply of antibiotic powder or
cream, bandaids, and antifungal medication. A simple scratch or
a mosquito bite can easily become infected in the tropics and this can
ultimately develop into a serious ulcer. Be sure to take more medication
than you think you might possibly need for yourself, as you will quite
possibly end up tending to local people’s minor cuts and scrapes as well.
If you are the sort of person who happens to come down with whatever
minor ailment is going around whenever you get run down, then perhaps
it would also be a good idea to pack a supply of nutritional supplements.
I also Wnd that if I am doing a lot of walking or other kind of physical
activity, small sachets of rehydration salts can be a life-saver.
Further reading: Samarin (1967: 71–4).

getting started 69
3.5 Funding
Linguistic Weldwork is seldom cheap, and unless you are a person of

independent means, you will almost certainly need some kind of funding
support to back you up in your endeavours. If you already have an
academic position, your salary costs will be taken care of, but if you are
a graduate student, you will need some kind of stipend, and possibly also
support for the payment of your fees. Obviously, opportunities will vary
from country to country, though sometimes funding agencies do allow for
grant applications to cover costs such as these.
Fieldwork is also going to involve direct travel-related costs. This will
include your fares to and from your Weld site, which may involve an
expensive international ticket as well as additional fares from the capital
to the local community where you are going to be working. Travel
insurance is crucial, because you never know what might go wrong
medically to require you to return to the capital, or perhaps even to
your home country for treatment. You are going to need reliable record-
ing equipment. You are also likely to be faced with fees for the issuing of
visas and research permits. You will need to have some money to cover
your living expenses in your host community, as store-bought goods in
remote locations can sometimes be fearsomely expensive. Finally, you
will need to have funds with which you can pay your ‘informants’.
It is obviously impossible to say precisely what kind of amounts you
should budget for, as this is going to depend on the speciWc situation.
I know of some linguists who have done all of their linguistic research
while staying in a hotel or a local guesthouse, while others have had no
option but to stay within the community in local-style housing. Obviously
there can be a substantial diVerence in costs involved. Of course, those
who choose to stay in a hotel when there is a possibility of local
accommodation within the community are likely to miss important
opportunities for observing natural linguistic interaction, and this is
a factor to be considered alongside that of cost.
Once you have worked out your budget, you will need to seek funding.
You may be lucky enough to be attached to a university where there are
internal mechanisms for funding this kind of research. Increasingly,
though, funding these days needs to come from external agencies.
There are some bodies which call for applications speciWcally to support
70 getting started
research leading to basic linguistic documentation of the kind that I am

talking about in this volume, including the following:
. The Hans Rausing Endangered Languages Project (www.eldp.

soas.ac.uk/hrelp_home.htm);
. The Foundation for Endangered Languages (www.ogmios.org/
home.htm);
. The Endangerd Languages Fund (www.ling.yale.edu/elf );
. The Volkswagenstiftung (www.volkswagen-stiftung.de).
None of these funding agencies is in a position to fund all projects that are
submitted, so they typically look for evidence that a project is well
thought out logistically, that it has community support, that it shows
promise of producing results that will directly beneWt the local commu-
nity in some way, and, of course, that it is of substantial academic merit.
In addition to these funds which are speciWcally geared towards
supporting linguistic research, there are other funding agencies which call
for academically sound applications from a broad range of disciplines
conducted by scholars in your country. Since you may be competing with
other projects to investigate evolving attitudes to mental illness in
sixteenth-century France, the deconstruction of the writings of postmodern
writers in Brunei, and the genetic modiWcation of newts in Britain, it is
much more diYcult to give speciWc advice on how to ‘sell’ a project
involving linguistic documentation. However, such funds typically stress
academic excellence and the proven research record of the participating
researchers, so it is important to pay close attention to these kinds of issues.
3.6 Permits
Depending on where it is that you intend to carry out your Weldwork,

there will be a range of bureaucratic procedures that you will probably
have to deal with. Even if there should happen to be no bureaucratic
requirements relating to access to a particular Weldwork project, there is
an overriding ethical requirement that the general aims of the project
should be understood by the community you will be working with, as well
as accepted by people in that community. A linguist should therefore not
just turn up on the beach or at the airstrip and expect immediately to
move into data-gathering mode.
getting started 71
Quite apart from normal passport and visa requirements, there may
well be a speciWc procedure to follow in some countries if a researcher is
planning to conduct linguistic research. A researcher may need to be
issued with a special research visa, which has conditions—and possibly
also fees—attached to it, which distinguish it from an ordinary tourist
visa. While a tourist visa will often be granted automatically for a short
stay on arrival at an international airport, a research visit may require
a longer stay, and immigration oYcers may not be authorized to issue
those sorts of permits on arrival at the airport.
You also need to be aware that your proposed site may turn out to be
one where national authorities do not welcome the prying eyes of
outsiders, in which case you may never be granted a visa. I understand
that it is very diYcult for outsiders to gain permission from Indian
authorities to visit parts of the Andaman Islands, so unless you have
very good local contacts and very good reasons for wanting to go there, it
is perhaps not worth trying. I had originally intended carrying out my
own doctoral research on a particular island in Indonesia in 1976, but my
application for a permit was met with prolonged silence from the
Indonesian end to the point where I was losing valuable scholarship
time. I then made my own decision to switch to Paama in Vanuatu as
an alternative Weld site. This was just as well, really, as it turned out that
the island in Indonesia that I had chosen to go to, unbeknown to me,
hosted a facility in which political prisoners at the time were rumoured to
be detained. That meant that I was possibly never going to get that
visa anyway.
Even if you are planning on doing research in your own country where
the question of visas does not arise, it is likely that you will need to seek
some kind of formal approval from the local community where you are
planning on conducting research. The nature of such local authorities, as
well as their powers, will no doubt vary from place to place, but it will
be your job to Wnd out what the proper procedures are, and to follow
those procedures. Simply arriving on the doorstep will probably not
ensure a warm welcome if a local community has already given consid-
erable thought to establishing a set of procedures to deal with the
involvement of people such as yourself within the community.
Research permits may need to be sorted out before arrival in the
country concerned, either through the country’s diplomatic mission in
your own country, or directly through the relevant agencies in the country
72 getting started
itself. Researchers who simply arrive in a country expecting to be granted

permission to carry out Weldwork may sometimes Wnd that they are
required to leave the country again and apply for permission from out-
side. There will normally be some local agency which has the primary
responsibility for initiating the issue of a research permit. This may be the
relevant National Museum, the Department of Immigration, or the
Department of Culture. It is always best to have some kind of local
contact in order that you know the correct place to address your appli-
cation, as well as the appropriate format. If you are unable to work out
the appropriate procedure, you may be able to get valuable help from
linguists in a local university. Fortunately, in this era of internet searches,
Wnding appropriate contacts is much easier than it once was.
Granting of permission to conduct research may be associated with the
payment of some kind of fee to the local institution. Local institutions
involved in culture and language are often not well supplied with funds
from their national governments, so this kind of contribution from visiting
foreign researchers helps in a small way in the running of the museum. It
also makes perfect sense for national museums to coordinate research
in this way, in part to ensure that there is complementarity between
researchers. It would not be wise for several linguists to descend on
a single language community when there may be other languages nearby
that have been completely neglected. Channelling research in this way also
ensures that there is not a sudden inXux into a single community at the
same time of a linguist, an anthropologist, a botanist, and a seismologist.
In order to reduce any possible negative impact of foreign visitors,
research should again be distributed as widely as possible.
Sometimes, the procedures for gaining permission to carry out research
in a community may be unclear, and the Weldworker can easily be caught
in the middle of competing powers. In the 1960s in Queensland, the
European ‘manager’ of an Aboriginal reserve had absolute power to
decide who could enter and stay on the reserve and who could not, and
he (it was never a she, as far as I know) did not have to consult with
members of the local community. By the early 1970s, local Aboriginal
communities were beginning to establish their own procedures for who
could enter reserves, but there was something of a power struggle between
the old managers and the new community councils.
When I Wrst arrived in far north Queensland to carry out my research
on the Uradhi and Mpakwithi languages, I had made arrangements with
getting started 73
the chairperson of the local community council, who was agreeable to my

research project. I duly arrived to start my research and just after
breakfast on my Wrst morning there, I received a message requiring me to
appear immediately at the manager’s oYce. I realized that I had been caught
up in a nasty mess and I was trembling by the time I arrived at the manager’s
oYce. I somehow managed to convince him that since I had the approval of
the community council, I should be allowed to stay, but it was quite clear
to me that he was not happy about this. I am convinced that he would have
put me on the Wrst plane out of there if he could have managed it.
In some parts of the world, the granting of oYcial permits, the payment
of bribes, and arbitrary vindictiveness seem to go more or less hand in
hand. Fortunately, my own rather more positive experiences leave me
completely unqualiWed to oVer any speciWc advice to any Weldworker who
faces these kinds of administrative norms. As advance warning for some,
and as entertainment for others, I simply recommend that you read
Barley’s (1983) The Innocent Anthropologist: Notes from a Mud Hut.
3.7 Equipment and supplies
While most of my Weldwork recordings were on reel-to-reel or cassette

tape, new recording technologies allow you to record high-quality digital
audio and video and then work with it directly on your computer.
3.7.1 Recording gear
Basically, the only equipment that an ordinary project will need in the
Weld is a reliable recorder and high-quality microphone, as well as an
ample supply of batteries. Some projects are more sophisticated than this,
calling for video recording of interpersonal communication or public
speeches to capture visual aspects of communication in conjunction
with ordinary grammatical patterns. And having now tried the luxury
of a laptop in the Weld, I can see why many other linguists might want to
do the same.
If you are going to rely on batteries, you will need to make sure that
you have a good supply of the best-quality batteries. If batteries are
available locally in the Weld, I have found that they are often the cheaper
batteries which drain more quickly, because local people are not able to
74 getting started
aVord the more expensive long-life batteries. In any case, if you need
a special size of battery, you may not be able to get what you need at all.
This means that you may need to estimate your battery needs for several
months and bring them with you. Of course, batteries are heavy and you
may Wnd yourself paying quite a surcharge if you have to pay excess
baggage to bring them with you on the plane.
The more equipment you have, the more likely it is that you will need
regular access to electricity. You may be lucky enough to be working in
an area where there is 24-hour electricity. Of course, electricity supplies in
some parts of the world operate on 110 volts while other parts of the
world operate on 220 (or 240) volts. If your equipment is not compatible
with both, you may need to carry a voltage adaptor. DiVerent countries
also operate with diVerent kinds of plugs and if you do not have a proper
converter, then your equipment may end up being totally useless in the
Weld where adaptors may well not be available locally.
In some Weld sites, if you are going to be dependent on an electricity
supply, you may need to acquire your own generator. The average
city-dwelling linguist has probably not had much contact with generators,
but even the most portable of portable generators is not really very
portable. They are, in fact, quite bulky and heavy enough that it takes
at least two people to carry one. So, with all of this equipment, you might
even need a vehicle (or a boat) to transport it all. These days, adventurers
and travellers are increasingly making use of portable solar panels to
charge the batteries of their equipment, and Weld linguists can make use of
this kind of technology as well.
The price of all of this equipment is starting to mount well beyond what
will be available for the average graduate student. It also means that a simple
trek across the island with your recording equipment in your backpack may
well be out of the question given the amount of additional gear that you
would have to bring with you to keep everything operating. What is most
practical and what is ideal often do not go hand in hand in Weld research.
Let’s go back to the audio recorder. Until fairly recently, any advice
would have revolved around what sort of cassette, recorder and cassette
tape to use. Many people still record on cassettes, though the information
that they contain is increasingly being digitized when it is deposited for
storage in professionally curated linguistic archives. Cassette recorders
are fast becoming obsolete and digital technology has advanced to the
point where it is preferable to record material digitally in the Wrst place.
getting started 75
There are several ways of recording digitally, but the preferred option at
the moment is to record with Xashram recorders.
Some linguists also record on minidisc recorders. While the quality of
sound reproduction of minidiscs is Wne for transcribing and listening to
texts, the proprietary playback mechanism, as well as the process of
transfer from minidisc to your computer, results in the loss of signiWcant
parts of the sound signal. This renders the material less useful for any-
body who might want to conduct instrumental analysis of your data in
the future. There are also questions about the archivability of the propri-
etary formats used by minidiscs, so they need to be converted to wav Wles
as soon as practicable. Best Weldwork practice at the moment is to record
with Xashram recorders in preference to minidiscs.
If you are going to be transcribing from a cassette recorder, you will
need to make sure that the equipment you have brought with you allows
for easy and rapid rewind, and that you are able to rewind short stretches
of text at a time. It can be very frustrating for both you and your
language-helper if you hit the rewind button and it takes so long for the
mechanism to react that you end up hearing all over again material that
you have already dealt with. The machine also needs to be quite sturdy, as
you are likely to be hitting that rewind button thousands of times
while you are in the Weld. A digital recorder will come equipped with
some kind of digital equivalent of a rewind button. It would be a good
idea to practise using this before you go to the Weld so that you can work
out how to reverse to the place that you want to get to.
Your audio recorder may need to be small enough that you can carry it
in your pack over sometimes rough terrain. It will need to be sturdy. I have
on one occasion had to dive into the ocean to rescue a tape recorder after
the outriggers on the canoe broke and the tape recorder and I went into the
drink together. (Fortunately, the tape recorder was in a waterproof bag,
and I was able to rescue it after only a few seconds on the bottom before the
salt water got to it.) You can also expect to get rained on from time to time,
and tropical rain tends not to be the gentle drizzle of many temperate
western cities. Then, there is the possibility of cyclones—also known as
typhoons or hurricanes, depending where you live—which can cause
considerable water damage to equipment, especially in thatch houses that
are built to allow for a cooling through-Xow of air in a tropical climate.
While many recorders come with an in-built microphone, you should
not use it as it will also record machine noise, and it is typically not of
76 getting started
suYcient quality for Weldwork recording. You will get far better quality
recordings if you use a plug-in external microphone. Microphones come
in all sorts of shapes, sizes, and capabilities, depending on what sort of
recording you are going to be doing, and you can Wnd good advice on
websites linked from your favourite linguistic archive.
One option would be a clip-on lapel microphone, which has the
advantage of being very discreet. However, this would be useful if only
one person were speaking. Even when I am recording monologues, I
always like to introduce my speaker, as well as giving the date and place
in which the recording takes place, and it would be an awkward way to
start a recording session if I were to get things rolling by speaking into
somebody else’s sternum. A Xat microphone that sits directly on the table
can be fairly discreet, while also picking up voices in more than one
direction, though this kind of microphone has a nasty habit of picking
up all sorts of noises of people Wdgeting and tapping on the table unless
you remember to lay it down on a piece of folded cloth. Finally, there is
the traditional hand-held microphone, which has the disadvantage of
making the Weldworker look like a journalist in a disaster scene, though
you can hold the microphone in the ideal position for the best-quality
recording, or put it in a stand that will soon be ignored by the speaker.
Further reading: Austin (2006).
3.7.2 Other needs
You will require a supply of stationery. This means that you will have to
anticipate how much notepaper you will need, as well as how many pens
you are likely to use. You will also need to anticipate expectations from
the community that your ample supply of paper and pens may be publicly
fairly available to allow people to write letters, as it may often be diYcult
for people to get hold of such items. Remember, though, that while
stationery is not expensive to buy, it can be heavy, and if you have to
transport it by air to your Weld site, you can easily use up most of your free
airline baggage allowance, leaving you facing a bill for excess baggage
that you had not budgeted for.
You need to remember that you may not always have a table to work
on. A clipboard is therefore a very useful item to bring with you to ensure
that you have a solid surface to write on. I would recommend that you
buy bound notebooks rather than loose-leaf binders or writing tablets
getting started 77
with tear-out pages. I have tried both of these options in the Weld in the past
and neither was very satisfactory. With loose-leaf binders, the holes on the
pages often tear through and your pages then become free-Xoating entities
which can either end up in the wrong place, or they can get lost altogether.
The same applies to tear-out pages from writing tablets. Such pages
are good for writing letters or for scribbled notes, but they are not good
for any information which you want to keep as part of a permanent record.
A Weld researcher needs to anticipate any possible special food needs.
Local food may be nutritious but bland, or it may be overly spicy, and
sometimes unfamiliar tastes and textures may be diYcult for the newly
arrived Weldworker to cope with at Wrst. Depending on the personal
circumstances of the researcher, it may or may not be easy to prepare
your own food in your Weld site. If you cannot prepare your own food and
you have an urge for familiar tastes, textures, and species, it is best to
bring some items with you. Some of your favourite items may be expen-
sive, and they may have to be shared communally. (However, I have on
occasion been known to hide tins of baked beans in my suitcase and eat
them when nobody else is around.) When the food is bland, your personal
supply of chili sauce, soy sauce, or tomato sauce may help.
Taking care of personal preferences can extend beyond your dietary
requirements. For example, if you have a resistance to the idea of using
coconut husks or leaves for sanitary purposes, you will need to bring in an
adequate supply of toilet paper, and you may need to protect it Wercely.
(I have in the past kept mine hidden under my bed until the precise
moment of intended use.) In a Weld site where there is no electricity, you
will need your own torch (or Xashlight), as well as a kerosene lamp. And
a kerosene lamp is not much use without matches.
If you know you are going to have to do some trekking, be sure to
invest in some properly Wtting hiking boots. One pair of boots that
I took into the Weld allowed too much forward movement of my foot
on downhill sections. When I arrived at my destination and took my
socks oV, one of my big toenails more or less came oV with the sock, and
the other one came oV by itself a few days later. A comfortable pack is
also a good investment. If you are heading for somewhere tropical,
a waterproof pack is a wise investment. Even if you don’t get caught in
a tropical downpour, you would be surprised how much sweat can run oV
your back, into the contents of your pack, and right through all of your
once fresh-smelling clothes.
78 getting started
Apart from your major equipment items, given what I have said so far
in this chapter, you should therefore consider including the following
items on your shopping list in preparation for leaving for your Weld
location:
. good hiking boots and pack

. kerosene lamp and torch (or Xashlight)
. special food treats
. clipboard
. pens
. one or two pocket-sized notebooks
. good supply of hard-cover bound notebooks
. batteries for torch and recording equipment
. tapes for recording
. regular prescription medication
. dietary supplements, rehydration salts
. injections and prophylactic medication
. antibiotic and antifungal medication, bandaids
3.8 Getting about in the field
The kinds of locations where Weld research takes place are typically fairly
remote. Arrival may be by small plane, coastal vessel, speedboat, canoe,
or on the back of a truck. Any trip that is more than a short one is likely
to involve a degree of discomfort, and some walking may be necessary to
reach your Wnal destination. For instance, when I was doing Weldwork on
Erromango, I faced a forty-Wve-minute Xight from the capital. I was then
faced with the possibility of a fairly expensive three-hour speedboat trip
if the sea was suYciently calm. If the sea was rough, or if the local
boat-owner ran out of fuel (which often happened), then I was faced
with a 22-kilometre walk. The Wrst 11 kilometres followed a logging track,
though with the suspension of logging, this had started to revert to bush.
The initial stage in the process of regeneration to bush involved
a dense growth of Mimosa pudica, a treacherous grass with razor sharp
backwards-pointing thorns that leave your legs from ankle to thigh
looking like you’ve just spent a bad night on a bed of nails. (The local
name for Mimosa pudica is gras nil ‘nail grass’.)
getting started 79
After that, the vehicular road gave out completely and the remaining 11
kilometres involved a single-Wle bush track that crossed a dozen or so
streams, including two major rivers, all without bridges. There was some
steep—and often slippery—climbing, and all of this is done while carrying
a pack. A Weld linguist has to be either Wt before arrival, or he or she very
rapidly becomes Wt in this kind of situation. Failure to cope with these
conditions can be unpleasant. I heard of one American Peace Corps
volunteer on Erromango who simply gave up in the middle of a trek.
He collapsed in tears on the path and in the end he had to be returned
stateside. In fact, I remember once in a state of exhaustion starting to
think very seriously about doing the same myself. I just wanted to curl up
beneath an overhanging rock and sleep on the ground to see if I had the
strength—or the will-power—to continue the next morning.
There may be other travel options. If you are going to be based on
a river or by the sea, it pays to learn how to row a canoe if that is one of
the local means of transport. This can be not only convenient, but also
relaxing for short journeys. Of course, you may lack advanced rowing
skills and the consequences can be dire. I have already mentioned my
misfortune in a canoe when an outrigger snapped oV, causing the canoe
to capsize. What I failed to mention at that point was that the accident
was my fault: I had failed to align the canoe properly according to the
direction of the swell in a turbulent part of the sea.
The potential unreliability of transport has to be taken as a major
factor in planning one’s Weldwork. For instance, if your only access in
and out of a particular location is by scheduled twice-weekly Xights or by
a weekly ship, you cannot always assume that the plane or ship will arrive
as scheduled. Anything at all could happen to leave you stranded for
longer than you had hoped. It might rain heavily and the grass airstrip
might take a couple of days to dry. The grass on the airstrip might not
have been cut by the local community so the pilot cannot land. The
passage of a cyclone can close down an entire air or sea transportation
grid for days. The plane or ship might break down with no replacement
possible. The plane or ship might be diverted at the last minute to deal
with a medical emergency somewhere else. A local politician might use his
or her inXuence to divert a plane or ship from its proper schedule to pick
him or her up. A Weldworker should, therefore, never rely on tight travel
connections, especially if you are dealing with your international Xight
out of the country. Always allow suYcient spare time in transit.
80 getting started
Of course, being delayed does not have to be an unmitigated disaster.

On a short non-Weldwork-related visit to the island of Malakula in 1988,
just as I was ready to Xy back to the capital to return to work, the heavens
opened up and the airstrip turned to a mudpit. A local man suddenly saw
an advantage in my distress. He took me by the hand and said, ‘I am one
of the last speakers of the Nāti language. Now that you are stuck here, we
will go to the local store and get some pens and some exercise books and
we will spend the time documenting my language.’ And that is what we
did for several days till the sun dried oV the airstrip and the planes started
Xying again (Crowley 1998a).
3.9 First contact
Your arrival in your chosen Weld site is likely to be the scariest part of
your entire Weldwork. There are some things you can do to make this
easier.
3.9.1 Arriving with an invitation
When I was talking earlier about choosing a language, I did not say that
the best way of all to choose a language is for the language community to
choose you. The luckiest Weld linguist is somebody who is invited into
a community (e.g. Hale 2001). Sometimes, the project design may be
predetermined by the community, in which case it is up to the linguist
to decide whether he or she can match the community’s expectations with
his or her own. In other cases, a community may arrange for a linguist to
come in and do certain things that the community wants done, with the
quid pro quo that the linguist will be allowed to write a grammar for his or
her doctorate.
Such invitations to participate in community-run projects can arise in
a variety of ways. In many parts of Australia, for example, there are
community-run ‘language centres’ which have been set up as archives
for linguistic data and for the storage of oral tradition and cultural
knowledge. Language centres like these often publicly advertise for lin-
guists to come and service their needs. In other cases, a community may
be keen to develop literacy materials for young children in conjunction
with moves at the national level towards greater use of local languages in
getting started 81
the formal education system and they may seek the help of a linguist.
A community may also be keen to see the Bible or some other important
religious document translated into their language. The Summer Institute
of Linguistics (www.sil.org) has an international network of trained
linguists who respond to such community requests in many diVerent
parts of the world.
Even if you do not have a formal invitation to come to work in
a community, it is always best to be introduced into the community either
by somebody who is a member of the community, or by somebody who is
at least well known to local people. My work on the Erromangan
language began more or less by accident. I accepted an invitation from
an Erromangan friend to come and spend Christmas with him and his
family in his village in 1993. It was such a wonderful place, and people
were so overwhelmingly welcoming, that my friend suggested that since
I was a linguist, wouldn’t it be a good idea to come and spend more time
in his village so I could document his language. This hadn’t entered my
mind during that initial visit, but being invited into the community in this
way seemed like an opportunity that was too good to miss. So, I Xoated
this idea past other people in the community. They responded enthusias-
tically, and I subsequently spent just over a year living on that island, and
produced both a grammar and a dictionary of the language, as well as
a collection of oral tradition.
3.9.2 Arriving without an invitation
Many intending Weld linguists don’t have the advantage that I had on
Erromango of already knowing people who can introduce you into
a community. This is most likely to be the case with graduate students
who may be just starting out on their Weldwork careers. In fact, this was
the situation that I faced when I had Wrst made my decision to work on
the Paamese language. I shudder when I think back to how I started, but
will contrast my arrival on Erromango with my arrival on Paama in order
to convince people that this is not the best way to start.
I arrived in the country unannounced, knowing not a single soul. The
Wrst thing I did was present myself to the local authorities for the neces-
sary permits. This did not seem to be a major problem, as nobody really
seemed to have given much thought to the need for any kind of procedure
for dealing with this kind of visitor at the time. I then presented myself to
82 getting started
the local museum, where an overseas linguist was in charge. He put me in

touch with some Paamese people living in town and I spent a few weeks
talking to people on a daily basis to record some basic linguistic
information, and also to learn something about the island itself.
At that time, there were no phones on Paama, nor was there an airstrip.
This meant that it was not going to be possible to forewarn anybody of
my impending visit in time for my planned arrival. The only regular way
to get to the island at the time was by ship. The use of that term suggests
a grandness that is perhaps a little inappropriate, as the MV Frances was
really just a largish boat that plied the open PaciWc Ocean. The one-day
journey to Paama was hugely uncomfortable. The only place to lie down
was on a canvas sheet that was stretched out over the cover of the cargo
hold. I was seasick. I suVered additional discomfort because I didn’t
realize that there was a toilet on board, and I was too embarrassed to
ask somebody. (In any case, on subsequent trips, it became apparent that
holding it in wasn’t necessarily the worst option.)
When the vessel Wnally arrived at Paama and I staggered onto the
beach, I do not remember exactly what my Wrst words were to the people
who I met, but I suspect it must have been something like ‘Take me to
your chief’. I do remember that I managed to get myself ensconced in
a guesthouse for visitors and then I set oV on foot to visit the chiefs of
each one of the twenty separate villages on the island to explain the
purpose of my visit. Although my visit was completely unexpected,
I remember thinking that people were remarkably welcoming.
Despite not having arrived under what I would now consider the proper
circumstances, I thought that I had probably made a reasonable sort of
Wrst impression.
3.9.3 Local reactions to the Weldworker
I also found out about twenty years later that some people thought that
when I Wrst arrived on Paama, I was a lunatic. I eventually observed that
local people were very tolerant of people with psychological disturbances
as long as they were harmless, which was probably just as well. My advice
now would be: if it is at all possible, arrange to be introduced to
a community by somebody who is known in the community.
Although my arrival on Paama must have seemed rather bizarre to
local people, I am sure that those few weeks that I had spent in the capital
getting started 83
carrying out some preliminary Weldwork certainly helped convince people

that I was not totally mad. There is a lot to be said for getting some kind
of introduction to the language before you go to your chosen Weld site.
Sometimes, even being able to say ‘hello’, ‘goodbye’, and ‘where is the
toilet?’ as soon as you arrive is going to impress people in the community
in a way that will help make people a little more relaxed about your
presence.
When I arrived on Paama, I was able to demonstrate that I knew a little
of the language already. People were keen to see how far my knowledge
went. At one point, a rooster sauntered past and somebody asked me
‘Do you know what that’s called?’ People waited to see if I knew the word
ato ‘chicken’. In fact, all I had recorded at that stage were the words
t
otoman ‘rooster’ and toletau ‘hen’, so I replied with the more speciWc
term. Evidently such a detailed and speciWc response had been completely
unexpected and people quite literally screamed with what I took to be
either amazement or delight. I had quite by accident shown myself to
be a good language learner, and this no doubt helped to get me started.
I may have been mad, but I wasn’t totally stupid.
However, the Weldworker must be prepared for the possibility that the
project that he or she sees as being thoroughly worthy may not be so
positively viewed from within the community itself. In addition to any
oYcial permits that may be needed for your research, you will also need to
gain community acceptance of your project. In some cases, a local commu-
nity may reject a project simply because it has misunderstood the intent of
the researcher. It is also possible for a project to become a convenient focal
point for the expression of local divisions, which can lead to its rejection.
For instance, when I proposed to go to Erromango to conduct research
in 1994, my request was initially rejected by the local council of chiefs.
The local sponsor of my project was not somebody who I chose, as this
aspect of the approval procedure was automatically handled by the
National Museum. It turned out that the person who proposed and
supported my project on the island was perceived as being sympathetic
to a political party that was disfavoured by some of the chiefs at the time.
In addition, that person came from a part of the island that had for a long
time resisted the granting of permission to Malaysian logging companies
to conduct their operations on land belonging to people in his area,
whereas some of the chiefs on the council were sympathetic to the
granting of logging permits.
84 getting started
Even though I had no idea that this sort of thing was going on, and the
chiefs in question had never met me, there was perhaps a suspicion that
I might be used as some kind of weapon against Malaysian logging
companies. Although my research project had no relevance to either
side of the political divide or to the granting of logging permits, it was
initially turned down because it provided a convenient way of ‘getting at’
somebody else on the basis of purely local concerns. It turned out in the
end that my own sympathies on this logging issue were very much in line
with those on the island who were opposed to the Malaysian logging
company. However, despite this sympathy, I was extremely reluctant for
a long time to voice any opinion either way on the logging issue because
of the potentially divisive nature of this issue, its lack of direct relevance
to my linguistic research, and the possibility that some chiefs might call
for my expulsion.
In some places, there is resistance to the idea of Weld research because
communities—and sometimes even individuals—feel as if they have
already been over-researched. There is a common joke that among
a certain group of people, the nuclear family consists of a husband and
wife, two children, and a linguist or an anthropologist. If there has been
a history of researchers passing through a community in a continuous
stream, it is understandable that they may be reluctant to welcome yet
another visitor. In some cases, you may simply have to accept that the
proposal to which you have given so much thought is not going to go
ahead for this kind of reason. You may need to learn to be humble and to
Wnd another Weld site where you are welcome. There is little point going
somewhere where you are not welcome.

4
Gathering your Data
One of the most commonly asked questions that non-linguists—and

sometimes even linguists of the armchair variety—have asked me about
the conduct of Weldwork is: How do you actually start when you are going
to describe the phonology, the morphology, and the syntax of a language?
That’s a perfectly reasonable question, of course, so let’s get on with it.
Further reading: Samarin (1967: 45–74), Austin (2006), Bouquiaux and Thomas
(1992: 27–38), Vaux and Cooper (1999: 21–2).
4.1 Choosing language-helpers
As soon as you are ready to get started, you are going to need people’s
help to learn the language. You may arrive with the idea that anybody at
all can perform this kind of role, as long as he or she can speak the
language in question. Of course, things can be rather more complicated
than this. A Weld guide such as this therefore needs to provide some kind
of advice in the selection of an ‘informant’.
4.1.1 ‘Informants’
But before we consider the issue of who is going to help you to work on
the language, I want to talk brieXy about terminology. In the past,
Weldworkers often referred to their ‘informants’ in the Weld. While some
Weldworkers still use this term, many people feel uncomfortable referring
to people in this way. This is because an ‘informant’ is often taken to be
somebody who tells us things that are not meant to become public
knowledge, rather like an ‘informer’.
86 gathering your data
The problem is to Wnd another word that clearly indicates the role
that our ‘informant’ plays without expressing this kind of negative
connotation. Some people use the term ‘consultant’, though this has its
own negative connotations in many parts of the developing world where
self-appointed experts are often contracted on highly paid short-term
‘consultancies’ to write reports that show little real awareness of the
situation on the ground. Some people use the term ‘Weld assistant’, or
‘language teacher’. Depending on how actively involved in the Wnal
materials this person is, it may be legitimate to refer to him or her as
also your ‘colleague’ or your ‘collaborator’. In the heading for this
section, I have chosen the compound ‘language-helper’.
All of these terms may sometimes sound a little forced, or they may not
quite match the kind of working relationship that develops. There are
therefore times when I try to sidestep these issues of stylistic awkwardness
by simply referring to ‘the person who I am working with’, ‘the person
who is helping me with my analysis’, or even ‘somebody who provides
native-speaker judgements’ if any of these expressions happens to sound
better in any given context.
Further reading: Newman and RatliV (2001: 2–4), Rice (to appear), Bouquiaux
and Thomas (1992: 31), Vaux and Cooper (1999: 10).
4.1.2 Language-helpers at home
You may be lucky enough to be in a position to do some work on the

language that you are interested in with somebody at home before you
even leave for the Weld. This person may be living in your city for a variety
of reasons. He or she may be there as a university student, as a recent
immigrant, as a refugee, or as a short-term visitor. One thing that you can
probably be certain of is that he or she did not come to your city in order
to become somebody’s language-helper. An approach from you for help
in this kind of matter will almost certainly come as a surprise, and people
may not always know quite how to respond.
I remember once approaching a speaker of Nauruan who was living in
Canberra when I was still a student there to see if she would be interested
in working with me and a small group of other linguists on her language.
We were particularly excited by this prospect because Nauruan was a
poorly described language which appeared to exhibit a number of
unusual features for a member of the Oceanic subgroup of Austronesian
gathering your data 87
languages. I visited this person at her home and explained the situation.
She seemed surprised but was willing to participate, so we arranged a time
to meet to begin recording sessions. Just as I was leaving and the door
closed, I heard very loud shrieks of laughter emanating from inside the
house as she told her room-mate what I had come for.
You may well be dealing with a well-educated member of the local
community who is in your country to undertake graduate study. If you
approach such a person to help you, that person may be reluctant, saying
that he or she does not speak the language well enough. This may be
based on a realistic assessment of that person’s competence, as many
people who are well educated enough to get to university overseas
have often spent a signiWcant part of their life away from their home
communities. If they grew up in town rather than in a rural area, they may
not speak their parents’ language well at all, or they may speak it with
reduced conWdence. People who harbour these kinds of feelings may be
concerned that they may give you incorrect data. Because such anxiety
may be associated with a sense of embarrassment, they may be reluctant
to get involved in a research project, and it may be diYcult for them to
say why because of this sense of embarrassment.
On the other hand, this kind of assessment of someone’s linguistic
ability may be unrealistic in the sense that he or she may well have a
perfectly good command of the kinds of things that you are interested in,
such as basic vocabulary, as well as a full command of the phonology,
morphology, and syntax. It is only natural that such people may not have
a full command of rhetorical styles or arcane vocabulary. However, if you
are just starting your work on a previously undescribed language, this
does not really matter all that much, as you would need to get this kind of
basic information down Wrst before you need to start thinking about
the frills. Your initial task, then, will be to reassure your reluctant
language-helper that he or she really can help you in a meaningful way.
When you start your initial round of elicitation, you should be
prepared to meet with a variety of kinds of responses to your requests to
record linguistic data from speakers of little-known languages. In general,
I have found that such requests meet with an enthusiastic response. People
are often keen for their language to be documented, as they may be aware
of what has happened to other languages which have suVered from
neglect in the past. There are likely to be other positive responses to
your expression of interest in describing their language. I once arranged
for a speaker of a Vanuatu language to pronounce some words for one of

my linguistics courses to give my students some practice in phonetic
transcription. Every time he said a word in his language, he giggled.
When I asked him what was making him giggle, he said he felt very strange
as he had never uttered words in his language to Europeans before. He said
it felt ‘odd’, to the point where he could not suppress the urge to laugh.
However, such positive responses may be tempered with a number of
other responses as well. You should not assume that working on some-
body’s language is automatically going to be a straightforward process
once you get started. Initial enthusiasm may gradually—or even
quickly—wane when it is discovered that this kind of work usually
involves a fairly long-term commitment on the part of the speaker of the
language. Initial enthusiasm can easily enough turn to boredom, as you
will sometimes Wnd it necessary to complete repetitive morphological
paradigms to get the full details of allomorphic variation. Also, the person
you are working with may agree to work with you when he or she is not
particularly busy, but if that person has other commitments, their work-
load may well build up to the point where they cannot spare the time to see
you as regularly as they could in the past. Missed appointments can be
frustrating for you, but they may be unavoidable for the other person.
You also have to be prepared for other kinds of reactions. I know of one
linguist who was well known for his ability to acquire a basic speaking
ability in any language that he had worked on, and to retain this ability for
many years. One story goes that this linguist was somewhere in Queensland
and he recognized a stranger in the street as being a likely speaker of a
language that he had worked on in the past in Papua New Guinea from his
facial features. He addressed this chap in a friendly sort of way in his own
language, but his response was to turn tail and run. In all probability, he
had entered Australia illegally across the somewhat porous boundary
between Papua New Guinea and Queensland, probably by canoe or small
boat, and had thought he could pass as a local indigenous person. Anybody
who is illegally in your country may be understandably unwilling to act as
a language-helper for the unwelcome visibility that this role may provide.
4.1.3 Language-helpers in the Weld
Carrying out research in the Weld is likely to proceed in a diVerent way

and at a diVerent pace from the way you might proceed when researching
the same language ‘at home’. Many linguists set out to choose a single
individual who is to become their language-helper on a regular basis.
Typically this person will be paid at a locally appropriate level for his or
her time. With any arrangement that you come to, you need to keep in
mind the fact that this person is going to have certain family and
communal obligations, which means that you will not always be able to
have access to your language-helper whenever you might want. Having
arrangements with several people at once is one way of overcoming this
problem.
Choosing the right person, or people, to work with can sometimes be
quite an issue. If you are going to be working at close quarters with
somebody for many months at a stretch, it is not just a question of certain
individuals being capable language-helpers and others being incapable.
Interpersonal considerations also come into play here, so somebody
who is a good language-helper for me may not necessarily be a good
language-helper for you, simply because we are two diVerent people.
If you enter into some kind of formal contract-like agreement with a
particular person, it can sometimes be diYcult—or impossible—to
‘dismiss’ him or her because this may cause more than just ill will between
you and your former language-helper. If that person happens to have
been an important member of the community—a chief, perhaps, or a
religious leader—then the ill will may come from the entire community.
I therefore try to have a much looser arrangement with language-helpers
whereby I can call upon somebody’s services when I need help, as long as
that person is available. This means that I can call upon somebody more
frequently or less frequently according to how successfully we have been
able to work together. No actual ‘dismissal’ ever needs to take place if
things don’t work out well, as I can simply gradually reduce the extent to
which I call upon that person’s help.
Local communities, once they understand the nature of your project,
will sometimes decide for themselves who they consider to be the most
appropriate individual, eVectively giving you very little choice about who
you work with. This arrangement can work well, as the person most likely
to be chosen is one whose linguistic knowledge is most highly regarded. In
my work on Malakula and Erromango, for example, it has been
assumed that I would work extensively with the local Cultural Centre
Fieldworkers, and these arrangements have worked well. These are
volunteers from within the local community who liaise with the national
body that coordinates cultural and linguistic research. They are typically
chosen both for their interest in this role, as well as their knowledge of
local traditions. Even if there are no individual projects under way in
a particular language area, these Weldworkers are given some basic
training, equipment, and encouragement to document various aspects
of the local culture. When some kind of research project is approved
within the local community, there is an expectation that these Weld
workers will be centrally involved in the project, provided with some
kind of wage, and also given some kind of further training or experience
by the visiting Weld linguist.
At the same time, there are potential pitfalls in having your language-
helper chosen for you. Sometimes, the person who you are expected
to work with will be an older member of the community. Such people
will typically be good at telling stories, but they may sometimes be less
good at paradigmatic elicitation. An older person may also have very
prescriptive attitudes toward language. This may get in the way of your
desire to be exposed to natural everyday language and it may result in
you being exposed only to relatively archaic or formal varieties of the
language. Such people may also try to make sure that loanwords which
are in everyday use throughout the speech community do not appear in
your Weld notes.
If you are working in a language that is spoken in a town or a city
rather than a small rural village, you will almost certainly have more
choice about who you are able to work with. You may Wnd it preferable to
choose somebody who is closer to yourself in age, temperament, or
experience. While in some societies, it may be perfectly acceptable for a
man and a woman to work together on a research project, there may be
situations in which it is a good idea to match the gender of the researcher
and the language-helper, particularly if one or both are unmarried, given
the possibility for signals to be misinterpreted in these kinds of situations
(Moreno 1995).
It may be that somebody with whom you naturally fall into a friendship
ends up more or less spontaneously becoming your language-helper. This
way, when you are not actually working, you can be hanging
out together, and your ordinary social interactions are likely spontan-
eously to produce additional linguistic data. For instance, you may be out
in the bush together and if you come across trees for which you have not
yet recorded names, you can easily add these words to your lexicon.
Similarly, if you are out in a canoe over the reef, there will be Wsh and
other sea creatures for which you will be able to record the names. If your
language-helper is an older person, however, he or she may have reduced
mobility and is perhaps more likely to sit around at home. This means
that the amount of ‘after-hours’ spontaneous language data that you will
be able to record will be less.
While there can be advantages in seeking to match your own age with
that of your language-helper, there may be limits on how closely you will
be able to do this. If you are a relatively youthful graduate student,
members of the community may not view it as appropriate for somebody
in their twenties to be the primary—or even the only—source of your
linguistic information. One way around this might be to work with
a younger person for certain kinds of tasks, such as transcribing stories
or making grammaticality judgements, while working with older people
who are more respected in the community for tasks such as lexical
elicitation, and for the recording of stories.
You may sometimes be expected to work with the very old because of
the community’s respect for their knowledge and experience. However,
older speakers are much more likely to tire easily so there may be real
limits to how much time you can work with them. Old people often also
have unclear articulation. I have recorded old people whose distinctions
between l and n, or between l and r, were almost impossible for me to
hear. Many Australian languages have phonemic contrasts between
dental and alveolar stops, nasals and laterals, but if an elderly speaker
no longer has any front teeth, it can be very diYcult to work out which
sounds are being produced.
I have also found that older people are much less likely to have an
ability to recognize paradigms or other grammatical patterns. That is, if
you ask someone how to say I sing, some people will very quickly learn
to also give you the forms for you sing, (s)he sings, we sing, they sing,
and so on. Older people, however, get very confused by paradigmatic
elicitation, and frequently give the incorrect forms. For instance, if you
ask how to say I sing, some people will actually give you the form for
you sing, and when you ask for you sing, you end up with the form for I
sing instead. Some older people quickly get bored by the whole process
of paradigmatic elicitation, feeling as if they are just repeating the same
word over and over. (‘I’ve already told you that word, why are you
asking me again?’) My experience tells me that it is younger people who
are more likely to understand that I am actually asking for diVerent

words.
However, the same people who get bored or confused with paradig-
matic elicitation may turn out to be excellent providers of vocabulary.
Some people seem to have an ability to think up new words that you
would never have thought of asking for, and they can provide additional
senses of a word that somebody else might not think of. Yet other people
often turn out to be particularly good at helping in the transcription and
translation of recorded stories, while other people Wnd this tedious and
frustrating.
However many people you end up working with, you are going to have
to provide them with some clear idea of what it is that you want from each
of them as you sit down to work together. While you may be quite
familiar with the idea of linguistic elicitation, this is something that
your language-helpers will never have been asked to do before. You
may therefore need to explain even very basic notions to people, such
as the fact that you need them to pronounce sentences slowly, word
by word.
When you ask somebody to repeat something, you will need to make it
clear that you need an exact repetition of the same thing rather than some
kind of a paraphrase. When you are transcribing a text, you will need to
make it clear how much text you can handle at a time before you have to
play the next section of the story. You will also need to make sure that
people repeat exactly what they hear on tape, even where it contains some
kind of error. Of course, you will indicate in your notes what the nature of
the error was so that this can be corrected in the version that you Wnally
make available to the community.
4.2 Phonological and grammatical elicitation
Social scientists often approach their research projects with ‘question-

naires’ which they ‘administer’ to their ‘subjects’. Field linguists often do
something like this, though we tend to talk instead about ‘prompts’
which we ‘present’ to our ‘informants’ (or, as I have already suggested,
‘language-helpers’). That is, you will very often start your data-gathering
by saying to somebody with whom you share a common language ‘How
do you say X in your language?’
4.2.1 Language of elicitation
If the major lingua franca of the area where you are working is English,
then you will need to Wnd somebody who speaks English well enough with
whom you can use English as your medium of elicitation, at least in the
initial stages. In a rural area, this means that your Wrst language-helpers
may need to be schoolteachers or other people who have completed more
than the minimal amount of formal schooling. If the major lingua franca
of the area is not English, you are almost certainly going to have to make
a fairly early switch to using whatever that language may be.
If there is a local lingua franca—such as Melanesian Pidgin in the case
of the Melanesian countries or Swahili in East Africa—it is possible that
this language will be typologically more similar to the local language than
English is. This means that the lingua franca may make more of the same
kinds of semantic and grammatical distinctions that are made in the local
language you are studying. English, for instance, does not have separate
inclusive and exclusive pronouns in the Wrst person plural, using the form
we for both situations. In Melanesian Pidgin, there is a systematic
distinction between inclusive yumi and exclusive mifala, and this distinction
mirrors what we Wnd in practically all of the Oceanic languages of
Melanesia. In eliciting verbal paradigms, therefore, it is often going to be
much easier to operate through the medium of Melanesian Pidgin than
English, even if your language-helper speaks perfectly good English.
Sometimes, of course, your own knowledge of the local lingua franca
may not yet be up to the task of elicitation, and you may not be able to
Wnd a single person who speaks English. In this rather unfortunate kind
of situation, you will have no option but to start the process of elicitation
using an interpreter. This means that you will need to Wnd somebody who
you can communicate with in English or some other language, and who
also knows enough of another language that local people know so that
your questions can be translated. You will then be able to transcribe the
response of the local person.
I, as somebody who speaks no Spanish, could therefore engage
somebody who speaks Spanish and English to interview somebody else
who speaks Spanish and Zapotec in order for me to work on Zapotec. It
does not take much imagination to see that this would not be a very
satisfactory way of doing Weldwork, at least in the long term. It is well
known that a competent bilingual ability does not automatically equate
to an ability to translate, and messages can easily get confused in the

process of transmission. This process requires that both the interpreter
and the language-helper grasp the point of my question. The merest hint
of ambiguity or lack of clarity at any point becomes the weak point in the
chain, as the children’s game of Chinese Whispers clearly illustrates.
Ultimately, you may want to move towards monolingual elicitation in
which you learn enough of the local language to be able to speak directly
to your language-helper in that language. Some linguists become very
conWdent at gathering information this way, while others continue to
elicit through the medium of the local lingua franca.
4.2.2 Working environment
You should try to organize a place in which you can work comfortably
with somebody while you are gathering your data. You will of course
need to operate within the limits of what is available locally, unless you
have been able to bring your own oYce furniture with you. For personal
comfort, you will need a table with a smooth surface of the right working
height, a stable chair with a comfortable back, good light, and good
protection from the sun and rain.
However, you can’t always get what you want, so you may sometimes
end up balancing your notebook on your knee while you are sitting on
a log or an upturned canoe. Make sure then, that you come prepared with
a Wrm clipboard, or that you have a notebook with a hard cover. If the
tables and chairs are made within the community, you may simply have
to make do with a wobbly table that is uncomfortably low or high, or
a stool without a back. If the place that you are expected to work in is a
poorly ventilated galvanized iron building in which it is hot enough to
bake a loaf of bread any time after 11.00 a.m., you will have to learn to
write in such a way that your forearm does not drip sweat all over the
page that you are about to write on. If your working conditions are less
than ideal, don’t try to push yourself beyond your physical limits, as
you are not going to be much use to anybody with a sore back or a
semi-permanent crick in your neck.
You also need to Wnd somewhere to work that is reasonably quiet and
private. You may Wnd that your activities attract all manner of interest,
and people may end up standing around watching. If you are an
exhibitionist, this may be Wne, but I Wnd myself wanting to scream after
just a short period of being stared at while I’m trying to work. Also, if
you’ve just heard a word and you can’t quite decide if the vowel was long
or short and you ask for it to be repeated, you risk getting half a dozen
other people repeating the word at not quite the same time, which is
practically no help at all. If young children are your main audience, you
can probably get rid of them fairly quickly, but if you are attracting an
audience of adults, this can be more diYcult to achieve without causing
oVence.
4.2.3 Initial elicitation
A Weld linguist should always begin a session of work by preparing

enough questions to last for two or three hours. To prepare yourself,
you will need to go over your previous notes so that you have a good idea
of the kinds of things that are likely to be interesting, or which still
represent problems in your analysis. It is not a good idea to come to an
elicitation session with a blank sheet with the intention of just thinking
of things to ask as you go along. From time to time, you are likely to
get stuck for something to ask, and as you sit there thinking, your
language-helper is likely to be getting bored. A bored language-helper
very easily becomes distracted, and this is where mistakes are often made.
It is a good idea not to aim to work with somebody for more than a
couple of hours at a time, or maybe three hours at the outside. Beyond
this length of time, I Wnd that language-helpers tend to get bored, and
I have already indicated that this can easily lead to mistakes being made.
In fact, I Wnd that much more than two or three hours at a stretch is about
all I can cope with myself, especially if my working environment is in any
way uncomfortable. This then gives you plenty of time in the same day to
analyse the data that you have gathered, and to plan your next session of
elicitation.
Elicitation should normally begin with some basic vocabulary,
i.e. words that are going to be easy to gather, and which are likely to
involve the smallest number of grammatical mysteries. The standard
Swadesh 200-word list is a good place to start (Samarin 1967: 220), but
there is no reason why you should not veer away from this in your
vocabulary elicitation if people seem interested in providing additional
words in particular semantic domains. This basic wordlist was also
compiled as a non-culture-speciWc list, so there are likely to be additional
words that relate to local culture, as well as the local environment, that
you can elicit very early on. Your primary purpose at this initial stage is
not really the recording of vocabulary as such, as it is highly likely that at
least some of your lexical material at this early stage will contain errors of
various kinds. Ultimately, all of the lexical data you record in the early
stages will need to be properly checked again for phonetic, semantic, and
grammatical accuracy.
One of the Wrst basic lessons in elicitation should be, ‘Never tell your
language-helper that he/she has given the wrong answer.’ Always write
down what somebody tells you, even if it is obvious to you that it is not
the answer to the question that you asked. The reason for the error may
be your language-helper’s, for mishearing or misunderstanding your
question. However, your own knowledge of the intermediary language
may not be nearly as good as you think it is, resulting in your question
being unclear. If the answer that you have just written down was not the
one you wanted, you should try asking for the same information in a
diVerent way.
Your primary purpose in the initial stages of linguistic elicitation is to
formulate working hypotheses regarding the phonological system of the
language, i.e. what sorts of phonemic contrasts are made, and what
phonetic diVerences represent conditioned or free variants of the same
phoneme. You will also be looking for patterns relating to the phonotactic
system of the language, i.e. can words contain consonant clusters? If
so, what kinds of clusters are permitted? Are clusters permitted only
between vowels, or do they also occur at the beginnings of words and
ends of words? These (and many more) questions are the kinds of things
that should be going through your mind at this initial stage of lexical
elicitation.
The Wrst few sessions with a speaker of a language will often sort out most
issues of phonemic analysis for most languages, with perhaps a relatively
small number of problems emerging as areas worthy of closer attention
once you manage to accumulate more data. The best advice at this stage
would be to suspend the search for solutions on some of these less imme-
diately tractable problems until you’ve done some work on some of the
other more basic aspects of the language. Some of these kinds of problems
end up solving themselves anyway when other kinds of data come in.
When you are carrying out your initial vocabulary elicitation, it is
a good idea to start with terms for body parts and the names of things
that you can see around you in your immediate vicinity, such as the
heavens, the ocean, the rivers, and the land. As I stated earlier, the aim
should be to operate with vocabulary in the early stage of elicitation
that involves few possibilities of grammatical complexity. Body parts,
for instance, may sometimes turn out to be an inappropriate area to
concentrate on in elicitation if it turns out that there is obligatory
marking of inalienable possession. Verbs are often best avoided in the
initial sessions because there may be obligatory marking of inXectional
categories such as subject and/or object, as well as tense/aspect/mood.
For many languages, the Weldworker will probably have gained a
reasonable idea of what constitutes the set of phonemic contrasts when
something like 300–400 words have been elicited. If there are still some
remaining mysteries, you could expand your vocabulary with a few more
sessions of lexical elicitation to see if this adds enough additional infor-
mation. Of course, it may be that for some languages, the phoneme
inventory is much more complicated, in which case you are likely to
have to gather a much larger working vocabulary before you can be at
all conWdent about the set of phonemic contrasts.
No doubt you will at some stage want to check on what look like
minimal pairs in your growing corpus of lexical data. You may want to
pronounce these pairs of words yourself and ask your language-helper if
you have pronounced them correctly. There is a real risk, however, in
doing this, as simply being able to understand what you have said, no
matter how badly you have mauled the sounds, may be suYcient
to warrant a positive response. It is always far better to get your
language-helper to repeat what he thinks you have said so that you can
check to see if that matches up with what you were attempting to say in
the Wrst place.
4.2.4 Eliciting grammatical patterns
Once your phoneme inventory looks reasonably secure, you could then
start moving on to the elicitation of grammatical data. You could start by
testing for the existence of diVerent possessive patterns by asking for
some of the nouns that you have already elicited, but asking how people
would express these with noun and pronoun possessors. That is, if you
already have asked the word for canoe, ask how to say my canoe and the
boy’s canoe. Once you start eliciting paradigms like this, it is important
that you elicit the full paradigm in case some parts of the paradigm
exhibit particular patterns of morphological behaviour. This means that
you should not ask for just my canoe, but also your (sg) canoe, his/her
canoe, our canoe, and so on.
In my experience, once you start working on grammatical patterns, it
becomes necessary to vary what you are doing during your sessions.
When you are eliciting vocabulary, you are likely to be asking for
words in closely related semantic Welds all together, e.g. body parts,
trees, Wsh, and so on. However, people are less likely to recognize from a
whole series of sentence prompts that you are interested in the expression
of the past tense, conditional constructions, relative clauses, and so on. If,
during your elicitation session, you notice the tell-tale signs that your
language-helper is suVering from boredom or nicotine withdrawal (or if
you are), then by all means break oV your elicitation and relax for a few
minutes before getting back into the swing of things, perhaps from
a slightly diVerent direction.
You could perhaps start looking at what sorts of categories are marked
on verbs, as there may be obligatory subject aYxes, or object aYxes, or
tense/aspect/mood aYxes. So, you could ask for simple sentences such as
My canoe is sinking, My canoe sank, My canoe will sink, and so on. You
could quickly move on to negative constructions by asking for sentences
such as Your canoe will not sink and Their canoe did not sink. Again, it is
important as far as possible to elicit full paradigms for each verb for all of
the categories that are associated with each verb in case there are regular
allomorphic variants to be found among any of the verbal suYxes
or preWxes.
As soon as you start eliciting pronominal paradigms of this sort, you
need to be aware of the kinds of confusion that can arise between the
prompt of the Weld linguist and the response of your language-helper.
Some people, when asked for the word (or phrase) for my eye will
translate literally what the linguist has said. Other people, when presented
with the same prompt, will instead give the equivalent of your eye. This
kind of thing can cause confusion for the Weldworker, which you will
need to sort out. Ideally, you should not attempt to correct your
language-helper. Instead, you should simply note down anything that
he or she says, and if something that is said later seems to contradict what
was said earlier, you could try asking the same question all over again to
see what sort of response you get.
Once you think that you have sorted out some of the basic patterns
of obligatory inXectional marking on nouns or verbs, you will be ready to
begin the elicitation of slightly more elaborate sentences. You will need to
establish what the basic constituent order of the sentence is, i.e. whether
the language is an SVO language, or if it exhibits some other kind
of pattern. To do this, you can present simple sentences for translation,
such as:
The man saw the dog.

The dog ate the meat.
As far as possible, you should stick with words that you have already
elicited when asking people for these kinds of sentences. It is also a good
idea to ask people to translate only those kinds of sentences that they are
ever likely to need to say in the language. Since kangaroos are herbivores,
it therefore makes little sense to ask a speaker of an Australian language
how they would say The kangaroo ate the wallaby. If you need to elicit the
verbal object form of the word for wallaby, use it in association with
another verb that makes much more sense, such as The kangaroo saw
the wallaby.
It may also be necessary to work out what sorts of meanings are
unlikely ever to be expressed for cultural reasons. For example, there
are societies in which the following is unlikely ever to be said:
The man wove a mat with his mother-in-law.
People may be unlikely to say a sentence like this for a number of reasons.
If only women traditionally make mats, you would be unlikely to need to
speak of a man who is weaving a mat. If mats are always made by a single
person, it may make little sense to speak of two people working together
on a mat. Finally, if a man and his mother-in-law are governed by social
rules that prevent them from sitting together in such close proximity, the
two of them would be unlikely to appear together in a single sentence of
this type.
Finally, you should stick to words and activities in your sentences that
you know people will have readily accepted words for in their language.
For many languages, it would not make much sense to start eliciting
grammatical patterns with sentences such as the following:
The student booted up the computer.

This is because in many parts of the world, this kind of meaning is likely
to be expressed using ad hoc loans from a language of wider communi-
cation rather than words from the local language.
While this is a very obvious example, there are other examples that are
perhaps less obvious. When I was eliciting vocabulary in Paamese, I asked
people for the word for ‘get married’, which I wrote down. Then I
asked the word for ‘get divorced’. The response was a brief pensive
look, followed by the comment:
We don’t get divorced.
When you are investigating the grammar, you will need to Wnd out how
a range of additional information is encoded in the simple sentence, such
as how peripheral semantic roles are expressed. You will therefore need to
present English prepositional and adverbial phrases for translation, such
as the following:
The man saw the dog in the river.

The dog ate the meat under the house.
The man will see the dog tomorrow.
The dog died because of the poison.
Finally, you will be able to move on to the elicitation of a range of

complex sentence constructions. You would therefore start asking for the
equivalents of relative clauses, e.g.
The man saw the dog which swam across the river.
or other kinds of subordinate clauses, e.g.
The dog swam across the river because it was hungry.

The dog drowned when it tried to swim across the river.
If the dog swims across the river, it might drown.
You would also need to elicit coordinate sentence patterns, e.g.
The dog swam across the river and the man spoke to his son.
The dog swam across the river but the man spoke to his son.
If I thought it might be practical to provide a universal elicitation

schedule which any Weldworker could work through in order to analyse
the grammar of any language, I would be happy to include this as part of
this guide to Weldwork. However, I think that this would be of only
limited value. Of much greater value are elicitation schedules that have
been designed speciWcally for languages in certain geographical or genetic
groupings (e.g. Samarin 1967: 110, Bouquiaux and Thomas 1992). Of
course, once you become aware of the structural features that are
commonly encountered in the language family that you are working on,
you may even be in a position to draw up a basic elicitation schedule
yourself.
Any intending Weldworker needs to survey the literature for closely
related languages to Wnd out what kinds of patterns are found in those
languages and what sorts of constructions are absent. For somebody who
is intending to work on an Australian language, it is essential to know
that in these languages there is often just a single possessive construction
for all kinds of possessive relationships, while in Oceanic languages, there
are often distinct possessive constructions depending on the nature of the
precise relationship between the possessor and the thing possessed.
Thus, the direction of your elicitation needs to be guided by patterns
that are revealed in your background reading for related languages.
4.2.5 Taking notes
When you elicit data in the kinds of ways that I have been talking about,
there are essentially two diVerent ways of keeping a record of what
people say. On the one hand, you can simply write everything down in a
notebook. Alternatively, you can use your audio recorder in much the
same way as a notebook by recording every response. When I Wrst started
doing Weldwork on Australian languages, I recorded pretty well everything
on tape. Anybody listening to those tapes in the archive will therefore
hear my prompt, followed by the response from the speaker of whatever
language I was recording at the time. I then transcribed these tapes in full
and my deposited ‘Weld notes’ are held in an archive in handwriting that
I have never since been able to match in terms of tidiness.
There are advantages in this kind of approach. In the case of the
moribund languages that I was working on at the time, we now have an
extensive record of the pronunciation of all of the words and individual
sentences that I recorded, and future linguists no longer have to rely solely
on my transcriptions of that material. However, if you are constantly
switching the recorder on and oV as you elicit, you are likely to distract
your language-helper to some extent. You are also likely to make it more
diYcult to take notes while you are recording. If you were to record your
language-helper’s responses directly into a notebook, you could much
more easily make little notes to yourself, oVering possible alternative
transcriptions and points to check, while also noting down interesting
asides from your language-helper.
While some linguists continue to make extensive use of audio
recordings, these days I use only notebooks for recording responses in
direct elicitation sessions. Of course, my notebooks are a horror for
anybody else to look at because they contain hurried handwriting as
I try to keep up with the rapidity of somebody’s speech, and there are
scribbled notes relating to all manner of interesting side observations. At
the moment, my notebooks are all locked away in a storage cupboard to
which only I have access, and I am not sure that I would ever want them
to go into a public archive.
Sometimes, a Weldworker’s notebooks may even point to aspects of his
or her personality or behaviour that are best kept private. I remember
reading through some of the original handwritten notebooks of one
amateur linguist who recorded valuable data from a large number of
Australian languages in the late 1800s and early 1900s. Most of these
languages no longer have any speakers and these notebooks, and
the resulting publications, represent the only information that we will
ever have.
Interestingly, the original notebooks occasionally contained infor-
mation that was excluded from the Wnal publications, such as words
with glosses like ‘the sound of bellies slapping during sexual intercourse’.
Of course, we can only guess at the circumstances under which such a
word could have been elicited, but it would not be diYcult for anybody’s
personal predilections to make their way into a notebook in this way,
only to be Wltered out in the Wnal selection of published examples.
4.2.6 Native-speaker judgements
Once you get beyond the introductory stages, you will need to begin
testing hypotheses about patterns that you should be beginning to
formulate in your own mind as you go. This can be done by presenting
forms to a speaker and asking for a judgement as to the correctness or
otherwise of that form. However, there can be problems associated with
this kind of introspective method of data-gathering and such judgements
should only be accepted when there is corroborative evidence, either in

the form of spontaneous utterances of the same type, or consistently
made judgements from several diVerent elicitation sessions, or similar
judgements from several speakers. This means that you should not
place too much faith in a single grammaticality judgement from a single
speaker on a single occasion.
One major problem in presenting forms for judgement in this way is
that your notion of ‘grammaticality’ can very easily be confused with the
speaker’s notion of ‘understandability’. That is, if what you have said is
understandable, your language-helper might say that what you have said
is OK. It is not always easy to ensure that this kind of judgement
corresponds to what people would actually say. Fieldworkers, when
they later Wnd out that sentences that were earlier given the OK are
in reality ungrammatical, have been informed in terms such as the
following: Well, you can say that, but we wouldn’t. Or: That’s alright, but
it’s not really what we say. Any comment like that should be taken as
a warning that you are possibly on the wrong track with your elicitation
strategy. So, you should avoid strategies such as the following in
your elicitation sessions:
Is X grammatical?
Can you say X?
Rather, you should formulate your elicitation sessions with questions

more like the following:
Would people say X?
It is also going to be very diYcult for you to control for all possible
variables when you present a sentence in isolation for a grammaticality
judgement like this. A simple change in intonation—which you may be
completely unaware of—may completely change the acceptability of an
utterance within any particular real-world context. Thus, somebody may
respond that your prompt is incorrect (or that it is correct), but you and
your language-helper may be thinking about quite diVerent real-world
contexts in which this would be the case.
One way of giving grammaticality judgements greater reliability is to
present people with two (or more) options and to allow them to indicate
which they think is the better sentence. Thus, rather than any of the
strategies above, it may be a better idea to ask somebody:
Which is better: X or Y?
When you are eliciting sentences, and perhaps more so when you are
transcribing texts, you are likely from time to time to encounter a pattern
that you simply do not understand. Maybe a completely new morpheme
has appeared in your notes for the Wrst time, or maybe a morpheme which
you thought you understood perfectly suddenly appears in a context
which does not Wt with what you had come to expect. No doubt you
will give a little pause, furrow your brow, and you will want to ask your
language-helper:
Why did you say X rather than Y?
Very often, the answer to this kind of question calls for much greater
linguistic sophistication than any linguistically untrained language-helper
is going to be in a position to provide.
My advice to the Weldworker is to ask this kind of question just once.
If you are lucky enough to have a language-helper who is intuitively gifted
in explaining patterns of language, then by all means continue to ask
these sorts of questions. But you need to understand that in response to
this kind of question, most people are just going to say something
unhelpful like the following:
We say it that way because that’s the way we say it.
Further reading: Bouquiaux and Thomas (1992: 3–70), Hale (2001: 81–95),
Mithun (2001), Samarin (1967: 106–50, 175–204).
4.3 Gathering words
The grammatical system of a language is something that is far more

circumscribed than the vocabulary of a language. While writing a gram-
mar can be challenging enough, as well as time-consuming, it can be much
more diYcult to compile a substantial dictionary of a language. We
see this diVerence reXected in the number of published grammars of
languages in comparison to published dictionaries of the same languages.
Some of the most widely cited published grammars have no corresponding
published dictionaries. This is no doubt in part because some linguists
see the lexicographer’s work as that of Samuel Johnson’s ‘harmless
drudge’, which is beneath their dignity. But I have little doubt that part
of the reason is also that there is simply far too much work involved if the
job is to be done at all well.
4.3.1 Direct elicitation
I have already described how you can start a lexical compilation for
a language by directly eliciting vocabulary via another language. You
should always ensure that you elicit vocabulary in semantic Welds rather
than running through a list that is organized alphabetically in, say,
English. If you have a list of words that some other linguist has gathered
in the language that you are working on and you want to check these
forms for accuracy, you should again try to put the words together in
semantic groupings Wrst, rather than presenting words to somebody in
what will seem like random order if they are listed alphabetically. There
are some useful wordlists that you may want to get hold of before setting
out for Weldwork: for example, in Australia there is Sutton and Walsh
(1979), and more generally, but with an African focus, is Bouquiaux and
Thomas (1992). Similar elicitation aids can be found for other regions.
Although body parts represent one of the easiest semantic Welds in
which to elicit vocabulary, you need to be careful not to try to elicit
names for all body parts at the very beginning of the relationship with
your language-helper. I have found that the initial stage of elicitation in a
new language sometimes attracts an audience, which makes the occasion
a very public one. I therefore tend to skip straight from navel to knee
when eliciting body parts if there are other people around. This is because
words for any of the intermediate body parts may cause embarrassment if
men or women do not use these words in each other’s presence in public.
In fact, there may be even more at issue here than embarrassment because
in some societies, the use of certain words may be completely taboo in the
presence of particular relatives. You can always wait till your relationship
with your language-helper becomes relaxed enough that you can ask for
these kinds of words in private.
You should be warned that these kinds of restrictions may not apply
just to the names of body parts between the navel and the knee. Mention
of certain cultural items, for example, may also need to be avoided
in public, though it may be more diYcult to predict what cannot be
mentioned, and when. On one occasion, I was eliciting vocabulary for
a number of cultural items which were described in an ethnographic
account of one society. I was working with an adult man but a number of
young boys came to sit and quietly listen to what we were doing. The two
of us had earlier been talking about the bullroarers that are used to make
a frightening noise when boys are being circumcised. I had forgotten to
write down the name at the time so I asked what it was again. My answer
was a quick kick in the shin under the table. I immediately realized that
this was a surprise that still awaited the boys in the audience and had to
delay my question to a more opportune moment.
It can sometimes be helpful to elicit vocabulary with small groups of
people rather than in one-on-one sessions with a single language-helper.
It can be remarkably easy for somebody on their own to be unable to
think of a translation equivalent for even a fairly basic vocabulary item,
but another person will often be able to step in straight away with the
appropriate word. When one person says one word, this may also remind
some of the other people present of other related words. People in groups
can sometimes also Wnd additional senses to words in discussion which
might not necessarily occur to somebody working singly with a linguist.
While group sessions like this can be a productive way to work, you
need to be aware that there can sometimes be a downside as well. One of
my language-helpers on Malakula once decided that it would be a
good idea to call for a meeting of interested people one Sunday after-
noon to help sort out some vocabulary relating to speciWc aspects of
the traditional culture. Something like Wfteen or twenty people turned
up. Things were not run as a formal meeting, so people sat around and
contributed when they wanted to, and they listened (and occasionally
dozed) when they had nothing to contribute. The problem was that every
time I wanted to check that I had transcribed a word correctly and I asked
somebody to repeat it, I would tend to get repetitions from several people
at once, meaning that I heard nobody’s repetition properly. Try as I could,
people did not seem to understand why I would only want one person to
repeat the word if I had called a public meeting of many people.
4.3.2 Beyond elicitation
Accumulating vocabulary simply by asking for translation equivalents

from the language of elicitation is going to come up against a blank after
a certain point. This is because many English words will not necessarily
have convenient equivalents in another language. There will similarly
be many words in the other language that do not have direct equivalents
in English. For instance, you would probably only come across a word
meaning something like ‘arrive on the scene just as somebody is talking
about you’ in a text, as you would almost certainly not think to ask for
such a word if you were getting translation equivalents from English.
Even if you did happen to think to ask for such a word, there is no
guarantee that any given individual would be able to think of that
particular word in response to such a prompt. Another issue related to
the direct elicitation of vocabulary involves the extended senses
of a particular word. While you might manage to elicit a word with
a particular meaning, your language-helper may not immediately think
of any of the extended senses of the word.
Any time that you see something new, you can ask what that thing is
called. The easiest way to Wnd new names for trees is to take a notebook
with you every time you go into the bush with somebody. Every time you
see somebody doing something new, you can ask somebody to say what
they are doing. Sometimes these items or these activities may not be
‘lexicalized’, i.e. there may not be a unique word to express that particular
meaning, but sometimes there may be and you will be in a position
to record it.
This kind of approach means that you should always remember to have
a pen and a pocket notebook with you. I have been known to forget and it
then becomes necessary to become creative in Wnding ways to remind
yourself of the new words that you have recovered. I was once in the bush
with some people who taught me some new words that I did not want to
forget, so I ended up scratching the words onto the husk of a green
coconut with a knife. Of course, I then had to face the inconvenience of
carrying that coconut with me for the rest of the day, or risk forgetting
those words. Of course, there are going to be times when you won’t be
able to have your pen and notebook with you. I suspect that my own
lexical collections relating to creatures living on the sea Xoor are less rich
than they could be for this very reason.
Another way in which you could add to your stock of words is to get
people working in brainstorming groups dealing with particular topics.
That is, perhaps you could encourage one group of people to work
together to come up with as many names of diVerent kinds of trees as
they can think of. Another group could be working on Wsh, and another
group could be thinking of names of diVerent kinds of birds. There could
be a large number of diVerent groups of people working on a wide variety

of words belonging to particular semantic Welds, such as terms involved in
house-building, the construction of canoes, the weaving of mats and
baskets, and so on. The advantage of having people working in particular
semantic Welds like this is that one word may easily lead to another word
in the same area.
There may be illustrated books which you can also use as prompts to
elicit vocabulary. In many parts of the world, there is a variety of
publications dealing with local Xora and fauna and people may be able
to provide names of trees, plants, birds, Wsh, animals, and insects from the
pictures that are contained in these volumes. However, there are often
limitations to the usefulness of books such as these as aids to elicitation.
For one thing, people sometimes Wnd it diYcult to recognize individual
species when presented with a small illustration, especially if there is no
indication of what other species are found nearby as part of the normal
environment in which that species is found. Sometimes, while there may
be a visual clue distinguishing one species from another, it may be that the
vital distinction in the mind of speakers of some languages depends on
one of our other senses. For instance, people may distinguish one
tree from another by the smell of its bark or its Xowers, or they may
distinguish one bird from another by their song, or by the kind of nests
that they make.
It must also be remembered that speakers of languages do not
necessarily classify the world in the same way that botanists and zool-
ogists classify things. This means that sometimes diVerent species that
are similar in particular ways may actually be referred to by the same
term. Sometimes, diVerent varieties of the same species may also have
diVerent names with no generic name for the overall species.
Another possibility is that people may have quite diVerent names for
the same species depending on the stage of growth it is at.
4.3.3 Spontaneous speech
While various strategies of direct elicitation can certainly help in the

search for new vocabulary, it is extremely important that elicitation be
supplemented by an extensive collection of spoken texts. I have already
indicated that the recording of texts is essential for a detailed understand-
ing of the grammar of a language, as well as for understanding how
sentences are linked together in discourse. But these same texts will also
contain a huge amount of very valuable lexical information. You will Wnd
that in the Wrst such texts that you transcribe, you will encounter many
new lexical items, as well as many examples of words that are used with
previously unrecorded secondary senses. You may also encounter a range
of words which appear in new idiomatic constructions.
The more texts you are able to record the better, and the more varied
these texts are in terms of content, the more new words you are likely to
discover. I have also found myself happy to agree to record somebody
telling the same story that somebody else has already told because it is
most unlikely that two people talking about the same topic will use
exactly the same range of vocabulary.
However, by the time that you have recorded a certain amount of
speech, each individual text will provide fewer and fewer new pieces of
new lexical information, except perhaps for texts that deal with specialized
knowledge of some kind. Before too long, you may Wnd that each
individual text may provide you with just one or two new pieces of
information. The Law of Diminishing Returns comes into eVect here. At
some point, you need to ask yourself whether the massive task of recording
new texts and then transcribing them is worth the relatively small amount
of new lexical data that you are likely to discover.
Of course, once you encounter any new word in a text, you can then
take that as an opportunity to present the word to people for another
brainstorming session to see if they can come up with any other words
that are related to it in the same semantic Weld. You can also question
people to see if these new words have any extended senses.
A good Weldworker will also have his or her ears open all the time as people
are speaking in the daily round. People will often use words and expressions
that you have never heard before and these should be noted down
immediately. Whenever you record a new word in this way, though, you
should be sure to check it later either with somebody who was taking part in
the original conversation, or with somebody else. You will need to remember
that your own knowledge of the language is not that of a native speaker.
No matter how well you think you know the language, your knowledge
will still be far short of how well local people know it, and it is surprisingly
easy to mishear a word, or to misunderstand what somebody has said.
If you cannot independently verify a word that you think you heard in
conversation, then it is probably safer to assume that you misheard it.
4.3.4 Sound substitutions
Another way to look for possible new words would be to adopt the sound
substitution method. By this method, you could start with all of the words
that you have already recorded and invent a complete set of possible
minimal pairs involving all of the phonemic contrasts in the language.
Thus, if you have already recorded a word of the shape pali, you could ask
people if they know of the word pari, or pati, or padi, and so on. Of
course, other sounds could be substituted so you could also present
people with prompts such as bali, dali, and tali, as well as pili and puli,
and so it goes.
Some linguists have gone to the trouble of taking the various segments
in the phoneme inventory, as well as a statement of the phonotactics
which governs the possible shapes of words, to provide a printout of all of
the logically possible words in the language. It would then be necessary to
go carefully through this list of potential words with speakers of
a language to cross out all of the nonsense words, and then to assign
meanings to those forms which people recognize as genuine words in the
language. The problem with these last two approaches is that they
could easily become very boring, both for the linguist and for your
language-helpers. A bored language-helper is likely to be a distracted
language-helper, and this is when errors are likely to start creeping in.
Let me show you something of the nature of the problem with an
imaginary language. A language might have Wve vowels i, e, a, o, and u,
and Wfteen consonants p, t, k, b, d, g, m, n, n, f, s, h, l, and r, and it might
have the simplest phonotactic pattern possible, with syllables consisting
of any consonant followed by any single vowel, and words could be one,
two, or three syllables long. With these Wfteen consonants and Wve vowels,
there would be 75 possible monosyllabic words. There would be 5,625
possible disyllabic words, and there would be 421,875 possible trisyllabic
words. Since a dictionary that contains 25,000 entries would ordinarily be
considered a fairly substantial piece of work, this means that people
would be expected to eliminate over 400,000 nonsense words.
It is hard to imagine anybody having the patience to stick with such
a mind-numbingly boring task. If one person were to do nothing but cross
oV non-existing words from such a list after thirty seconds of thought for
each word, and that person worked for eight hours a day, Wve days a week,
the task would not be Wnished until just over one year and eight months
later . . . and that is assuming no time oV for annual leave, illness, or lunch
and coVee breaks! This is clearly not a wise path to follow.
4.3.5 When is the dictionary done?
At some stage, you are going to have to ask yourself when your dictionary
might be ‘complete’. Any decision about when to stop collecting new
words and to start the process of writing up the dictionary is necessarily
going to be somewhat arbitrary, as there will never be a time when you can
say that you have Wnally recorded the ‘last’ word in a language. Not only is
the vocabulary of every language absolutely enormous, but vocabularies
are constantly changing in any case. I have found that even in very
small languages with just a few hundred speakers, after an absence of
a year, new slang expressions can become established in the community.
You will be able to record your Wrst 300 or 400 words in just a day or
two. You should be able to reach your Wrst thousand words after a week
or two of eVort, in conjunction with progress that you have made on the
grammar and the phonology. In order to reach the 2,000-word stage, you
can probably expect to take several months and to have to put in a lot
more eVort than it took to reach the Wrst thousand words. Getting to
3,000 words may take you a year or so if you are also involved in the
writing of a grammar at the same time. Each additional thousand words
is going to be increasingly diYcult to discover.
There is no single correct answer to the question: How many words
should a dictionary contain? There are dictionaries of some local languages
which may contain up to 20,000 entries, but these dictionaries have often
beneWted from a century or more of lexicographical tradition, with the Wrst
eVorts having been much more modest in scope. If you are the Wrst person
to ever write a dictionary of a language and you are working on a grammar
of the language at the same time, it would be quite a respectable eVort
if you could manage a dictionary of between 5,000 and 10,000 words.
Further reading: Hale (2001: 95–8), Samarin (1967: 205–17), Vaux and Cooper
(1999: 37–49), Bouquiaux and Thomas (1992: 401–687).
4.4 Keeping track of data
Because the writing of a grammar and a dictionary is likely to be spread

out over several years, you are going to have to develop some kind of
system for making your information accessible, as well as for keeping it

safe. While you are in the Weld, your methods of data organization may
have to be slightly diVerent from what they would be at home where you
have an oYce with a computer and limitless supplies of stationery.
However, you will still need to be able to locate important examples in
the Weld without too much trouble, and you will need to be able to
recognize the diVerence between data which is reliable and data which
calls for further checking.
4.4.1 Daily records
Anthropological Weldwork tends to revolve around the maintenance of

quite elaborate diaries in which the details of each day are laid out for
future reference. A linguist clearly does not need to be encumbered by this
kind of requirement, though we do still need to keep a linguistically
informative record of every piece of information that we observe. I have
already referred to the notebooks that you will be working with in the
Weld. It is probably a good idea to have separate notebooks for elicited
data, for text transcriptions, and for lexical data. If you end up recording
data from more than one language, it would also be extremely wise to
have a separate set of notebooks for each language. Including data
from more than one language in a single notebook increases the possibil-
ity that you may accidentally ascribe forms from one language to
a diVerent language.
Whenever you have a working session, it is a good idea to indicate the
date, the place, and the name of the person who provided the data. You
should also indicate where that person comes from because there may be
dialect diVerences of which you are not yet aware. If you end up with
large amounts of data recorded from large numbers of people, you at
some point have to decide about the reliability of one piece of information
vis-à-vis some other piece of information. The date of recording may be
relevant, as older information may be less reliable than information that
you have gathered more recently. The speciWc person who provided the
information may also be relevant if you eventually come to realize that
certain people are more likely than others to get confused about certain
kinds of things that you ask for.
If you are eliciting data by having somebody translate from a
lingua franca into the local language, it is a good idea to prepare your
prompts fully ahead of time in your notebook. Be sure to leave plenty of

space in relation to each example so that you can clearly indicate whether
your language-helper found the sentence acceptable or not. Your language-
helper may also spontaneously oVer alternative ways of saying the same
thing. It is a good idea to record such information as well, so allow
space for this.
When I am transcribing texts, I typically transcribe with two or three
lines of free space beneath each line of text. This allows me space to
provide any kind of translation notes as I go on another line to the actual
text. I also transcribe onto only a single page of my notebook, leaving the
other page blank. This means that I can use that page for any kind of
further grammatical or lexical elicitation that might occur to me while
I am doing the transcription. Using diVerent pages clearly distinguishes
the elicited material from the transcribed material.
4.4.2 Filing systems
Some guides to Weldwork (e.g. Samarin 1967) advocate quite extensive

systems of data-Wling in the Weld so that individual examples of a wide
variety of structural patterns can be easily located. I have never managed
to summon the dedication or Wnd the opportunity to develop such an
elaborate system of card-Wling for grammatical data, so I would suggest
that unless you really cannot Wnd your way around your corpus, the
amount of work involved in putting such a system together in the Weld
might be better spent on more elicitation and more text transcription. It is
probably going to be a better idea to leave such detailed organization of
your data until you return home where you can make use of sophisticated
electronic systems of Wling such as Toolbox.
With regard to lexical data, there is a variety of ways in which infor-
mation can be recorded in the Weld. One of these ways could be called the
page-Wling method. Just as the name suggests, this is a method of keeping
track of information that involves putting words onto pages in a book. If
you can Wnd a solid notebook in which sections are marked by letters
of the alphabet, this would be a good addition to your Weldwork kitbag. If
you can’t Wnd such a notebook, you could of course always add your own
tags to show where you can Wnd words beginning with A, B, and so on.
There is a variety of ways in which you can organize your lexical data in
notebooks in the Weld. It might be a good idea if you keep all words
referring to trees together in a single place, and all words referring to

birds. You might also want to keep all of your verbs together in a single
place, perhaps even having separate places for your transitive verbs and
your intransitive verbs.
Within each of these groups you can organize your information alpha-
betically to make it easier for you to Wnd your way through these lists later
when you are looking for a particular word. Of course, when you start out
on such a notebook, you should leave plenty of lines between your early
entries to allow for subsequent entries to be placed in appropriate alpha-
betical order. It would also be a good idea to leave several blank pages
between each diVerent section of your notebook to allow for expansion in
the number of entries within each section.
When you have collected a large number of items, however, you are
likely to Wnd it diYcult to keep things in strict alphabetical order because
items are not ‘movable’ once you have entered them in your book. If you
were to write items in pencil, you could erase them and move them down
the page to make space for a newly inserted item above it. You could also
write entries in pen but use white-out to shift an item down the page.
However, this is not a good idea, especially if you end up using white-out
over white-out, because eventually this can crack and break oV and you
will lose all or part of whatever was written there.
While some of us may have wonderfully tidy handwriting, this is
certainly not true for all of us. With my handwriting, I sometimes Wnd
that written v and r look remarkably similar. I am unlikely to Wnd it too
much of a problem trying to work out whether a word should be van or ran
in English, as the context is likely to provide me with relevant clues. There
has therefore been relatively little motivation for me to tidy up this aspect
of my handwriting. However, when I am recording lexical data by hand in
the Weld, I have often been uncertain later as to whether I had intended to
write a v or an r. If you are entering lexical data by hand, you will need
to be aware of any potential confusion in your own handwriting and to
compensate for this in your notebook entries. I could compensate for
my own tendency to confuse r and v, for example, by writing such
words extremely carefully. Alternatively, if I still do not trust myself to
distinguish these letters, then I could perhaps write these words in capitals
rather than lower case, where this kind of confusion would not be possible.
Another way of organizing lexical data in the Weld is to use a card-Wling
system. By this method, each separate entry is written on a separate small
card. As you Wnd additional words, new words can be inserted in the
correct alphabetical order. These cards can be kept in a simple shoebox or
in a specially made cabinet of some kind, and the information that is
contained on each card can ultimately be transferred to a printed page.
Each individual card will be large enough that a number of diVerent kinds
of information can easily be included for each word. You can therefore
include information about word class membership, as well as examples of
sentences in which that word appears.
While your cards are probably best kept in alphabetical order, it is
relatively easy for you to pull out all of your words referring to trees, or all
of your intransitive verbs, if you wanted to do some kind of testing on
these subsets of words. You could make this easier by using cards of
diVerent colours for diVerent kinds of words: green cards for trees, pink
cards for animals and birds, blue cards for Wsh, and so on. Of course, once
you’ve pulled all of the cards out for a particular category, it is no major
problem to put them back into alphabetical order again.
Although card-Wling has its advantages, there are some disadvantages
too. The cards that you can buy in stationery stores under the names
index cards or system cards are more like thin cardboard than ordinary
paper. This has the advantage of making them durable, but it also makes
them relatively expensive to buy in large quantities. It also makes them
quite heavy, and therefore potentially expensive to transport, particularly
if you were anticipating the recording of 5,000 lexical entries on 5,000
diVerent cards.
Of course, you would also need something to keep your cards in. While
shoeboxes are good for keeping larger cards in, if you keep smaller cards
in a shoebox, you will Wnd that they will move around in the box and
they will easily get out of order. The smallest cards would therefore
probably need to be kept in a special drawer, rather like the old-fashioned
library catalogue drawers that we used to use before catalogue informa-
tion was fully digitized.
When I Wrst went to Paama, I took a large bunch of small index cards
with me and I kept them in a two-drawer metal box. The box itself was
not all that cheap, but that was not the major problem that I faced with it.
Being metal, it was also quite bulky and heavy, hence expensive and
inconvenient to transport. It certainly was not something that I could
carry with me to diVerent people’s houses when I wanted to record
new vocabulary.
Another disadvantage of a card Wle like this, in comparison to lexical

material that is stored in a notebook, is that you can drop a notebook,
pick it up again, and continue along your way after dusting oV the cover.
But I once had the misfortune to drop my card Wle. The drawers Xung
open and a couple of thousand cards all dropped out onto the ground. It
then took me several hours to collect all of the cards up and to make sure
that I had put them all back correctly in alphabetical order.
I kept that metal card box for many years out of a sense of sentimental
attachment, even after I had abandoned all of the cards with the Paamese
lexical information on them when I Wnally produced the published
version of the dictionary. It started to rust after many years in the tropics
and the drawers became hard to open. I eventually threw it out, and I have
never missed the horrible rusty grey monster.
4.4.3 Computer storage
When you have completed your Wrst Weldtrip and you have had an
opportunity to organize your data at home in preparation for your next
visit, one of your priorities should be to organize your lexical data. You
will of course have access to a computer, so all of the information in
your notebooks or on your cards should be carefully transferred to
a computer Wle.
You will then be able to produce a tidy printout of all of your lexical
data. When you do this, you should be sure to leave lots of white space,
and perhaps also to use double spacing. I typically then spiral-bind my
printout and this then becomes my working notebook to which I add my
new lexical data on my next Weldtrip. If you leave lots of white space in
your printout, you can insert new words by hand while you are in the
Weld. It is also a good idea not to copy the dictionary pages back to back.
If you have a white page opposite every page of printed text, this gives you
plenty of additional space on which to insert new material.
Then, when you get back home again, it will be a relatively simple
matter to insert any new material, or to correct any of your old material
on your computer Wle. Of course, if you are planning any further trips,
each updated printout of the lexicon can become the basis of your new
lexical notebook. Be sure, though, to include the date of each version on
the printout so that you can always tell which information represents an
update of which older information.
Some Weldworkers will be in a position to enter data directly onto

a computer while they are in the Weld. If you are fortunate enough to be
in this position, then by all means make use of the technology because
of the ease with which you will be able to access and organize your
information. However, you should keep in mind that there are some
risks in doing this. Some information is likely to be more reliable in
your mind than other information and there is a very real danger that
some fairly tentative information may end up indistinguishable in your
computer Wle from more deWnitive information. This means that you will
need to have some way of clearly marking which information needs
further checking and which information you are more certain about.
This could be as simple as remembering to key in a question mark after
information that needs checking.
In one dictionary that I am currently working on, for example, I
have entered words that I have only encountered in a published
anthropological source that is almost eighty years old in a diVerent Weld
of the database to distinguish these from entries that I have gathered myself.
There are other entries which derive from a wordlist compiled by another
linguist but which I know from experience is likely to contain many errors.
Where I have not yet been able to check these items, I have entered these
into another Weld. Finally, there are forms from my own data about which
I am uncertain for a variety of reasons which I have marked with a query
in a separate Weld. This means that I know that only data which appears in
a certain Weld of the database should be treated as fully reliable.
Further reading: Samarin (1967: 151–74), Austin (2006).
4.5 Archiving
A linguist should go beyond merely documenting a language, as the

archiving of any materials gathered is also of considerable importance.
While your Wnal grammar and dictionary will hopefully appear in pub-
lished form, it is likely that much of your textual data, as long as your
notes, will not be published as such. However, your text transcriptions,
and possibly even your Weld notes, may contain much valuable data that
other linguists in the future may be able to make excellent use of to discuss
issues that you have not addressed in your own accounts of the language.
For example, since I published my own grammar of Paamese over

twenty years ago (Crowley 1982), I have become aware of many aspects
of the language that I would treat diVerently today. There are also parts
of the grammar that I can now see were largely ignored in my earlier
description. In particular, the animacy hierarchy (human nouns >
non-human living things > inanimate nouns) is involved in a wide variety
of grammatical constructions. Thus, for example, whether a noun is or is
not marked for plurality is dependent on its position in the animacy
hierarchy, with highly animate nouns much more frequently overtly
marking plural and inanimate nouns seldom marking plural. I was
recently able to quantify these tendencies on the basis of observations
of the actual behaviour of nouns in my collection of Paamese texts. There
are possibly other features of Paamese grammar that linguists in the
future may want to investigate when I am no longer around, but since
my full collection of texts has been placed in the national archive in
Vanuatu, this should not be a problem.
The importance of archiving is perhaps best illustrated by what
happens when data is not properly stored. I have worked on one
moribund language in Vanuatu where the local chief informed me that
about twenty years earlier another linguist had done some of the same
kind of work on that language. He told me the name of that linguist.
I was involved in the preparation of a comprehensive publication of all
published sources relating to all of the languages in the country (Lynch
and Crowley 2001), and I never came across any publications in that
linguist’s name. In violation of the ethical requirement that the results of
linguistic Weldwork be provided back to the local community in some
locally understandable form, this linguist also appears not to have
supplied copies of anything that he wrote to local people. A search of
the national archive for cultural and linguistic materials suggests that this
linguist also failed to deposit copies of any of his notes or tapes in that
archive. If these potentially valuable materials are still in existence, we
have absolutely no idea where they are.
I have also heard too many stories of valuable records of little-known
languages, or even of languages which have become extinct, that were
once kept in an old oYce that burnt down or which was destroyed in
a cyclone. There are also stories of valuable linguistic records which have
simply been lost or misplaced while no copies of these notes had ever been
deposited in an archive for safe back-up.
When you make your data available for archiving, you need to keep in
mind the fact that some of the data you have recorded may need to be
treated in speciWc ways. Some of your audio recordings or your notes, for
example, may involve matters that were told in conWdence because they
identify particular individuals in ways that should not be made public.
Some information may be seen as belonging exclusively to some
particular lineage, or to some particular group in society. There may
also be information which only women or only men are permitted
to know.
Such restricted material should only be deposited in an archive if there
are strict provisions in place for determining levels of access to that data.
Archivists should be able to devise cataloguing systems which list
material as being of ‘open’ access, which anybody can listen to or read,
while other material may be listed as being ‘closed’. Such material may
then only be accessed with the direct permission of the researcher who
deposited the material, or of people holding particular positions in
the local community.
A Weld linguist need not become actively involved in the task of
archiving. Any detailed discussion of archiving, then, probably does not
belong in a Weldwork manual such as this. However, in this day and age,
the Weld linguist needs to appreciate that the task of archiving is likely to
be made easier if any audio or video data is recorded in a format that
facilitates digital rather than analogue archiving, as this kind of material
is less subject to erosion of quality over time. Such recordings call for
the acquisition of special recording equipment. Although this is more
expensive than traditional analogue recording equipment, linguists
should keep this in mind when submitting applications for funding
for Weld research. Indeed, some funding agencies now require a clear
statement from applicants about the type of equipment to be used, the
formats in which data will be captured, and where and how the data will
be archived.
However, analogue recordings (and even Weld notebooks where this is
relevant) should be copied and presented to an appropriate archive for
safekeeping. This serves not just to protect the interests of future
generations, but even individual researchers themselves. Notes and
tapes can be damaged in anybody’s oYce (or in the Weld) by any manner
of disaster, which could be averted by placing copies for safekeeping in
a professionally curated archive.
It is not just notes and unpublished materials which should be

presented for archiving. In many countries where research permits are
issued for linguistic research, linguists who publish materials arising
out of their research are also called upon to provide copies of their
publications in an appropriate library or archive. Even if you are aware
of no explicit requirement that you do this, linguists should consider it as
both common courtesy and an ethical obligation to return copies of all
oVprints of articles and chapters, as well as copies of all books, to local
authorities. Not only does this enable local people to see what you have
done with the data that you have been given access to, but it also makes
future research easier for subsequent researchers if all work from the
area that they plan to work in is held in a single location.
Further reading: Austin (2006). For more information on archiving see the Open
Language Archives Community (OLAC) homepage: www.language-archives.org.
Another useful source is the Digital Endangered Languages and Musics Archive
Network (DELAMAN) page: www.delaman.org
5
Beyond Elicitation
I have already indicated that elicitation can take you only so far in your
study of a language. In this chapter, I will be discussing some of the issues
that we face when trying to record natural speech.
5.1 The textual corpus
An account of any language needs to be based primarily on a substantial

corpus of continuous spontaneous speech. Of course, obtaining such
a corpus can present the Weldworker with a variety of problems.
5.1.1 Early corpora
Linguists in the past had no choice but to gather their textual corpora by
having speakers of the language dictate their stories while they were
transcribed ‘live’. Such a process is to be avoided these days, for
a variety of reasons. First, it has to be enormously taxing on the linguist.
Secondly, in the absence of any recording that can be played back later,
any mistranscriptions can never be checked. Thirdly, the manner in which
the story is told is almost certainly going to result in a style of speaking
that is diVerent in unpredictable ways from ordinary language usage given
what must seem like a fairly bizarre process for the language-helper.
Some earlier textual corpora were compiled instead by speakers of the
language writing the stories themselves. Franz Boas, for example, trained
speakers of some of the Amerindian languages that he worked on so that
they could write thousands of pages of oral tradition in their languages.
European missionaries developed a writing system for Maori in New
Zealand in the early 1800s and taught this to the local people so that
122 beyond elicitation
they could read the Bible that had just been translated into their language.
Maori rapidly became more literate in their own language in the Wrst half
of the nineteenth century than European settlers were literate in English.
Maori began publishing newspapers in their own language, writing letters
to each other, and they also wrote down a huge amount of oral tradition
and genealogical information, much of which has survived in manuscript
form to the present. Valuable as these kinds of linguistic corpora are,
however, it must be remembered that written texts often diVer from the
natural spoken language in a variety of ways.
5.1.2 The recording of speech
A modern grammar should therefore be based primarily on the spoken

form of the language. Before the mass production of reel-to-reel tape
recorders in the 1950s, it could be very diYcult to obtain this kind of data.
The only recording that we have of any of the Tasmanian languages is an
almost inaudible—and quite short—stretch of Fanny Cochrane-Smith’s
voice recorded onto an Edison cylinder in the late 1800s. These days, of
course, even reel-to-reel tape recorders have become museum pieces, and
cassette recorders are similarly becoming superseded technology so that
digital recorders are increasingly becoming standard equipment.
Before we even start looking at more speciWcally linguistic issues relating
to the recording of texts, there are some very basic technical issues that you
need to sort out Wrst. Before you even go to the Weld, you need to remember to
test your recording equipment to make sure that it actually works. If you are
out in the Weld a long way from anywhere and your recorder won’t go, you
may have only yourself to blame for the inconvenience.
These days, with the advent of digital recorders, there are often more
options on your recorder than you may be used to dealing with on
traditional tape recorders. This means that there might be quite diVerent
ways of recording, playing, and pausing. Make sure that you become well
and truly familiar with the operation of your digital recorder before you
go out into the Weld. People in the Weld who have agreed to tell a story for
you won’t be too impressed if you record their well-rehearsed narrative
text, only to Wnd that you hadn’t correctly put it into ‘record’ mode. This
sounds utterly basic, but even the most experienced of Weldworkers can
get things wrong. On my last Weldtrip to Malakula in Vanuatu, I tried
to record a story. I had the recorder in readiness in—I thought—‘pause’
beyond elicitation 123
mode, so that I could release the pause button to record the story. When
I went to play the story back, I discovered to my dismay that when I
thought I was releasing the pause button, I was in fact pausing it, and all
I recorded was my storyteller clearing her throat at the beginning and end
and I had missed the entire story.
Always keep an eye on your battery levels on your recorder. Losing
valuable data because your batteries run out halfway through a story can
be extremely frustrating. However, don’t assume that it is only your
recorder that has batteries that need checking. At the beginning of one of
my Weldtrips to Vanuatu recently, I recorded a story which somebody was
very keen to tell. When I went to play it back, it started out a little muZed,
and it became increasingly inaudible to the point where I could barely
make out anything by the end. I tried recording some more material and it
was similarly muZed. The playback mode seemed to be functioning Wne,
and the battery level meter was looking perfectly healthy. I was baZed.
In frustration, I bought a return air ticket to the capital worth about
£100, and spent three nights in a hotel—none of which had been budgeted
for in my research proposal—so I could go in search of a new tape
recorder that worked. In the end, all that I could Wnd that was portable
and battery operated was a cheap and nasty little dictaphone with an
in-built microphone. I took it back into the Weld and recorded quite a lot
of additional material on the new machine. It worked, though the sound
quality was far from brilliant. Since I was recording a moribund language
with only about Wfteen living speakers, that was a major disappointment.
However, when I eventually returned home and was putting my
original equipment back into the cupboard, I noticed a piece of paper
that fell out the box that the microphone came in. There was a picture of
a little battery inside the handle. It couldn’t be so, I thought. But it was.
I had been using the microphone for possibly seven or eight years without
realizing that the microphone was gradually losing its oomph, and it
eventually gave out on me on that trip to Malakula. I didn’t need to
spend all that money and waste all that time looking for a second-rate
tape recorder. I could have Wxed my problems with just the cheapest
of batteries that I could have bought out in the Weld.
Given the number of potential disasters that can befall the Weldworker
when recording texts, I always like to replay a story in full immediately
after I have recorded it. This serves two functions. First, it reassures me
that the recording itself is of good enough quality to be transcribed.
Secondly, it provides an immediate element of feedback to the person

who just told the story. Many people may never have heard their voice
on tape before, so immediate playback gives a clear message to the
community of what your project is about.
While you don’t have to be a studio technician in order to produce
valuable recordings in a language, there are some basic points to which
you should pay attention. Try to record indoors if you can, to avoid the
problem of wind blowing into the microphone. Even a light breeze that
you may be totally unaware of can leave you with a muZed recording
without your realizing it. Alternatively, you could bring a windshield that
slips over the microphone to keep from recording the wind noise. If you
are recording indoors, buildings made of traditional materials such as
wood, bamboo, and thatch are usually better than buildings with concrete
or galvanized iron, because you pick up less echo.
Try to record at a time when there are relatively few people outside.
The sound of children playing in the distance is easily picked up on a high-
quality microphone, so it is a good idea to record when children are at
school if this is possible. I’ve made the mistake of recording when some-
body in an adjacent building was grating yams so the story comes with
a constant ktttthhhh ktttthhhh ktttthhhh ktttthhhh throughout. And you
may also need subtly to ask people not to pick nervously at the table or to
strum their Wngers while they are talking. I once asked somebody else to
record a story on my behalf, and the storyteller was scrunching a plastic
bag in his hands as he talked. When I went to transcribe the story, it
sounded like he was talking through a hailstorm because the microphone
was set down very close to where the scrunching was taking place.
I normally introduce any story with a brief comment about who I am
talking to, the name of the place that we are in, and the date. This could
be in English if you are thinking of some future archivist, but I have
generally given the introductions in my Vanuatu recordings in Bislama,
the national language, so that my storyteller will feel less uncomfortable
with the process. I then indicate to the storyteller that the microphone is
his or hers so that the story can begin.
5.1.3 What to record
When you go to record your Wrst stories, you should try to get somebody
to tell you a story which is fairly short. Ideally, your Wrst story should also
relate to something that you know something about yourself. Perhaps

you could ask somebody to tell a story about the day that you Wrst arrived
in the Weld location, or you could ask him/her to give a simple set of
instructions about how to light a Wre. The Wrst story I ever recorded in
Paamese was a set of instructions about how to prepare a yam. It was
a very short and particularly unexciting text that went like this:
The yams are dug up. They are brought back. They are grated. They are
washed clean. They are thrown into a saucepan on the Wre. When they are
cooked, the water is poured away. The coconut is grated and it is squeezed.
They are eaten.
This text was certainly not a story of earth-shattering importance, but it

was just what I needed at that early stage in my Weldwork.
Recording a story about something that you already know something
about is going to be helpful to you in the early stages because the more
that you and the storyteller share by way of background knowledge, the
easier it is going to be for you to understand what the story is about. Your
Wrst story is going to be diYcult enough to transcribe even in the most
favourable of circumstances, so you will want to ensure that there are
a minimal number of ‘surprises’. When you eventually do get to the stage
of recording longer stories about more arcane topics, you will from time
to time Wnd yourself confused about who is saying what to whom, or what
is the signiWcance of such-and-such an event, and this is a confusion that it
would be best to avoid in the initial stages of text recording if you can
possibly manage it.
But at the same time, you should reassure your language-helper that
they should feel free to say anything at all that they want to say. If they
get anxious about ‘linguistic purity’, you should stress that they should
not feel uncomfortable about mixing words from other languages if that
comes naturally to them. Usually, the moment people become conscious
of any pressure to speak a language in some unnaturally ‘pure’ form, they
become hesitant and lose Xuency. Linguists should be in the business
of describing, not prescribing, in any case.
You can reassure somebody who feels in any way uncomfortable
because they have produced what they feel to be an ‘unsatisfactory’
story because of the presence of borrowed words by telling them that
the story can always be corrected in the written version that you are going
to produce afterwards. Of course, if you are going to make this promise to
people, you have an ethical obligation to keep that promise. It would be

most unfair publicly to distribute a version of a story that the storyteller
felt was in some way inadequate where it directly reXected what he or she
had actually said.
Once you have become familiar with the basic patterns of the language,
and after you have recorded a few texts, you will Wnd that the process of
transcription gradually becomes easier and the task will be less time-
consuming. At this point, you can start asking for stories about any
kind of topic, and stories of any length. You can ask people to tell you
stories about anything that they are interested in talking about. People
will often concentrate on traditional themes, preferring to tell stories
about animals and people in the past that have been passed on from
generation to generation.
Stories do not have to relate to traditional themes in order to present
linguistically interesting features. I have found that people often have
stories about a range of lifetime experiences that they are keen to tell,
including stories around any of the following themes:
. the arrival of the Wrst missionaries and the establishment of the

Christian church, and conXict between Christian and traditional
practices;
. dealings with the Wrst non-missionary Europeans, such as labour
recruiters or government agents;
. events associated with the presence of the US military in Vanuatu
during the Second World War;
. a major disaster such as an earthquake or a cyclone;
. facing death when lost at sea or when seriously injured somewhere;
. a major trip, such as somebody’s Wrst trip overseas.
Sometimes, younger people who have been educated in school may oVer
to tell a story that hails from another island, or which is clearly some kind
of hybrid between a traditional story and an introduced story from
outside because of references to non-indigenous animals such as cats.
Once you have made a recording, you need clearly to name the
resulting Wle so that you know what is on it, using a consistent naming
convention with no unusual characters. If you are using a tape recorder or
minidisc (which I am not advocating!) you should make sure that you
identically label both the cover of the tape or minidisc as well as the tape
itself. You would be surprised how easy it is to label only the covers, only
to discover at some later stage that the unlabelled contents are all spread
out on the table in front of you and you no longer know what is on any of
them!
Since you are going to be interested in any kind of talk, it does not
matter in principle what kind of talk it is. Five minutes of gossip and
scandalous lies can be just as grammatically and lexically informative as
Wve minutes of a traditional story about the creation of the world. Of
course, it would be very unwise to insist on recording anything that you
knew to be scandalous and untrue because other people in the community
may be outraged by this.
You are probably not likely to face this kind of situation anyway, but
something a bit like this may be an issue. When I Wrst started recording
stories in Paamese, I would arrive with my tape recorder and ask some-
body if he or she would be willing to contribute a story to my collection.
Most people expressed interest in the idea but some were reluctant
because they said that they didn’t know the full story. I remember that
my response at the time was something like, ‘That doesn’t matter, just tell
me what you know.’ Generally, people were not persuaded by my attempt
at reassurance. I can see now that it was proper that they should not have
allowed their incomplete knowledge to be recorded because other people
hearing or reading their story could easily have criticized them (and me)
for getting things only half right. Now, if somebody says that they do not
know a story well enough to tell it, I never encourage them to tell half
a story. And if somebody says that they need to check the details of a story
with somebody else before I record it, I am happy for them to do that.
5.1.4 Reasons for recording
Stories have the potential to provide a variety of diVerent kinds of

valuable new information. In particular, stories are used for the following
purposes:
(i) Grammatical expansion. You cannot expect that elicitation from

English will cover all possible grammatical patterns, and it is almost
certain that new patterns are going to arise in spontaneous speech.
When I had still recorded relatively little textual data on the
Naman language of Malakula, I had the impression that serial
verb constructions of the type He cut split the log were unusually
rare in this language in comparison to other languages to which Naman
was closely related. However, when I had substantially increased

my textual corpus, I came to realize that my initial suspicion was
unjustiWed and that this language had serial verb constructions that
are as productive as those of any other nearby language.
Of course, any new constructions which appear in textual data can be

supplemented by additional elicitation. Elicited translations from English
or some other language may point to the existence of a separate category
of past tense on verbs in a language. With careful elicitation, you can
complete the full paradigms for the past tense and you may think that you
have done all there is to do. However, once you start recording stories,
you may Wnd that past tense meanings are occasionally expressed by quite
diVerent forms. Further investigation may reveal that the original past
tense paradigm that you recorded only related to the immediate past, and
there is, in fact, a completely separate paradigm for the distant past.
However, your textual attestations of this new paradigm will possibly
not provide you with forms for the full paradigm. You would therefore
need to supplement the data from your text with further elicited data in
order to complete this new paradigm.
(ii) Discourse patterns. Many linguists are satisWed simply to produce

an account of the phonology, morphology, and syntax of the
language. However, it is becoming increasingly important these
days also to attempt to describe the discourse patterns of
a language, which show how sentences are organized into longer
utterances. This is something that cannot be studied simply by
taking translations of sentences from English, as it is highly likely
that the resulting discourse structures will simply mirror those of
English. The only way to get genuine discourse patterns is to get
somebody to produce spontaneous speech in the language.
(iii) Recording oral tradition. I discussed the need for the Weldworker to
provide feedback to the community in Chapter 2. One possible
beneWt to a community of your research is the recording of oral
tradition in written form. As a linguist, you probably would not
expect to be able to record a complete set of oral traditions, but
there are cases where the recordings of a linguist have become
the most important cultural documentation for some people where
traditional patterns for the transmission of oral tradition had been
completely disrupted.
5.1.5 How large a corpus?
I have referred to the recording of texts in the plural, as it is essential that

you record both a substantial amount of material, as well as a varied
range of people: young and old, male and female, and perhaps educated
and uneducated, urban and rural. It is a good idea to record stories that
are both traditional and modern in content so that you can see how new
cultural and technological concepts are expressed in spontaneous speech,
in contrast, perhaps, to the more prescriptive judgements that you may be
oVered in direct elicitation sessions.
I have not, however, tried to give any kind of Wgure for how large
a corpus of texts you should aim to produce. For somebody who is
working on a moribund language, it may be diYcult to obtain more
than a few short recordings of continuous speech. My own study
of Bandjalang, for example, was based overwhelmingly on elicited
data because my language-helper felt she did not have the verbal conW-
dence in the language to provide more than a few short texts.
While even a little bit of textual data is far better than none, the greater
your amount of textual data, the more reliable your Wnal grammar is
going to be, and your dictionary is also going to be so much the richer as
well. Combined with well-directed elicitation, about an hour’s worth
of spontaneous speech can provide enough information for a moderately
detailed grammar. My Erromangan grammar was based on about
six hours of speech which amounts to several hundred pages of careful
transcriptions. Some linguists, however, have assembled written
corpora amounting to several thousand pages of transcribed text. While
such large corpora clearly oVer wonderful opportunities for the statistical
study of variability, I would question the need for such a large corpus
if your goal is to produce a dissertation-length account of the grammar of
a language.
It is less important to oVer any speciWc guidelines about the size of
a corpus than to stress the need to aim for some kind of ‘equilibrium’ level
(Samarin 1967: 68–71). By this, I mean that you should aim to record
enough texts that signiWcant new grammatical patterns seldom show up
in new texts, and any instances of variability have settled down at
reasonably stable proportions with each new text. Once you have reached
this stage, you can legitimately shift your attention from the gathering of
texts to the detailed analysis of the texts that you have already recorded.
5.1.6 DiVerent genres
I have so far been talking about the recording of just a single style of
speaking, i.e. narrative monologues. These are relatively easy for the linguist
to transcribe and translate, so they represent a good way to widen your
structural and lexical horizons. However, once you have acquired a good
selection of narrative texts, you may want to consider widening your
horizons by recording other kinds of speech.
You may, for example, want to consider turning your attention to the
recording of conversations. Of course, this can be much more diYcult to
achieve. If you have a group of people speaking together, it is almost
inevitable that some people are going to end up speaking at the same time.
People are likely to laugh or overlay the speech with other kinds of noise
that can at times make transcription extremely diYcult. Even when
people happen to be taking turns at speaking with clear gaps between
each speaker, so much of the meaning in a conversation is taken from the
surrounding context that as soon as you have decontextualized the speech
in an audio recording, much of what is said may make little sense even to
your native-speaker language-helper.
There is also the problem of setting up a workable recording situation
for a conversation. For ethical reasons, you cannot record people’s con-
versations when they are not aware that this is being done, so you need to
set up your recording equipment in such a way that its presence is obvious,
while at the same time not being too obtrusive. You also have to make sure
that your equipment is placed in a position where it can clearly capture
the contributions of everybody in the conversational group. Finally, of
course, there is the problem of how natural people’s speech will be in
this kind of situation. Arguably, one of the biggest conversation-killers
would be the instruction: ‘Speak naturally while I record you speaking.’
There are other kinds of situations where it would be legitimate to
record people speaking in order to gain access to a wider range of speech
styles. You may be able to record sermons or similar kinds of utterances
with the permission of the people involved. You may also be able to
record meetings while they are taking place. This may involve various
kinds of formal speeches which you would not be able to hear when
recording ordinary stories. Of course, you may face diYculties in Wnding
a suitable place for your microphone so that you are able successfully to
record the contributions of everybody involved. Also, the venue may well
be one which does not allow for good quality recordings. You would
therefore need to decide how valuable this kind of data would be for your
project if you are going to work to overcome the sometimes extensive
technical issues involved.
A certain amount of the communicative content of a conversation or
people speaking in a meeting is likely to be non-verbal as well as strictly
linguistic. In order properly to understand some of what you have recorded,
it may be necessary to see some of the various non-verbal cues such as
facial expressions, gestures, or body position. Arguably, then, successful
conversational analysis or interpersonal exchange can only be carried out
by means of video recordings along with audio recordings. While this kind
of data can be enormously informative, it clearly requires a lot more
specialized recording skills and much more expensive equipment. And, of
course, it makes the recording process so much more intrusive.
Further reading: Chelliah (2001), Samarin (1967: 75–84).
5.2 Getting it right
I was somewhat reluctant at Wrst to admit this, but I have not always got
things right in my Weld notes. We linguists like to believe that we are
a pretty clever lot and that especially in basic things like phonetic tran-
scription, we are facing a straightforward task. However, I Wnd myself
immensely comforted by the following words of Samarin (1967: 22), a Weld
linguist with a considerable amount of experience:
There is probably no linguistic researcher alive, nor will there ever be, who can
write down at the Wrst hearing everything which is linguistically signiWcant.
He then kindly goes on to list some of my past errors, as well as the errors
about which I have felt so smug when I have discovered them in the work
of others:
The errors can be of several types: not hearing enough phonetic diVerences,
hearing diVerences in the wrong places, wrongly segmenting the stream
of sound into phonological units, and so on.
Of course, we have to sort out the inaccurate material from the (hopefully
correct) bulk of our notes before we produce our Wnal write-up so that
our descriptive account is an accurate reXection of the language as it is
used by its speakers.
5.2.1 Evolving analyses
In many cases, somebody who records a little-known language is going to be

the only person for a very long time who records anything of that language. I
would guess that for a long time, no linguist will want to go and write a new
grammar of Paamese in Vanuatu. I would not want to argue that my
grammar of the language is perfect or totally complete. However, there
are so many other languages that are completely undescribed, that there
would not be much point in somebody writing another grammar of Paa-
mese, unless it were to oVer a completely diVerent perspective. This means
that whatever I have said about Paamese enters the published literature as
the only source of information on that language. Of course, if you are
working on a moribund language, you may well be the only person ever to
describe the language in print, as further work may be impossible once the
current generation of speakers has gone. This means that you have to be
very sure of the accuracy of the information that you present.
As I indicated above, there are some aspects of the grammar of
Paamese that I originally published about which I have since changed
my mind. I was careful enough to ensure that the actual facts that I was
dealing with were correct, but there were some points on which
my analysis of the facts has since changed. For example, there was
a category of what I originally referred to as verbal adjuncts, i.e. a kind
of adverbial, that I eventually came to realize were a kind of serial verb
construction. I am talking here about constructions such as the following,
in which ‘spoke follow’ form a serial verb construction:
Na-selūs usil meatin tāi.

I-spoke follow person one
‘I spoke about someone.’
5.2.2 Native-speaker error
You must be prepared for the possibility that even a very knowledgeable
person with a native-speaker competence in a language can make
a mistake while speaking. This means that you cannot automatically
assume that because somebody says something, it must be right. People
sometimes make slips of the tongue, choosing incorrect words, mispro-
nouncing words, or misusing grammatical constructions for all sorts of
reasons. If, when listening to a recording, somebody insists that some
kind of linguistic error was made, and you have no evidence that this
represents systematic behaviour, then you should not incorporate
this into your Wnal description.
I have occasionally been trapped in conversations at parties with
non-linguists who, when they have found out what I do for a living,
suggest that the kinds of descriptions that I have written are unreliable
because of the possibility that I may have been deliberately misled by
speakers of these languages. I will ignore the objectionable presupposi-
tions here about there being an honest and reliable ‘us’ and a dishonest
and scheming ‘them’ and will deal more directly with the question of
whether it is possible for somebody deliberately to fool a linguist into
producing an inaccurate description.
There are certainly cases where, it is argued, observers have been
deliberately fooled into recording questionable data, though this seems
to be more likely to aVect anthropologists than linguists. Margaret Mead,
for example, when she was conducting anthropological Weldwork in
Samoa in the 1920s, administered questionnaires to adolescent girls to
Wnd out what they were up to sexually. Some of those girls—who are now
old women—have recently claimed that they made some information up
at that time, partly because they felt that was what Margaret Mead had
wanted them to tell her.
In linguistics, it would be possible for somebody deliberately to mislead
you in the area of vocabulary. Of course, it would be diYcult to imagine
what beneWt there might be for somebody in doing this, which reduces the
likelihood of it happening. One possible scenario, though, might be where
particular vocabulary that is associated with secret rites may be withheld
from you (Samarin 1967: 28–9). However, it is almost unimaginable
that a linguist could be deliberately misled in grammatical matters.
To fool even a fairly inexperienced linguist about linguistic structures in
a language would be similar to a speaker of English deciding to fool
a learner into thinking that the past tense of verbs was marked by
the suYx -ick rather than -ed. Just think how diYcult it would be to
consistently say walkick instead of walked. Such a deception would be
almost impossible to maintain for any length of time.
While deliberately misleading a linguistic Weldworker is likely to be
rare, there are times when speakers unintentionally provide inaccurate
information. Some people in a society are likely to be more reliable than
others for providing certain kinds of information. Particularly when it
comes to lexical information, for example, younger speakers of a language

often do not have the same detailed kind of knowledge that older speakers
have, so it is best not to rely too much on younger speakers for the
collection of information that requires any kind of specialist or traditional
knowledge. It is also possible that somebody who is only partly sure about
a word’s form or meaning may present their information with greater
certainty than his or her knowledge warrants, as a way of showing
how ‘knowledgeable’ that person is. For this reason, it is important to
cross-check lexical information with other speakers of the language, rather
than relying on information from just a single speaker.
Speakers of a language can give you wrong information for other sorts
of reasons as well. Your command of the intermediary language may well
not be anything like as good as you think it is and it may be that your
question was unclear or ambiguous. If, on any occasion, you receive
a response which goes against your expectation, you should at least
consider the possibility that your interviewing technique may be at fault.
It is also possible for a speaker of a language to adjust his or her way of
speaking in the direction of yours. If you have not yet mastered the details
of the local language—and this is likely to be the case, otherwise, why else
would you be there?—then your language-helper may subconsciously be
moving in the direction of your own semi-correct grammatical patterns.
In my experience, however, native speakers of languages are much
more open about the limitations on their knowledge and they often tell
me which information they are sure of and which information I need to
check with somebody else with more specialized knowledge. A careful
linguist needs to make note of reservations such as this and to do the
appropriate cross-checking. It is very easy to forget to add a question
mark to a particular lexical entry in your notes and for this form then to
make its way into your permanent lexical record. The error in this case
would be yours, and not due to your language-helper.
5.2.3 Fieldworker error
A careful Weldworker will not just question the legitimacy of data that
comes from his or her language-helpers, but will also be constantly
questioning the reliability of his or her own representation of that data,
as well as his or her interpretation of what has been recorded. It is not
uncommon for one’s early data to contain various kinds of errors.
Sometimes data doesn’t make much sense because you have broken
a single word into two separate words, neither of which makes any sense.
Even if you have managed to work out the phoneme inventory, it is often
possible to mistranscribe individual words containing sounds that you
would never have imagined you would Wnd any diYculty with, and this
can also lead you to analyse elements of the grammar incorrectly.
The kinds of errors that I have just mentioned are situations where the
speaker of the language provides accurate information, but it is not
correctly represented or analysed by the linguist.
It is always advisable to write your notes in bound notebooks rather than
on individual sheets of paper. Writing on sheets of paper that can be torn
out of a writing tablet is not a good idea because once they become
detached from the rest of your notes, it can often be diYcult to work out
if they represent earlier or later notes. You should always clearly indicate at
the beginning of each day’s writing what the date is, the name of the person
whose speech you are writing, and where that person comes from. The
advantage of including the date is that it is often helpful to know which
form of a transcription represents an earlier—and perhaps less reliable—
representation and which is the more reliable later representation. While it
may have been perfectly clear at the time that you were writing who you
were speaking to and where that person came from, it may well be far
less clear who you were talking to and where that person was from when
you are back in your oYce and there is nobody who you can check
things with.
It is surprisingly easy for the linguist to get the meaning of a word
wrong. This means that a lot of what you record, especially in the early
stages, needs to be cross-checked with other people, or with the same
person, but much later on in the analysis. One linguist devoted
a considerable of eVort to recording basic wordlists of about 200 items
from nearly 200 diVerent speakers, each coming from a diVerent place.
That means the linguist asked the question ‘How do you say X in your
language?’ about 40,000 times. From time to time, the language-helper
and the linguist were clearly speaking at cross purposes, and incorrect
information ended up being published. On more than one occasion, in
response to the question ‘What is your word for X?’, the person replied
with the local equivalent of ‘Um . . . ’, and the linguist dutifully wrote
down the relevant hesitation form as if it were the actual word.
This then ended up in the permanent record for those languages.
One common way in which a linguist can make a mistake in the

dictionary deWnition of a word is to assume when somebody gives you
an example of what a word can refer to, that this represents the actual
meaning. This means that your entry might end up being semantically far
too narrow. For example, my early notes on Paamese include the word
romarom for which I gave the translation equivalent edible fresh leaf.
Later on in my notes, I recorded the same word, but this time my notes
gave two translation equivalents: tip of something, end of a story.
When I Wrst recorded this word, somebody must have been explaining
to me that the fresh tips of the leaves of a certain plant were edible, and
they must have used the word romarom to refer to these. I evidently did
not realize at the time that romarom does not have to refer just to the tips
of leaves, as it can refer to the tip of anything at all, and it can even refer
to the end of a story. Fortunately, I discovered my initial error before my
dictionary was published in this case, but I would have to admit that I
cannot guarantee that I have avoided the same kind of problem with
every word in every dictionary that I have ever published.
The easiest way for a grammatical error to be incorporated into a Wnal
description would be for a linguist to rely too heavily on what may in fact be
his or her own clumsy attempts to elicit grammaticality judgements from
native speakers. I have already warned about the diYculty of getting
reliable judgements from people by simply asking them the question ‘Can
you say X or not?’ It can be very tempting to base a particular piece of
linguistic analysis on a single response to a question of this type from a single
speaker when other speakers may give rather diVerent sorts of responses.
The variability in responses does not necessarily mean that there are distinct
varieties involved here. The reality is that grammaticality judgements some-
times need to be expressed as matters of degree rather than as an absolute
distinction between grammatical and ungrammatical. Any attempt to treat
the data in this way may, therefore, represent an error on the part of the
linguist rather than an error on the part of your language-helper.
5.2.4 The problem of tunnel vision
Linguistics is a science, and being a scientist means that we arrive at

our conclusions on the basis of empirical observations. Empirical
observations, of course, are based on objective reality, without any element
of subjective judgement being involved. As scientiWc linguists, we would
like to believe that our Weld notes are completely free of judgements and
that they contain nothing but objective reality. While we may readily jump
to the conclusion that some other linguist’s notes may be inaccurate, we
sometimes seem to be less willing to say the same about our own notes.
On the basis of my own experience, however, I am not sure that we can
guarantee that our own notes are invariably perfect. We need to recognize
that for at least some of us, our perceptions of what we are hearing may
sometimes be inXuenced by what we were expecting to hear in the Wrst
place. The linguistic Weldworker needs to be constantly struggling against
the tunnel vision that often comes with whatever happens to be our
current hypothesis and which seems to guide our perceptions in the
direction of Wnding further support for that hypothesis, even if it should
later turn out to have been wrong.
In the Naman language of Malakula that I am currently working on,
there is a phonemically contrastive schwa, along with Wve peripheral
vowels, giving a vowel inventory of /i e a @ o u/. While I have very good
reasons for arguing that schwa represents a distinct phoneme in this
language, there are some words in which I have found it diYcult to decide
if a particular vowel is schwa, or if it is instead one of the other vowels.
For instance, in the word for ‘inside’, the phonetic transcriptions in my
notes contain both [mb@ª@t] and [mbuªut].
At certain periods during my work on this language, I have leaned
towards [@] being the correct vowel in this word with the forms with [u]
representing ‘transcription errors’, while at other times, I have convinced
myself that [u] is correct and that I had been mishearing this vowel as
schwas. Depending on which stage of my Weldwork I was at, I rather
suspect that I tended to transcribe the word for ‘inside’ with whichever
vowel I was favouring at the time. In saying this, I am eVectively
admitting that I may not be the world’s best phonetician. However,
I am prepared to admit this because I am fairly conWdent that I am
probably no worse a phonetician than most other Weld linguists.
5.3 Text transcription
The corpus of texts that you have recorded are going to be useful to you
only if they have been transcribed. If you are going to produce these in
some kind of written form for the beneWt of the community, the Wrst step
in doing this is also to produce a transcription of the original text.
If battery power for a laptop is not a problem for you in the Weld, I have
found that the easiest way of all to transcribe is to download your audio
Wles from your digital recorder onto the laptop. You will then be able to
block the wave form for a section of the text and replay precisely the same
stretch of speech as many times as you need. If you are careful enough,
you can even zero in on a particular word that you are not sure about and
play it as many times as you and your language-helper need in order to
decide on the correct transcription. If you are going to transcribe directly
from your recording device, then you are going to have to develop a Wne
touch on the rewind button in order to transcribe material eYciently.
When you Wrst start transcribing recorded texts, the process of
transcription will probably be a painfully slow process, for both you
and your language-helper, even if the story is fairly simple and short. It
is impossible to provide a strict formula for how much transcription time
you will need per minute of spoken text, as this will depend on all sorts of
considerations such as how familiar you are with the language, how
familiar your language-helper is with the process of transcription, how
clearly articulated the text is, how good the quality of the recording is,
how many disturbances there are while you are transcribing, and so on.
As a rough estimate, however, you can probably expect to spend
something like twenty times the length of the original text on your initial
rough transcription with the help of a native-speaker language-helper.
That is, in order to transcribe a Wve-minute story, you are going to need to
sit down with somebody for nearly two hours of transcription time. I
always check a transcription at least once afterwards by myself, which
may take that much time again. Any further checks tend to be somewhat
quicker, as each additional time that the transcription is checked, the
number of potential problem areas is hopefully going to be reduced.
You will need to transcribe exactly what was said on tape, as well as its
meaning. You will Wnd that some language-helpers are better at giving
you this kind of help than others. Some people, for instance, may want to
rephrase what has just been said, on the basis that it was not properly said
the Wrst time around. Other people may not want to repeat what was
recorded because it contains factual errors of some kind, and they may
insist on correcting this. After all, people get details wrong all the time
when they are speaking, saying things like 1979 when they mean 1989,
saying John when they mean Harry, saying uphill when they mean
upriver, and so on. However, you should stress that, while you want to
get an accurate record of what appears on tape, you ask people clearly to
indicate all corrections which should appear in any version of the story
that is going to be made public.
Another problem that you can face when transcribing texts is that some
people take what they hear on tape as an invitation to say something else
that follows on from this, and it is almost impossible to get them simply to
repeat exactly what was actually said. In one language that has only half
a dozen elderly speakers left—Ura, spoken in Vanuatu—I once recorded
a story from an old lady who is regarded as the best speaker of the
language. Unfortunately, when the time came to transcribe and translate
her story, she took each section that I played back on the tape as an
opportunity to produce yet more language, which I obviously could not
record because I was already playing back something else that was
recorded on tape.
Sometimes, it can be useful to work with somebody other than the
original storyteller to transcribe a story. If the narrator is fairly old and
has unclear articulation, you might Wnd it useful to get a younger person
to help you with the transcription process. Not only is the younger
person likely to have clearer articulation, but you may Wnd it easier to
persuade this person not to launch into a related monologue in response
to hearing a selection of text.
So far, I have been suggesting that diYculties in transcription can be
put down to problems associated with your language-helper, but it is
important to realize that you will often be a major part of the problem
yourself. I have already indicated that we often develop a kind of analyt-
ical tunnel vision that leads us to represent what we hear to some extent in
terms of what we expect to hear. When transcribing texts, we are quite
likely to represent word boundaries, and even the shapes of individual
morphemes, according to the particular ways we are analysing the
language at the time of transcription.
No matter how careful I think I am being with my transcriptions, from
the very Wrst text to the very last, for every language that I have ever
studied in the Weld, I have had to re-transcribe my earliest texts in the light
of new analyses that have come to light by the time I got to my later texts.
Not infrequently, new material that comes to light in these re-transcribed
early texts then leads to new ways of thinking about some of the material
in the later texts and those transcriptions then need to be modiWed. You
can probably expect to be transcribing and re-transcribing your texts
until you get to the Wnal stages of your linguistic analysis and write-up.
Because each of your texts is likely to go through various diVerent forms
during your Weldwork, it is always a good idea to date each particular
version in case you pick up an old version of a text and you are no longer
sure how it Wts into your current analysis.
Some linguists prefer to reduce their own role in the process of text
transcription by arranging for a language-helper to do this for them. This
can be diYcult where there is no written tradition in the language in
question, as you will either have to teach the person to use (and become
conWdent in) an orthographic system that you may have to devise
yourself, or you may even have to teach your language-helper to
transcribe texts in the IPA. However, where there is a local tradition of
literacy, this can be a good solution. Not only will it free you up to do
other things if you have a language-helper produce your transcriptions,
but those transcriptions will also be produced by somebody who has
native-speaker intuitions about the language.
However, there are possible pitfalls that you would need to watch out
for. If there is a local tradition of literacy, this may also be accompanied
by a set of somewhat prescriptive attitudes towards the language.
If a speaker happens to use a form that diVers somewhat from the way
that a word is traditionally written, your language-helper may automat-
ically write the word as it is habitually spelt rather than how it was
actually pronounced. It will still be necessary, therefore, for the Weld
linguist to cross-check transcriptions.
As a Wnal warning, let me caution against succumbing to any tempta-
tion that you may have to take an untranscribed text back home with you
with the intention of transcribing it yourself later. Remember, the reason
that you have recorded textual data is so that you can encounter new
vocabulary and new patterns, so if you do not have access to somebody
you can ask about a text, you will not be able to Wnd out what is going on.
I have occasionally tried doing a transcription by myself if my
language-helper was not available that day, my intention being that I
would be as prepared as I could for when he was available. I would
typically Wnd that the Wrst few lines would be no problem at all. There
would then perhaps be a single word that I could not clearly make out, or
perhaps a word that I had not heard before. That would reduce the
amount of sense that I could make of that whole sentence, and if

understanding part of the next sentence depended on a good understand-
ing of the previous sentence, I would rapidly get completely lost. Yet the
same text, when my language-helper was there to help me the next day,
would miraculously become absolutely clear.
5.4 Participant observation
The notion of ‘participant observation’ is one that is widely used in

ethnographic research in the discipline of anthropology. If scholars
want to know how people from a particular culture organize their lives
and what they believe, there are essentially two contrasting ways of
getting access to this information.
5.4.1 Extracting and observing language
One approach to anthropological data-gathering involves the anthro-

pologist living alongside, rather than with, the people they are studying,
and asking them a lot of questions. Margaret Mead’s work in Samoa was
similar to this approach. When she was in Samoa in the 1920s, she lived
with an American doctor and she administered questionnaires to people
about their behaviour. She was interested particularly in the behaviour of
adolescents, and some of her questions involved very intimate details
about the sex lives of adolescent girls. In recent years, however, scholars
have questioned the validity of some of her central Wndings, arguing that
her methodology was potentially Xawed, as people do not necessarily
respond honestly to direct questions of this sort, especially on topics
such as this one that are regarded as sensitive. I have already mentioned
the recent claims of some of the women who were interviewed by
Margaret Mead that they used to make up some of the things that they
said when they answered her questions.
Participant observation is a contrasting approach to the gathering of
anthropological data, which involves living with the people who you are
studying, and doing the same kinds of things they do, and waiting for
things to happen, and simply observing them when they do happen.
Somebody wanting to study what people do in Vanuatu when they
drink kava, for example, would not do this by asking people questions
about what they do when they drink kava. Rather, you would need to
go along with people when they are drinking kava and observe what
happens.
Anthropologists tend to recognize diVerent kinds of participant
observation, ranging from passive participation to complete participation.
Passive participation requires you to be physically within the community,
but as far as possible to act as a bystander to events so that you can
eVectively become an overhearer to what happens. This means that in any
situation, you need to organize yourself so that you occupy a position
physically and socially unobtrusive within a particular context, but which
at the same time gives you maximum exposure to the data that you are
aiming to record.
It depends on the kind of data you are trying to record how unobtru-
sive you would need to be. If you are simply trying to Wnd out what the
local word for dog is, or if you are trying to Wnd out how the paradigm for
the present tense of verbs is marked, you clearly won’t need to be in a blind
spot at all, and you can ask people quite direct questions. But there are
certain kinds of observations for which it is much more essential to
be unobtrusive. Studying the discourse patterns within interpersonal
conversations, for example, would make it much more essential for the
Weldworker to be as close to invisible as possible.
Being unobtrusive can mean a number of things. At the simplest level,
it can mean being physically as hard to see as possible. If you wanted to
record a formal meeting of some kind, it would deWnitely not be a good
idea to be moving around with a hand-held microphone from speaker to
speaker. If you are going to be doing this kind of recording, you would
need to have microphones set up ahead of time in the right places so
that you can be discreetly sitting oV to the side somewhere, perhaps
monitoring the sound levels on the recorder.
Part of the same requirement is the need to make people from whom
you are recording narrative text as comfortable as possible with your
recorder. You have to remember that while you are going to be very
familiar with your equipment, it may look extremely sophisticated, and
even intimidating, to somebody who has not seen it before, particularly if
you have a large and bulbous hand-held microphone that you have to
thrust in front of somebody’s face. I have found myself making
a recording in which somebody was audibly nervous, and he was
constantly Wdgeting with his hands while he was speaking. The resulting
story had a far greater proportion of false starts and other dysXuencies
than I had found with other speakers of the same language. In retrospect,
perhaps I should have allowed the speaker a trial run, perhaps wiping the
initial version of the story if the narrator was happier with a second (or
third) attempt when he felt more comfortable with the recording situation.
Being unobtrusive may also mean that you should be as quiet as
possible. This means that when you are recording talk in a meeting, you
should avoid turning the pages of your notebook, or Wddling more than is
absolutely necessary with your recorder. This kind of behaviour may
eVectively bring you out of your carefully chosen unobtrusive position,
thereby aVecting the data that you are trying to record in some
unpredictable way. When I am lecturing to a classroom full of university
students, I Wnd that it sometimes takes only a single distraction to
throw me oV my stride—somebody’s mobile phone rings, two people
have a whispered exchange—and your Wddling may have a much greater
impact on the proceedings that you are recording than you might imagine.
Completely passive participation in a situation is often—perhaps even
always—impossible. If you are an outsider, you can never be completely
invisible, no matter how much you might convince yourself that you have
become truly ‘accepted’ by members of the community you are working
with. No matter how frequently you have used your recorder, it is still
something that people can see, and which is diVerent from what happens
in their day-to-day lives. Some level of active participation is therefore
unavoidable in most cases. Arguably, the more ‘foreign’ your appearance,
behaviour, and language, the more diYcult it is going to be for you ever to
blend in, so there is a danger that all of your observations might be
contaminated to some extent by your presence.
What we are really interested in as linguists is empirical data,
i.e. data that is based on observed fact. It is axiomatic that there are no
single-style speakers of any language. This means that everybody’s speech
is going to vary in at least some ways according to the non-linguistic
context of the utterance. People will typically use varieties that are
considered to be more ‘formal’ in situations where strangers are
present while they will use more ‘informal’ varieties with family and
friends. People are also less likely to use formal varieties when speaking
in groups with other people who they are socially close with—this is
peer group pressure in action—even if a stranger might happen to be
present.
All of these considerations lead us into what is often called the Obser-
ver’s Paradox. The aim of linguistic research in a community is to Wnd out
how people speak when they are not being systematically observed, yet
the only way we can obtain that data is by systematically observing them.
Very few people can speak ‘naturally’ when they have been asked to
‘speak naturally’, especially when there is a microphone and a recorder
in front of them, and there is a total stranger from some university sitting
there pressing the record button.
The Observer’s Paradox comes into play most dramatically in
situations where there is a major power asymmetry in the social groupings
of the person whose speech is being studied and that of the person
conducting the interview. For instance, if one ethnic group in a society
is seen historically as being an ‘oppressor’ and another group as being
‘oppressed’, the presence of a member of the dominant group interview-
ing a member of the subordinate group is particularly likely to bring the
Observer’s Paradox into play.
Many of the conclusions that were reached about the linguistic and
cognitive skills of ethnic minority children in the United States in the
1950s, for instance, were based on data that was severely tainted in this
way by the eVects of the Observer’s Paradox. Groups of Black children in the
USA were often judged to be linguistically deWcient after having had their
language skills assessed in formal interview situations with, for the most
part, adult White interviewers, who were not previously known to the
children. Hardly surprisingly in such situations, children’s linguistic re-
sponses tended to be as monosyllabic and as non-committal as possible.
Such responses then tended to be interpreted as a sign of linguistic deWciency.
William Labov’s pioneering work in recording natural speech in the
1960s clearly demonstrated the methodological problems associated with
some of these earlier studies. He referred to one record of an interview
with a young Black boy in New York City that was used as a basis
for conclusions that children like him were linguistically deWcient.
The boy went into a room where there was a friendly White interviewer.
The interviewer showed the boy a Wre engine, and then said (Labov
1972: 184–5):
Tell me everything you can about this.
To begin with, the child responded with twelve seconds of silence. The
rest of the interview went like this:
Interviewer Child
What would you say it looks like? 8 seconds. A spaceship.
Hmmm. 13 seconds. Like a je-et. 12
seconds. Like a plane. 20 seconds.
What colour is it? Orange. 2 seconds. An’ whi-ite.
2 seconds. An’ green. 6 seconds.
An’ what could you use it for? 8 seconds. A je-et. 6 seconds.
If you had two of them, what would 6 seconds. Give one to somebody.
you do with them?
Hmmm. Who do you think would 10 seconds. Cla-rence.
like to have it?
Mm . . . Where could you think we At the store.
could get another one of these?
This interview illustrates the brief and ‘linguistically deWcient’ responses
that a lot of investigators had reported in studies of Black children’s
speech in the United States.
Obviously, though, the child here was faced with an asymmetric testing
situation. This child is quite possibly suspicious that if he does or says
something wrong, something bad might be going to happen to him. From
his experience in school to date, it is quite possible that he had already
developed a number of what we might call survival strategies, one of
which is to avoid saying anything at all. That is, maybe it is better to say
nothing at all than to say the wrong thing.
But diVerent sorts of testing situations with children of the same age,
educational level, and socio-economic and ethnic background produced
quite diVerent results. Labov organized tests which incorporated the
following features:
. The interviewer brought along potato chips, to make the interview

more like a party.
. Children were interviewed in the company of one of their friends.
. The interviewer sat on the Xoor instead of a chair, to reduce the
height imbalance.
. The interviewer and the children were matched for ethnicity.
. The interviewer introduced rude words into the discussion.
What these tests showed was that in these kinds of testing situations,
children were actively competing to talk. They would interrupt each
other, and their contributions were anything but monosyllabic. Here is an

extract from one of these interviews (Labov 1972: 188–9):
Interviewer Children
Is there anybody who says Yee-ah!
your momma drink pee? Yup!
And your father eat doo-doo
for breakfas’!
Ooh! [laughs] And they say your father - your father
eat doo-doo for dinner!
When they sound on me,
I say CBM.
What that mean? Congo booger-snatch! [laughs]
Congo booger-snatcher!
[laughs]
And sometimes I’ll curse
with BB.
What that? Black boy?
Oh that’s a MBB.
MBB. What’s that? ’Merican Black Boy.
Ohh.
In order to overcome this problem of the Observer’s Paradox, we

eVectively have to divert people’s attention away from the fact that they
are being formally interviewed. Labov’s work in the 1960s demonstrated
that these so-called linguistically deWcient children, when approached
in a testing situation in which they felt less threatened, proved in
fact to be highly verbal. It was in his famous article ‘The Logic of
Non-Standard English’ that the Linguistic DeWcit Hypothesis was con-
vincingly rebuVed. He was able to do this by reducing the degree of
asymmetry between the children and the people doing the testing, and
the data that he recorded showed that these children were in fact as verbal
as anybody else, though their language did possess certain systematic
structural characteristics that other children’s English did not possess.
While formal interviews and recording sessions are not necessarily
going to produce the most natural patterns of speech, the eVect of
recording sessions in the kinds of descriptive work that many of us intend
to carry out on previously undocumented languages is clearly not going
to be the same as we have seen as a result of Labov’s work. For one thing,
you will almost certainly be recording adults rather than children.
For another thing, you are going to be very clearly the learner, and you
are not going to be seen as somebody who is administering any kind of
a threatening test.
5.4.2 Overhearing in the Weld
One way of gaining access to natural speech is simply to observe speech

that is going on around you. Of course, this is something that you are
almost certainly not going to be able to begin with when your own
understanding of the language is likely to be minimal. Arguably, in
order to be able successfully to ‘overhear’ a language of which you have
no previous knowledge could take many months, possibly even years, of
exposure. Let me talk about the kinds of diYculties that you might face.
My data on the Erromangan language of Vanuatu contains both /s/
and /h/, which appear to contrast phonemically, as speakers of the
language have provided me with minimal pairs such as nmas ‘big’ and
nmah ‘death’. However, while people claimed that this was a minimal
pair, I soon found that in ordinary speech, the word for ‘big’ is sometimes
pronounced as nmah rather than nmas. While the word for ‘death’
is always pronounced as nmah, when people write the language, they
generally—but not always—represent it as nmas rather than nmah.
This kind of variation between s and h is widespread in the language,
and does not apply just to this pair of words. In fact, almost any word
containing either of these sounds can be found alternating with the other
sound, though some words seem to be more frequently found with s,
while other words are more frequently found with h. At one stage,
I decided to listen carefully for these two alternating pronunciations to
see if I could quantify people’s preferences. For four days, I noted every
instance of the word meaning ‘big’ that I heard around me to Wnd out how
often people said nmas and how often they said nmah. Although I could
have chosen any number of other words, I decided that this was a good
word to investigate because it is a high-frequency item. What I found was
that 59.7 per cent of the time, people produced nmah, while in the
remaining 40.3 per cent of the time, they said nmas.
Of course, those Wgures are only as reliable as my ability to recognize
instances of this word in daily conversation around me. I had spent many
months on Erromango, but my speaking ability in the language was never

developed very far. It is a language that exhibits greater morphological
complexity than any other language I have ever worked on. I was also
only able to spend three months at a time on the island, so although my
cumulative period of residence was more than a year in total, every
time that I felt I was about to make some kind of conversational
breakthrough, I had to leave and start almost from scratch on the next
trip. This means that even with the most careful listening for those four
days, I may easily have missed a substantial number of instances of
people pronouncing either nmah or nmas.
In principle, of course, if I were ever going fully to document this
language, I would need to study the distribution of s and h in every
word in the language. Given that it took me four days of doing nothing
else but noting down people’s pronunciations of the word for ‘big’, this
would be a huge operation. While frequently attested items such as
the word for ‘big’ would not present such serious diYculties, with
infrequently attested items, we face real problems in assembling
a representative sample of speech to work out what the appropriate
statistical percentages might be.
For example, in Erromangan, I recorded the word ulehei ‘variety of
ginger’, which meant that it could be entered into my dictionary Wles in
that shape. I subsequently recorded the word from someone else spoken
in a story, where it appeared instead as ulesei. In my two instances of this
word, I now have a 50/50 distribution between the s and h pronunciations.
Clearly, two instances of a form could hardly be described as a
representative sample, so I would need to record the same word from
a variety of speakers in such a way that people are not conscious of the
kind of linguistic information that I am trying to obtain. If I were simply
to wait for people spontaneously to produce this word in ordinary con-
versation, it could easily take me years before I got enough examples that
the Wgures would be reliable. The alternative might be to test people more
directly, but it would be diYcult to contrive a natural enough testing
situation that would elicit a statistically representative set of occurrences
of this word.
Even if I could have done this for this particular word, it would not be
practical for the remaining hundreds of words that I have recorded with
variable s and h in the language. Given that I was attempting at the
time to describe a little-known language for an audience of language
typologists and comparative linguists, there were, quite frankly, more

interesting things that I would rather have spent my time on. And this
is a dilemma that many Weld linguists may end up facing.
5.4.3 Beyond linguistics
We can’t do everything all by ourselves of course, and this is especially the

case when we come to recording specialist information, as in plant names
and uses, musicology, animals, celestial activity, and so on. Cooperation
with specialists in these areas can be extremely productive for all
concerned and is to be encouraged, regardless of the pressure from our
institutions and funding bodies to work as single researchers.
6
Problems and Pitfalls
I have tried to present linguistic Weldwork so far as something of an

exciting challenge. However, it would be unfair to suggest that Weldwork
is always going to be easy, and that if you face a particular problem, all
you have to do is look up a Weldwork guide for how to solve it. Life is
often so much more complicated than that. In this chapter, I want to
make you aware of some of the challenges that you may face.
6.1 Analysis in the field
I have deliberately avoided writing this volume as a guide to how to

approach the task of carrying out linguistic analysis and of writing up
your Wnal grammar, or the compilation of a dictionary. You will Wnd
that these can be tasks of enjoyment tempered not infrequently with
frustration. However, these kinds of issues lie well and truly outside the
scope of this handbook. Linguistic analysis and lexicography are skills that
you should have already acquired during your various descriptive linguis-
tics courses before you ever set out on your Weldwork. In any case, there are
plenty of well-written volumes already available on the diVerent kinds
of linguistic patterns that you can expect to encounter in diVerent
kinds of languages, e.g. Shopen (1985a, 1985b, 1985c), Whaley (1997),
Payne (1997), as well as manuals for lexicographers, e.g. Landau (1989).
In a guide to linguistic Weldwork such as this, however, it is relevant to
talk about some of the issues that relate to carrying out linguistic analysis
while you are in the Weld. While you are doing this, you can expect to face
particular problems which are diVerent from the issues that you may have
to deal with when you are doing your linguistic analysis back in the
comfort of your oYce at home. You might be tempted to think of your
152 problems and pitfalls
Weld location simply as a data-gathering site, leaving the entire task of

analysis for when you get back home. However, you will need to be doing
at least some kind of linguistic analysis as you go.
I have already indicated that a Weld researcher needs to be well
prepared for each session of Weldwork with a language-helper. This
means that you should come prepared with some kinds of expectations
about what you would like to investigate in that particular session. If you
are planning on investigating the marking of objects on verbs, for
example, you should at the very least come prepared with a list of
transitive verbs around which you can construct examples. If you want
to investigate certain aspects of the morphophonemics of verbal inXec-
tion, you will need to have some kind of preliminary idea about what
sorts of allomorphic variants there are in the data that you have already
gathered, and what sorts of environments each variant appears in. You
therefore need to have checked the behaviour of diVerent types of verbs
with a range of diVerent inXectional aYxes in order to come to any kind
of tentative conclusion that you want to check in an elicitation session.
Linguists seem to vary quite a lot with respect to the amount of further
linguistic analysis that they do in the Weld. I have often envied those who
seem to come back from the Weld with a set of detailed ideas about the
structure of the language they are working on. It turns out that many such
people were living in a hotel or some other kind of private guesthouse
when they were in the Weld and they then ‘commuted’ to where they
are going to meet their language-helper. People who are working in
these kinds of circumstances are likely to have the privacy and working
conditions that such conditions entail, and possibly also access to
electricity and a good working surface.
In my early work in northern New South Wales, for example, I was
able to return after a recording session in the local community to the hotel
room that I was staying in nearby. There, I could carefully replay my
recordings and come up with tidy Wnal transcriptions before preparing for
the next day’s elicitation session. The language that I was working on at
the time was moribund and I was working with just a single person.
Living within the community would, therefore, not have provided me
with any exposure to additional data from ordinary spoken conversation,
as the language of the community was English. This meant that there
was only advantage and no disadvantage to me in living outside
the community in terms of carrying out my linguistic analysis. On just
problems and pitfalls 153
one of my Weldtrips to that community, I did make arrangements to stay

in the community. While I enjoyed the greater amount of social
interaction that I had with people who I was getting to know as friends
by that stage, I deWnitely missed the opportunities to work that my rather
more convenient hotel room oVered.
In many Weld sites, comfortable hotel rooms or other private lodgings
outside the community are not always available. In any case, there are
often very good reasons why it is preferable to live within the community
rather than ‘alongside’ the community when conducting Weldwork. If as
a Weld researcher you Wnd yourself living in the community, you eventu-
ally become something like a participant observer, doing many of the
same kinds of things that other people do. This means that you will Wnd
yourself, depending on your own abilities and interests, helping people to
plant their food gardens, bring in the Wrewood, cook meals, go Wshing or
hunting, look after children, weave mats, or build a house. If you are
doing any of these things with other people, you are going to be in
a position to gather much richer and more varied linguistic data. You
will therefore be hearing people using the language with each other all the
time and in a varied range of situations.
However, the more successfully you make the transition into partici-
pant observation while living within the community, the more diYcult it
is likely to be for you to carry out detailed analysis of your data in the
Weld. One of the reasons for this is that people from diVerent cultures may
have quite diVerent sorts of expectations from you about the need for
privacy. You may be carrying on your linguistic analysis when you are
sitting down by yourself with a pen in your hand, whereas members of the
local community may see you as bored or lonely, a problem that can be
easily remedied by somebody coming and sitting with you, and perhaps
even engaging you in conversation.
Your notion of ‘work’ and theirs may also be quite diVerent. For local
people, work is perhaps more likely to involve sweaty physical travail.
While you may think of yourself as ‘working’ while you are sitting by
yourself in order to think about the details of verbal morphophonemics
or interrogative constructions, your hosts may well regard your apparent
sloth as an invitation to engage in recreational talk.
Even where people do not directly interrupt you while you are working
in this way, your activities may attract unwanted attention. On a recent
Weldtrip to Vanuatu, I took with me for the Wrst time a laptop computer,
which allowed me to download recorded texts. I found it much more

convenient to transcribe texts from the laptop, as I could easily block
a stretch of text and repeat it as often as I needed. However, many of the
people in the village that I was staying in had never had the opportunity
to watch somebody working on a computer so I would often end up
attracting an audience. People would generally recognize that I was
working and they would try to be quiet. But they were watching, inces-
santly, over my shoulder. And they would occasionally bump the table.
Some people can work when there is a crowd of people watching like
this, but it was driving me absolutely nuts. If it’s a crowd of children, you
can generally solve the table-bumping problem by being a little bit Werce
and they will often scatter. But you cannot do that with adults unless you
want to alienate people. You simply have to accept that your work is
going to be of interest to some people who are passing by. If you Wnd this
as distracting as I do, this means that quality time to conduct analysis in
the Weld can be somewhat limited. You may simply have to learn to suVer
in silence and continue working. Alternatively, you may be able to
turn your disadvantage into an advantage by calling on people in the
assembled multitude to help with certain points by asking questions such
as ‘What was it that he said just there in the recording?’, ‘What does
that word mean?’, ‘Could you also say such-and-such?’, and so on.
Alternatively, you may have to be prepared to abandon your plans to
work as soon as an audience gathers and do something else.
Fieldwork, therefore, often seems to present us with something
of an unfortunate choice between good data access with restricted
opportunities for analysis on the one hand, and reduced data access
with improved opportunities for analysis on the other.
Further readings: Gil (2001), Rice (2001).
6.2 Learning to speak the language
Probably every Weldworker would love to be able to acquire a good

speaking knowledge of the language that they are working on. Some
experienced Weldworkers advise that it is only through learning to speak
a language that you can fully analyse it (Everett 2001). It would be
interesting to know how many grammars of languages are written by
linguists who do acquire a speaking ability in the language that they
are studying. There are some legendary Wgures in linguistics who have
shown enormous Xair for learning languages very quickly in the Weld,
including Ken Hale, Kenneth Pike, and Stephen Wurm.
6.2.1 Reasons for trying
There are several reasons why it is a good idea to learn to speak the
language that you are working on if you can. First, you will be in
a position to broaden your range of structures and vocabulary through
exposure to spontaneous conversations that are taking place around
you all the time. Secondly, you will be able to engage in monolingual
elicitation sessions rather than simply presenting prompts in
a lingua franca for translation into the local language. And Wnally, you
stand to have a much more enjoyable time mixing with people. Through
local people’s appreciation of your knowledge of their language, you will
no doubt improve your own motivation.
One does not need to be able to speak a language in order to analyse it.
I indicate in the preface to this volume that I have published accounts—or
am currently working on accounts—of as many as eighteen diVerent
languages, though I never tried to carry out even a basic conversation
in most of these. Many of the descriptions that I have produced involve
moribund languages, so I faced few natural opportunities for conversa-
tion in the language in any case.
However, while I was working on my doctoral dissertation on
Paamese, I found that after about six months on the island I was
increasingly able to live monolingually in Paamese, and by the time that
I left after twelve months, I was resorting very little to Bislama, the local
lingua franca. I Wgured that I had made a real advance when I fell asleep
while some people were playing cards nearby. I woke up because I was
feeling cold and my card-playing friends poked fun at me because they
said I had been talking in my sleep. (I learned a new expression that
day: selūs ramobong ‘talk in one’s sleep’.)
Since I had never known myself to be a sleep-talker, I asked, ‘What did
I say?’ They replied, An gat vāreinaus. This literally means something like
‘Cold is just a lot biting me’, which is a normal way of saying in Paamese
‘I am really cold’. I asked, ‘What language did I say it in?’ They replied,
‘You were speaking in Paamese’. That surprised me, but what surprised
me most was the form of my message, so I asked, ‘Did I say An gat
vāreinaus (a construction which I didn’t actually know about yet) or

An gatinau vārei? ’. This is a simpler and more English-like construction
that I already knew about, which literally means ‘Cold is biting me a lot.’
They conWrmed that I had used the construction that was not con-
sciously familiar to me, but which I had presumably internalized from
conversations that I had heard around me, but had not yet dealt with
in my direct elicitation sessions.
6.2.2 Reasons for success or failure
The authors of most published grammars, however, are surprisingly

reticent about their own conversational abilities in the language in
question. Perhaps this is because they are being modest, but my suspicion
is that many are somewhat reluctant to let it be known too publicly that
they did not manage to become very conversational in the language.
I have certainly not been able to match my own success in learning
Paamese in some of the other Weld situations that I have been involved
in. While I was working on the island of Erromango, for example, I did
get to the stage where I could say things to people in the local language,
but any kind of extended spontaneous verbal exchange was beyond me.
I suspect that there are several reasons for my relative success in
learning Paamese and my relative lack of success in learning the language
of Erromango, and these factors may be relevant for other Weld linguists
as well. First, having published a grammar of both languages, I would
argue that the Paamese language exhibits fewer structural complexities
than Erromangan. Of all of the languages that I have ever described,
Erromangan is the most complex, with verbs especially exhibiting
considerable morphotactic complexity, as well as involving a substantial
amount of morphophonemic variation in the shapes of many of the
aYxes, including a fair amount of unpredictability and irregularity.
The relative ease of Paamese and the relative complexity of Erromangan
is matched by the vastly diVering extents to which outsiders have acquired
a speaking knowledge of the two languages. There are many non-Paamese
from other parts of Vanuatu who have learned to speak Paamese, while
there are very, very few non-Erromangans who have ever learned to speak
Erromangan.
Second, the size of the Paamese-speaking community in comparison
to the Erromangan-speaking community may be a factor. There are
probably more than three times as many speakers of Paamese as of

Erromangan. Large numbers of speakers of Paamese also live in the
capital, while there are relatively few Erromangans living in town. This
meant that whenever I went to town, I was able by and large to continue
to live in a Paamese-speaking milieu, while this would be almost
impossible in the case of Erromangan. Paamese households in town
would often be exclusively Paamese-speaking, but households involving
Erromangans would very often be linguistically mixed, resulting in much
greater use of Bislama, the lingua franca.
Third, I have detected something of a diVerence in outlook between
Paamese and Erromangans in regard to their languages. While both
groups of people are proud of their languages, the Paamese almost
seem to be aggressively proud of theirs. I never had to ask people on
Paama to dispense with Bislama and to speak to me in Paamese. They
initiated the switch themselves. They would constantly say things to me in
Paamese. Only if I didn’t understand would they then repeat it in Bislama.
Eventually, when I understood enough in Paamese and even began
replying in Paamese, people simply refused to speak to me in Bislama.
The Erromangans, however, never demonstrated any similar kind of
linguistic chauvinism.
But my success in learning Paamese and my failure with Erromangan
was not due just to these three factors. The way in which my time in the
Weld was organized was also a factor. While I spent a total of just over
twelve months living on both Paama and Erromango, my Weldtrips to
Paama involved two visits of six months’ duration, while my time on
Erromango involved four visits of three months each. It wasn’t until I had
spent six months on Paama that I felt conWdent enough to carry out
spontaneous conversations in the language. By the end of my shorter
stays on Erromango, I would typically feel that I was about to make
a breakthrough in terms of my speaking ability, but then I would leave.
When I came back, I would Wnd that I was more or less back to
square one.
DiVerent individuals are presumably going to have diVerent kinds of
language learning abilities and are going to require, therefore, diVerent
periods of exposure before they can become conversational. For me, it
seems that six months of continuous exposure was enough to acquire
a knowledge of Paamese. Samarin (1967: 54) seems to think that this
period should be enough for most, though I have been envious of others
in the past who have managed this after three months, and I have found
myself feeling smug about other people’s linguistic inabilities after twelve
months. The only way you are going to know what your own abilities are
is to go to the Weld and Wnd out.
The factors that I have just discussed are unlikely to be the only factors
that are involved in acquiring a speaking knowledge of the language of
your Weld site. Another possibility to consider is the extent of the role that
the local lingua franca already plays within the linguistic repertoire of the
local community. If the lingua franca is well established as a part of
the local linguistic scene for certain communal functions, then it may be
more diYcult for members of that community to wean themselves oV
a reliance on the lingua franca in their dealings with you.
One of the reasons that I chose to work on Paama in the Wrst place was
the fact that this was a relatively small island on which no other language
was spoken. My rationale was that the Paamese people would be unlikely
to use Bislama with each other, except perhaps when an outsider was
around. My prediction turned out to be correct, and when I was on
Paama, the only outsider was in fact myself. This meant that I was
exposed to almost nothing but Paamese from the outset.
Before I had decided on Paama, I originally considered a number
of other Weld sites. I gave some thought at one stage to working on
Malakula. I decided against this because there are at least two dozen
actively spoken languages on this island, as well as another dozen or so
languages that are now moribund. I guessed that with such diversity on
the island, Bislama would play a much greater role within the various
local communities. I am now involved in a long-term project on
Malakula and can conWrm that this is very much the case.
I have spent a total of about twelve months living in the village of
Vinmavis. While the primary language of this village is very clearly
Neve‘ei, Bislama also has an important role. Because this language is
spoken in just a single village with a population of about 500, many men
have married women from other language groups. In some cases, these
women have learned to speak Neve‘ei, but many of the younger wives use
Bislama with their husbands and their children. Because these women
become fully integrated into the community, on most communal
occasions, Bislama has become the dominant language. Thus, community
meetings and church services are almost invariably conducted in
Vinmavis in Bislama rather than the local language. One of the church
congregations in Vinmavis has also recently set up a private school which

caters to children from a wide area of northern Malakula. Students who
attend this school all board with local families, but because these children
all speak diVerent local languages, Bislama also comes to be a household
language even where the husband and the wife both come from Vinmavis
and speak Neve‘ei.
6.2.3 Individual diVerences
Some people are likely to be more successful than others, partly because
of the circumstances in which you Wnd yourself, and partly because of
diVerent individual aptitudes. You will therefore need to become com-
fortable with whatever level of success you can achieve. However, you
almost certainly would not need to become a proWcient conversationalist
in the language in order to impress people in the local community. For
many small local languages, few outsiders will ever have acquired
a speaking ability in the language, so people will not necessarily expect
you to manage this anyway. You can possibly expect great tolerance of
your eVorts to speak the language as long as it is clear that you are trying.
Although there is no single formula by which anybody can successfully
learn a language, there are some tricks to help make thing easier. If you
have an aYnity with young children, they can often be your best informal
language teachers. Young children are more likely to be monolingual, so
they will not Wnd it strange to speak to you in their language, and they will
not have acquired the kinds of social graces that prevent them from
correcting you when you have made a mistake. I remember once walking
barefoot through a Paamese village with a young boy and I hurt myself
when a sharp stone poked into the underside of my foot. I cried out in
pain and the boy asked me in Paamese what had happened. I responded,
Ahat tā dilı̄n hēk ‘A stone poked my foot’. My young interlocutor
perfectly straightforwardly said that the proper way to say this was
Ahat tā sal hēk, which literally means ‘A stone speared my foot.’ I went
Ah to indicate that I had got his point. However, this was not enough
for him and the little boy said, ‘OK, say it then.’ So he forced me to repeat
the sentence with the correct choice of verb. I have never forgotten
that lesson.
Adolescents and adults, by way of contrast, may have quite diVerent
attitudes to the use of the local language. They will typically have learned
a lingua franca and may be reluctant to use the local language with you,
Wnding it easier to stick with using the lingua franca. Somebody who has
been to school may also want to interact with you in English in order to
practise their English, while you are looking for people with whom you
can practise your knowledge of the local language. An adult speaker of
Paamese would probably have simply accepted what I said when the
stone hurt my foot. This is because my meaning was no doubt obvious,
in spite of the apparently odd way in which I framed it.
I have already referred to my lack of conversational conWdence in
Erromangan. During some of my time on the island, there was an
American Peace Corps volunteer working in the same village that I was
staying in. I had heard her speaking Bislama, in which she demonstrated
adequate proWciency, though I would not have described it as
a sophisticated command of the language. However, when I visited
other villages, people would speak in raptures about this woman’s ability
to speak the local language. I was struggling quite badly with the
language at the time and I have to admit that I was rather envious. I felt
this particularly strongly since I was there to do nothing but language
work, while the Peace Corps volunteer was sitting down with somebody
for a few hours at weekends when she was not otherwise fully occupied
with teaching in school.
My envy, it turned out, was somewhat misdirected. The American
approached me at one stage for help in language matters and it became
apparent that she had not even been able to work out how some of the
most basic of noun phrase structures worked. It was clear that she had
made little progress in working out the grammar, and she could not
possibly have been genuinely ‘speaking’ the language. But she had
certainly done the right thing in learning the main greetings and linguistic
formulae and she used these regularly. Not only this, but she had also
learned a small number of basic vocabulary items. Her ability to say just
good morning and goodbye, while also knowing that a nup was a yam,
impressed local people hugely.
Conversational Xuency might be too high a goal to set for many people,
while an ability just to express greetings is surely too low a goal. Most
people should probably aim for something between these two extremes.
Unfortunately, there is no magic bullet when it comes to actually
acquiring a speaking knowledge of any language and this is something
that can only come with hard work and with constant practice. In the case
of a previously undescribed language, you will obviously not be able to

buy a set of tapes and language lessons before you go to the Weld and you
will just have to learn on the job as best you can. You are simply going to
have to get used to asking people how to say things.
You will then have to try using the patterns that you have uncovered
with other people. Of course, you may have to develop a thick skin, as
you are almost certainly going to make mistakes, and sometimes people
are likely to Wnd at least some of your mistakes uproariously funny. One
of the reasons that young children typically make much better language
learners than adults is that children have fewer inhibitions, and they are
more willing to learn from their own mistakes. An extroverted adult is
therefore likely to make a better language learner than an introvert. If you
are more introverted, you can of course still succeed at language learning,
but it will possibly take you longer to realize that you can get your
message across if you force yourself to give it a go.
However, I have already indicated that local people will often be
remarkably generous in their judgement of your linguistic abilities. I
don’t know how often I have heard Paamese people say that I speak
Paamese ‘better than the Paamese’. This is absolute nonsense, of course,
but this story seems to have started some years ago when I was able to
show people that I had learned to count up to twenty (and beyond) in
Paamese. Since most younger speakers of the language can only count up
to Wve in the local language and beyond that they use numerals from the
lingua franca, Bislama, my counting ability seemed impressive. In reality,
my conversational abilities in Paamese were probably never worth
much more than a Bþ, but a little reverence of this kind can certainly
spur you on.
6.3 Going troppo
I have talked about a wide range of issues and problems relating to the
conduct of Weldwork. Fieldwork is almost certainly going to be a major
learning experience for you as a Weldworker. Much of the impact of
Weldwork on the individual is positive, in that you gain exposure to
a new culture, to a new language, and to new friends. At the same time,
however, you cannot expect everything about Weldwork to be positive.
Before my Wrst Weldtrip to Paama in 1976, I was already experienced

enough in the practical side of Weldwork from my previous work on
Australian languages that I did not anticipate any major problems
adapting to my new situation. Admittedly, I was going to have to learn
a new language in order to carry out my Weldwork. However, before
arriving on Paama, I spent about a month in Vila doing some basic lexical
and structural elicitation with Paamese speakers there, and I had even
recorded some short texts in the language which I had transcribed.
I used this month to acquire a basic conversational competence in
Bislama. By the time that I arrived on Paama I felt that I was ‘set to
go’, and I was really looking forward to the experience.
I fairly quickly found a place to stay, and found that people in the
village that I had settled in were very friendly and welcoming, inviting me
to all social functions that they themselves attended. I did not aim to do
any serious linguistics for a while, as I thought it would be useful to get
the ‘lie of the land’ before I tried to do too much. Once I decided that
I was ready, I started recording linguistic data, and things were progressing
well. I worked this way for about three months, and was making
satisfactory progress.
After about three months, however, things began to change. I began to
feel that my work wasn’t progressing as quickly as it should. I began to
get a little worried because I had a doctorate hanging over my head that
depended on my gathering a healthy amount of data. My arrival on
Paama had unwittingly ended up coinciding with the time of the annual
yam-planting season. Clearing the bush, preparing the ground, planting
the yams, and tending the growing vines is heavy work that the entire
community has to be involved in. It simply wasn’t ‘on’ for me to expect
people to absent themselves from their community responsibilities to sit
down and work with me on linguistic matters. As the yam-planting
season progressed, I was getting increasingly anxious that my own
linguistic work was suVering.
Given that everybody was away from the village working most days,
I simply tagged along with the work parties. I even joined in the work
where there were things that I could do. I was able to gather some
yam-planting terminology in this way, though the amount of data that I
was recording for the physical eVort that I was putting in was fairly small.
I continued to worry, though I was at the same time having a wonderful
time mixing with people and learning new things.
However, the anxiety that I was feeling about my linguistic project and
the social mixing that I was doing were beginning to feed on each other.
After about three months of being constantly in the company of a village
full of people, my slight worry was starting to turn to stress. I had up till
then been used to living by myself while I was a university student in
Australia. When I came to Paama, however, I found that I was unable to
take oV by myself whenever the fancy took me. I could not go for a walk
by myself in the bush. If anybody saw me heading oV by myself, at least
one other person would feel obliged to tag along. It turned out that there
was a reason for this: if I went by myself, I would be subject to all manner
of supernatural forces in the bush that I was not equipped to handle.
I might, for example, accidentally walk through a taboo place, which, it
was believed, could place either my life or health in danger, or the life or
health of other people. In the view of the Paamese, it seemed that I needed
constant baby-sitting.
My inability to spend time by myself, coupled with a growing anxiety
about the lack of progress in my work, led me to go troppo. This is
a somewhat traumatizing experience in which you crave for familiar
experiences, and you begin to resent the unfamiliar. This leads to
occasional hostility towards people who feel they have acted in
a completely friendly and supportive manner towards you. A touch of
paranoia seems to be part of the collection of symptoms, and it is far from
pleasant.
Nobody had told me to expect this to happen before I went to Paama.
Once you begin to experience culture shock, possibly the best way to
overcome it is to have a short break away from the Weld location and
come back refreshed. Eventually, my rather negative feelings passed and
I got used to my situation. In fact, once you overcome this fairly
uncomfortable period, your time in the Weld can become immensely
enjoyable, and you may eventually experience great sadness upon leaving.
DiVerent people may experience diYculties with diVerent aspects of
a particular culture. In most parts of Vanuatu, people have converted to
Christianity in the last century or so and people today are generally
devout Christians. People in most communities attend church regularly.
The Sabbath is fairly strictly observed in many places, with no manual
work being permitted, and there are sometimes even limitations on doing
anything that could be construed as having fun, such as frolicking in the
sea. The idea that one might be an agnostic or an atheist is diYcult for
many people to understand, so some attendance in church by a visiting

linguist is possibly a good idea. Many westerners do not say grace before
meals, but people’s religious traditions in Vanuatu generally call for
somebody to say grace before the meal is started. The visiting linguist
may even be called upon to contribute from time to time, so it is worth
being prepared.
Whether one is a man or a woman can also inXuence one’s reaction to
particular elements of the local culture. In many traditional rural
societies, men have rather more freedom than women, and this may
also include visiting linguists. Women may also be expected to exercise
modesty in dress and behaviour that may seem restrictive at home. In
Vanuatu, for example, while men are free to smoke and drink alcohol,
women are not. Women are also not expected to wear slacks, and they
certainly do not wear shorts and T-shirts. Visiting western women may to
some extent be granted ‘honorary male’ status in that some aspects of
non-standard behaviour may be tolerated, but it is always necessary to be
fairly subtle when testing what may be acceptable and what may not.
You can also expect that sexual mores will be quite diVerent from what
you are used to at home. Of course, local people will have their own
preconceptions of the sexual mores of westerners, who may be perceived
as not being subject to any kind of restraints. What may to you be
a completely innocent happening, such as a conversation between
a single female Weldworker and a young unmarried local man, may have
quite a diVerent meaning altogether. I remember once having a perfectly
innocent conversation in broad daylight with a female visitor to
Erromango from New Zealand and behind her back, but in my line of
sight, the local young men were giving me all sorts of encouragement with
their winks and smirks. For these kinds of reasons, it is often best for
female Weldworkers to work with female language-helpers, and for male
Weldworkers to work with male language-helpers, unless there is
a substantial age diVerence between the two.
Not everybody goes troppo. Those who do experience it Wnd that it
comes in diVering degrees, and that it manifests itself in diVerent ways. In
my case, there was some depression, as well as irritability towards
particular people (especially those who had become my friends, and
who were therefore with me most often). There are probably just as
many solutions to culture shock as there are manifestations of the
phenomenon in the Wrst place.
6.4 Getting caught up
I discussed participant observation as a way of gaining access to linguis-

tically valid data. I did not mention at that point that you can expect
participant observation to lead to a variety of diVerent kinds of social
relationships within the community. Some of these relationships can be
immensely rewarding, while others end up being more like entanglements
than rewarding relationships.
6.4.1 Becoming part of the scene
If somebody is more than just a short-term visitor, it often becomes

necessary for them to be incorporated in some way into the local
community structure. In some cases, you may well end up being ‘adopted’
by a particular family, thereby acquiring a mother and a father, brothers
and sisters, and nephews and nieces. While this kind of welcome can be
Xattering, it must be remembered that children in paid employment may
have extensive obligations to support their rural parents and siblings
Wnancially, and this privilege may end up being extended in your direction
as well.
While this could be viewed as an inevitable Weldwork expense, it is
something that you may Wnd it diYcult to justify to your university’s
accountants. You may therefore be forced to do a little creative accounting
by which that pair of running shoes or that watch that somebody
persuaded you to buy mysteriously appears as an amount of money
under some other budget heading.
Alternatively, you may end up having to fork out from your own
pocket, though this will depend on how deep your pocket is. Trying to
explain to somebody that you simply cannot aVord some particular
item may not be easy. People may Wnd your protestations of poverty,
even if you are a graduate student, diYcult to believe, because you are
still likely to be much better oV than local people in many parts of
the world. I have been quizzed in the Weld about my income, and people
have only wanted to know my total income. They have not been the
slightest bit interested in hearing about the rather smaller amount
left after national and local government taxes, mortgage payments,
insurance, deductions for health care, compulsory union fees, savings
for retirement, and utilities.
You may occasionally also be questioned about the possibility that you
are going to make lots of money selling books about the language that
you are working on. While some linguistic descriptions have no doubt
earned good money for both the authors and the publishers, none of the
linguistic descriptions that I have produced fall into this category.
Relatively few people are aware that most academics earn little or no
direct income from royalties, so you may have to explain something
about the economics of publishing to members of the local community.
(Of course, there is a sense in which we do gain less directly from the
products of our research, in that we can go for a promotion, or seek
a more prestigious or a better-paid job at another university on the basis
of our publication record.)
6.4.2 Unexpected entanglements
There is a host of additional kinds of involvements that you may end

up being caught up in, sometimes completely unintentionally, and the
ultimate outcome of these involvements may be much more diYcult to
predict. I once did some very brief grammatical elicitation on the Neve‘ei
language of Malakula with a speaker of the language who was studying in
New Zealand. I subsequently accepted an invitation to spend Christmas
with his family on Malakula to see in the year 2000. This was an invitation
that I was keen to accept, as I decided that I would prefer to say farewell
to the 1900s well away from the hoopla that seemed to be enveloping the
rest of the world over the advent of the new millennium. A small
remote village with no electricity, hence no TV, seemed like an excellent
option to me.
Shortly before I left for Malakula, I discovered a number of previously
completely unanalysed texts in Neve‘ei that were recorded by the British
anthropologist Bernard Deacon when he visited the area in 1926. That
was at a time when people were still living a largely traditional lifestyle up
in the interior of the island. While his material was suYciently well
transcribed to allow it to be recognized as Neve‘ei, it needed some careful
checking by a linguist to correct some of the phonetic transcriptions.
On the day after Christmas when people in the village were relaxing,
I took out my photocopied pages of Deacon’s stories and explained that
I had found some stories in their language in an old book which I now
wanted to write out correctly. Since Neve‘ei is not a written language,
I expected that people would Wnd it diYcult to recognize words from the
written page, particularly as a number of unfamiliar phonetic symbols
were used. I therefore adopted the strategy of attempting to read what
Deacon had written and I then asked people either to indicate if what
I had said was correct, or, if it was not correct, I would ask them to say
what it should have been.
When I had Wnished going through each story word by word, I then
read it back in its entirety to check that I had it right. The Wrst time I
did this, the reaction from my audience—which had grown quite
substantially out of curiosity from the single individual who I had initially
approached—was one of considerable amusement, not to mention
amazement. After all, here was this khabat ‘European’ who, after having
been in the village for a mere three days, seemed to be able to speak the
local language.
Word rapidly seemed to spread about this unusual khabat, and people
began dropping around with requests for me to ‘tell a story in the
language’. Although I said I couldn’t ‘tell a story’ in the language,
people were happy enough for me to read one, especially since practically
nobody in the village had ever had an opportunity to read their language
themselves. This quickly developed into something of a party trick that
I would be asked to perform, and these story-readings became sources of
amusement and amazement for people of all ages. While some of the
stories that I was ‘telling’ were well known, one was of particular interest,
as most people had never heard it. However, one old man informed
people that he had in fact heard the story himself when he was a boy
and that it was correct in all of its details.
By this stage, I must admit that I was enjoying the attention somewhat,
as it is often diYcult for someone from another culture, especially on
a short visit to a place such as this, to Wnd something that one is
suYciently good at that one can make a serious contribution. Some
people were apparently giving more serious thought to the signiWcance
of my party trick than I was. My host’s father volunteered the informa-
tion that I was not the only person in history with this particular ability.
Long ago, he said, there had been another khabat who surprised people
by turning up with the same ability. The Neve‘ei-speaking people have
only been Christians since the 1940s. Prior to this, people occupied
a series of bush hamlets, practising their traditional culture in the interior
adjacent to the coastal area where they now live. At that time, a khabat
visitor arrived speaking the language, and also participated in their

traditional dances and took part in the grade-taking ceremonies that
Malakula is famous for in the anthropological literature.
I was told that he was an Englishman by the name of Teken. This was
very obviously Deacon’s name in a Neve‘ei form, which had been
enshrined in oral tradition since 1926. Although Deacon was welcomed
initially, things at some stage apparently went wrong. I was told the story
of how he took it upon himself to visit a place called Lebelang up in the
interior where he should not have gone. This place is still regarded today
as being very powerful, and the consequences of trespass in the
pre-Christian 1920s were severe: the old people resolved that Teken
should be killed and eaten. However, he was spared, possibly because
people would have been aware of likely serious consequences from the
colonial authorities of such an act against a British citizen.
Deacon left the area for the southern part of Malakula, where he had
already carried out a considerable amount of more detailed ethnographic
work. According to the story as it was told to me, when he got back to
southern Malakula, he fell ill and died shortly afterwards, in 1927
(see Gardiner 1984: 75). When a young and otherwise healthy person
dies on Malakula, this is frequently attributed to a variety of mysterious
and malevolent forces. It is easy to imagine that people at the time would
have seen Deacon’s death as some kind of natural consequence of his
transgression against their taboos.
I was told that the tradition was for people to be buried with their heads
to the setting sun. The people at Southwest Bay, however, were reported
as having buried his body with his feet facing in the direction of the
setting sun, with the intention that this would make him ‘come back’.
Presumably, those responsible for this interpretation of the manner and
the signiWcance of his burial had a better impression of Teken than the old
people from the Neve‘ei-speaking area who had wanted to kill and eat
him, and who probably saw in his death some kind of appropriate
retribution.
Meanwhile, back in Vinmavis in 1999, when Christmas was over and
before the new year, this was interpreted in Vanuatu—as just about
everywhere—as representing the end of the old millennium. This came
to be associated by some people with potentially earth-shattering events.
On two occasions, I remember being asked at the time by local people
whether I thought that the world would end at midnight on 31 December.
There was also an almost comical rush of weddings in the capital, as

people who had been living together for years in unsanctiWed ‘bush
marriages’ were keen to have church marriages performed, apparently
so that they could non-adulterously face their maker on New Year’s Day.
People in Vinmavis seemed to me to be approaching the new
millennium much more rationally. The local Presbyterian church building
was showing very visible signs of having endured many years of cyclones,
earthquakes, and the normal wear and tear that buildings are subject to
in a coastal environment in the tropics. It was decided that the last
communal act of the old millennium would be for the entire congregation
to come together to demolish the old church building. The date set aside
for this was 28 December 1999. People would then rebuild their church
from new foundations at the beginning of the next millennium.
The task of disassembling the church was completed with great speed
and eYciency, leaving just a crumbling concrete foundation after two or
three hours of sweaty and dusty work. However, a few hours before the
Wnal service to deconsecrate the old church was due to begin, I was
approached by a member of the church committee and asked if I would
agree to participate in the programme by telling one of the stories that
Teken had written down. My initial response was to say that I would feel
very uncomfortable about the idea of standing up in front of the entire
village and being presumptuous enough to dare to tell them one of their
traditional stories in their language, and probably not very well at that.
The church committee member persisted and he was clearly not going
to take no for an answer. It quickly became apparent that there was more
at stake here than just my own sensitivities. This was being seen as
a historical occasion at the turn of the millennium. My coming to visit
so many years after Deacon was buried with his feet to the setting sun,
and having an ability to tell the stories that he had recorded, was surely
too much of a historical coincidence for such a contribution to be missed.
I therefore acceded to the request, though not without considerable
trepidation on my part.
The story that I chose was the one which I already knew to be no longer
widely known, but whose accuracy had already been conWrmed by the
older people. I will not say much about the story itself, other than to
comment that it related to some particularly un-Presbyterian subject
matter involving corpses, cannibalism, and defecation. I judged that
my telling of the story went surprisingly well despite my own extreme
nervousness and the apologies that I expressed beforehand in case I spoke

unintelligibly, made mistakes, or mentioned subject matter that some
people may have objected to. Whatever shortcomings there were in my
delivery, I clearly spoke intelligibly enough because it was a story that
I am certain was intended to provoke laughter, and the congregation
laughed in all the right places. When I Wnished, I scurried back to my pew
to a round of spontaneous applause, feeling relieved that the experience
was over, but gratiWed and humbled by the response.
I have already indicated that the word that the people of Vinmavis use
to refer to Europeans is khabat. According to the traditional belief
system, the same word is also used to refer to a ghost-like individual
with superhuman powers who is of great cultural signiWcance. Khabat
were also supposed to be able to be reborn as children after their death.
These khabat are supposed to have been able to perform various kinds of
magic, and also to have had white skin. It is not too surprising to Wnd that
the Wrst Europeans to visit Malakula also came to be referred to as
khabat. Not only do Europeans qualify in terms of physical appearance,
but their technology, even in the nineteenth century, could easily have
appeared magical. For instance, they had long sticks which could be
pointed at a pig or a person and which would make a loud noise.
The pig or person would then suddenly fall over dead without anybody’s
direct touch. This long stick—a riXe—came to be called a nivis khabat,
because it could be used to do the same thing as the traditional nivis ‘bow’,
but in a much more mysterious and dramatic way.
Although Malakulans today still refer to Europeans as khabat, I am
quite certain that people do not regard Europeans as ghosts. Europeans
are not assumed to have superhuman powers, and everybody now knows
that the secret of the riXe is that you go to the store and buy bullets.
Judging by the way that the people in Vinmavis who I got to know were
interacting with me, I never felt as if I was being treated as anything but
fully human. Whenever small children burst into tears of fear at the sight
of me, this was not because they are told that white people are ghostly
bogey men, but because they are doctors who give them much-dreaded
injections.
Some may be tempted to argue that the use of the word khabat to refer
both to powerful ghostly beings and to Europeans is an indication that
Malakulans think that Europeans are really returned ghostly ancestors
of some kind. There are certainly plenty of examples of reports from
elsewhere in the PaciWc of Europeans at Wrst contact—and even of

linguists and anthropologists well after Wrst contact—being said to be
‘ghosts’. However, when I have carried out Weldwork in Vanuatu, or
when I have simply been visiting people socially in rural areas, I had
never been faced with any kind of situation where there could have been
even the remotest possibility of this kind of interpretation, so I tended to
be a little sceptical of claims of this kind, suspecting that people were
perhaps overinterpreting—or even misreporting—their experiences.
Deacon’s appearance in 1926 would have been early enough that in
many parts of Malakula he would probably have been the Wrst European
that some people had ever seen. He may therefore even have been
assumed by people to have been a khabat in the traditional non-human
sense, rather than in the modern human sense, and he hinted at this
himself in a letter from the Weld to a correspondent in the UK: ‘Here
come the men and I must talk to them about oVerings to ghosts. I am
a ghost’ (Gardiner 1984: 45). If a signal was sent to the people of
Malakula that Deacon was planning on returning after his death by
being buried with his feet facing in a particular direction, this may well
have sat comfortably with the belief system of people at the time.
This all leads to some fairly heavy questioning of the role of the
ethnographer and of the linguist in the community which he or she is
investigating. Many traditional anthropological and linguistic descrip-
tions make practically no reference at all to the investigator or to local
people’s reactions to him or her. However, all observations in which an
observer is present are based on what happens while he or she is engaged
in observing people, and it is the outsider who is then interpreting that
behaviour for the reading audience. It is because of this that I feel that
I could not ask people in Vinmavis the very question that anthropologists
would probably most like an answer to: Am I Teken returned? If so, in
what sense might this be the case?
For some people, it would surely be a ludicrous question to ask, as well
as pretentious and/or insulting. Such a question arguably operates on the
presupposition that Third World people who are frequently represented
as ethnographic subjects might be somehow less rational than westerners,
as has been argued by Obeyesekere (1992). Yet, for other people, this may
not be such an impossible idea to suggest, though it would have to be
recognized that there are diVerent levels at which people could hold such
beliefs. Some people may take the literal view that if Deacon was to be
regarded as a khabat in the traditional sense (as well as in the modern

sense), then any suggestion of a later reincarnation is tantamount to
blasphemy and to be rejected outright by any modern Christian. On the
other hand, other people may view any possible ‘return’ of Teken more
metaphorically, with my appearance in Vinmavis being Deacon-like in
certain unexpected respects at the turn of the millennium. Finally, of
course, there is the possibility that the interpretation may be taken
literally.
I suspect that it would not be possible to oVer any single interpretation of
the events which I observed—and in which I unintentionally ended up
participating—in Vinmavis over the Christmas/New Year period of
1999–2000. My presence, as well as my ability to tell a long-forgotten
story in the local language, was obviously a fortunate coincidence which
allowed me unwittingly to become part of the millennial symbolism in
Vinmavis, rather than the simple bystander as I had originally anticipated.
Because of the role that was thrust upon me, however, it is diYcult for me
to seek the precise meaning that my presence had at that historic moment.
Further reading: Kulick (1995: 268) (Appendix: On being a ghost).
6.5 Linguists behaving badly
Successful Weld research depends on an ability to maintain good relations

both with the individuals with whom you are working and with the
community at large. In some cases, relationships may become strained
and this can even threaten the viability of a project. In extreme cases,
similar kinds of projects which might be proposed for some time in the
future may even be jeopardized. Sometimes, an individual may be
personally responsible for having the welcome mat withdrawn, while in
other cases, things may go wrong for reasons that are well outside your
own control.
At the level of interpersonal relations, people from diVerent cultures
can generally Wnd ways of getting on together in spite of their diVerences.
It may be something of a cliché, but all it takes is a little sensitivity and
a willingness to learn. Obviously, I could not hope to oVer practical
lessons in how to behave in a culturally appropriate way in every cultural
group, so you will need to discover for yourself what the local faux pas
might be so that you can avoid them.
But it is not diYcult to make a mistake, in which case you need to hope
that people in the local community are going to be tolerant of your
weaknesses. For instance, when carrying out Weldwork in Vanuatu, it is
most unwise for a visiting linguist to drink a fair amount of wine and to
follow this up with the consumption of a fair amount of strong kava.
(The wine, of course, facilitated the misjudgement in the Wrst place.)
The result of this kind of behaviour is that a certain amount of cleaning
up becomes necessary, and one’s sense of dignity can take something of
a beating. However, the visiting linguist may also be called in the next day
for a more formal talking-to from the local chiefs, which will require
a certain amount of humility, along with a formal apology.
Of course, sometimes, people do the wrong thing as a result of their
own straightforward stupidity or even deliberate pig-headedness. So
serious may the blunder be that one researcher may sour a community’s
attitude towards more thoughtful researchers in the future. I remember
hearing of one linguist who decided to jump into a community’s supply of
drinking water to have a wash. Not only was that linguist not
invited back, but I could well imagine that local people would have
been somewhat wary of inviting any other linguists into the community
for some time to come.
If your Weldwork is being conducted from the vantage point of
a temporary resident of a local community—arguably, I have repeatedly
suggested, the best way of obtaining the richest and most reliable
data—you also need to remember that you are somebody’s guest. You
will of course have included some provision in your Weldwork funding to
compensate your hosts in some culturally appropriate way. As
a thoughtful guest, you should also be prepared to muck in and help
with household tasks. If you end up in a working party of people digging
a hole for a new toilet or bringing in next week’s supply of Wrewood, you
should be prepared to treat this as a new experience, as well as an
opportunity to hear the language being used in new contexts.
Having no doubt hosted visitors of your own in the past, you will also
be aware that sometimes guests can outstay their welcome. You need to
bear in mind that while people may genuinely be pleased to host an
overseas visitor such as yourself, your presence may sometimes also
represent something of a burden. Before I started Weldwork on the island
of Erromango in Vanuatu, I visited a friend there socially for a week while
I was on holiday. As soon as I arrived, my friend’s mother said something
to him in a low voice which, it turned out, was an anxious question,

‘What am I going to cook for him?’ I quickly assured his mother that she
didn’t have to put herself out at all. I would be happy, I said, to eat
anything at all that the family was eating. My only special requirement
was a cup of tea Wrst thing in the morning, and I made sure that I brought
my own packet of tea with me that I could share with the family.
Local people typically get their Wrst cup of tea when the women of the
household have Wnished cooking the bananas, and this tends to be around
8.00 a.m. Unfortunately, by then I would be getting the Wrst pangs of
a caVeine-withdrawal headache, as I am very much a morning person,
and Wrst thing for me is around 7.00 a.m. (at the very latest). To cater for
my needs, my host’s mother could be heard chopping the Wrewood to get
my cup of tea ready shortly after birdsong in the early morning while
I was still lying in bed.
While I was perfectly capable of lighting a Wre and boiling the
tea myself, I realized that it would probably be a major task for me
every morning to get things organized. As I was a visitor, I wouldn’t
necessarily know where the axe was that morning to chop the Wrewood,
where the matches were to light the Wre, where the kettle was, where the
cups were, and so on. Since I was intending to be there for just a week,
I thought it would probably be best just to accept my host’s mother’s
generous accession to my special caVeine-related need. When I returned
on my next visit, this time with the intention of staying for three months
of Weldwork, this kind lady immediately reverted to her early morning
tea preparation mode, just for my beneWt, and once it had started,
it became extraordinarily diYcult to do anything about it.
This is just a small example of how easy it is for your presence to aVect
a local family’s routine. I am hoping that this particular burden that I had
accidentally imposed was not too onerous, as the Wre would have to be lit
every morning, whether I was there or not. However, there are times when
the burden could become a problem. I am currently working in a village
on Malakula in Vanuatu. The area where this village is located is prone to
extended dry periods. While water is piped into the village from a distant
spring, a land dispute was causing somebody to interrupt the Xow of
water from time to time in order to make a point.
At one point, for a whole week, we were having to bathe either in the
sea, or in slightly smelly (and certainly not drinkable) well water. I began
to think that if this water problem persisted, the whole community would
begin to face serious problems with drinking water. I decided that if

things got much worse, I would have to consider leaving. While my
departure would certainly not solve the drinking water problem, it
would at least have saved people from having to worry about me in
addition to their own problems.
People in the past have sometimes appropriated valued cultural arte-
facts and taken them away in violation of local custom or wishes. In many
countries, laws have been passed to prevent this kind of behaviour, but this
has not stopped ‘art collectors’ from stealing artefacts in violation of
national laws. Although Weld linguists typically do not get involved in
such activities, there has sometimes been a residue of suspicion that
attaches to us, because local people do not always Wnd it easy to distinguish
between a linguist, an anthropologist, a legitimate art dealer, and an art
thief. In some cases, the climate of suspicion has been strong enough for all
social science—including linguistics—to be banned for extended periods.
Some linguistic Weldworkers become so obsessed with the need to
become participant observers that, at some level at least, they try to
convince people that they have crossed the boundary between ‘us’ and
‘them’. Of course, it takes far more than an ability to speak the local
language, much more than a willingness to go around barefoot, and more
than a penchant for wearing some form of traditional dress or adornment
to change one’s ethnicity. While it is perfectly possible for a linguist to
adopt local traditions as a way of being respectful, and local people may
even encourage you to do this, nobody will be fooled into believing that
you have become one of ‘them’, even if you might believe this yourself.
One linguist has reportedly been correcting local people for years when
they are speaking their own language. He appears to have no idea how
insulting educated members of this language community Wnd this kind of
behaviour. I know of another Weld linguist who made a point of using one
of the local languages in a formal meeting with government oYcers where
all of the local people were using the oYcial languages of the country.
Although the linguist presumably wanted to make the point that the local
languages should be treated as being important, nobody at the meeting
understood the content of his talk. The experience was apparently quite
embarrassing for all concerned, apart from the linguist who was oblivious
to the reaction.
Sometimes, the insensitive behaviour of Weld linguists extends also to
their own colleagues. There was a linguistics conference that was held at
the University of Papua New Guinea about twenty-Wve years ago in

which most of the papers were presented by people who were doing
Weldwork on some of the many diVerent languages in the country.
Presenters and discussants in the various sessions—who were almost
exclusively non-Papua New Guineans—were repeatedly saying things
like ‘in my language, we say X’ and ‘in my village, people say Y’. This
use of phrases such as ‘my language’ and ‘my village’ can be taken as
an arrogant usurpation of somebody else’s language and identity by
English-speaking academic linguists from overseas, or as a thoroughly
naive and precious attempt to demonstrate to other linguists how
successfully he or she has been able to ‘go native’.
Further reading: Vaux and Cooper (1999: 10–21).

7
Salvage Fieldwork
The term ‘salvage’ Weldwork refers to the study of languages that are
moribund, i.e. the language is being documented because it is realized that
it is rapidly disappearing. This kind of Weldwork often involves particular
problems that go beyond any of the topics that I have discussed already in
this volume.
7.1 Getting started
This subheading is the same as the title of an entire chapter which

appeared earlier in this volume. While some of the issues relating to
salvage linguistics are the same as those that we encounter when working
with communities where vibrant languages continue to be spoken, there
is a whole raft of additional issues which relate speciWcally to the
commencement of salvage Weldwork that must be discussed separately.
7.1.1 Responding to the enthusiastic
Finding linguists who are willing to conduct this kind of Weldwork is one
issue. A lot of descriptive grammars these days are written by graduate
students carrying out research for the doctoral dissertations. Some
linguists may write only a single grammar, one which is written for their
doctorate. When prospective graduate students are looking for languages
to describe, the preference is overwhelmingly to work on languages which
are still being actively spoken within vibrant speech communities rather
than on languages with just a handful of remaining speakers.
I was recently approached by a member of the Orkon community,
which originally occupied part of the west coast of the island of Ambrym
178 salvage fieldwork
in Vanuatu. This language is now moribund, with fewer than twenty

speakers living in two diVerent villages where another language has
become the primary medium of exchange. On the basis of my ongoing
research involvement in Vanuatu, I also regularly receive requests from
intending graduate students from a number of diVerent universities
around the world for advice about suitable language communities in
which to carry out doctoral research. I would love nothing more than
to meet requests from these graduate researchers with this direct request
from a speaker of Orkon by advising somebody that they should carry
out research on this language, which very clearly does not have
a healthy future.
However, I do not feel that I could do this. A grammar that is detailed
enough to meet the requirements of a doctoral dissertation needs to be
based on information that can only come from exposure to a language
that is still actively spoken by an entire community. While some of the
remaining speakers of Orkon no doubt still have an excellent command of
the language, it would be almost impossible to observe the language being
used to any great extent in a natural way. All speakers of Orkon almost
certainly spend most of their time speaking other languages rather than
Orkon. It seems to me that it would be unfair to send a graduate student
to record data under these kinds of circumstances when so much depends
on their ability successfully to complete their doctorate.
Ideally, this kind of situation calls for an experienced Weld linguist who
already has a doctoral degree and for whom it is not nearly as important
to be able to produce a detailed and comprehensive grammatical descrip-
tion. In this case, there are several linguists who could be called upon to
help, including myself. However, established linguists are also quite likely
to have extensive teaching commitments during term time, as well as
longer-term research commitments that are already under way, which
can make responding to such a request diYcult. The end result is that
languages that are in greatest need of documentation are sometimes those
which are least likely to be described.
What is particularly gratifying in the case of Orkon is that there is
a member of this community who is keen to see the language documented.
We can only hope that this enthusiasm succeeds in bringing a linguist on
board to carry out the kind of salvage Weldwork that is so clearly needed.
My own work on the moribund Nāti language of Malakula (Crowley
1998a) came about as a result of a similar burst of local enthusiasm.
salvage fieldwork 179
At the end of a short social visit to Malakula, when I became stranded

with the closure of the airstrip due to heavy rain, one of the last speakers
of this language recognized an advantage to him in my misfortune and
took me to the local store to buy some exercise books and pens, found
somewhere quiet where we could work together without interruption, and
proceeded to get me to write down as much of his language as I could in
the few days that I had available before the airstrip was reopened.
Interestingly, when one of this man’s relatives saw us working together
on his language, he was asked how much he was paying me to record all of
his linguistic data, which indicates how valuable a task this was seen as
being. (Needless to say, I did not accept any money for the task that I
had performed.)
7.1.2 Persuading the reluctant
However, despite the obvious value of the documentation of moribund

languages in situations such as these, it is sometimes surprising for
linguists to Wnd that proposals to carry out this kind of salvage work
may be met with varying degrees of interest from the communities
involved (Gerdts 1998). My own experiences of Weldwork on moribund
Australian languages in the 1970s involved acceptance of my activities at
best, with some outright hostility, but never, as I cast my mind back,
completely enthusiastic support. Aboriginal people in the areas where I
worked had suVered major demographic decimation, communal degrad-
ation, and personal exploitation and humiliation as a result of many
decades of social policies and communal attitudes that were paternalistic
at best, and downright destructive at worst. This led to strong feelings of
linguistic and cultural loss. All of this made it very easy for people
to regard any unfamiliar European linguistic Weldworker either as yet
another intrusive and exploitative agent of government, or as yet another
patronizing and interfering (or even racist) member of the community
at large.
Under these conditions, salvage Weldwork was far from easy, and a Weld
linguist had to acquire a thick skin quickly. One reputedly good speaker
of the then poorly documented Gumbaynggir language in northern New
South Wales, for example, simply turned her back on me when I said that
I was not in a position to pay her thousands of dollars for the privilege of
recording her language. Of course, I did have some funds at my disposal
with which to pay those who made their time available to work with me,
though the amounts were far more modest than what was expected on
this occasion. While other people did not demand large sums of money,
they certainly did not always go out of their way to make themselves
easily available to work with me.
I encountered other less negative reactions as well, and many people
did allow me to take up their time to record what they knew of their
ancestral languages, even if that was sometimes only very little. However,
my memories from the 1970s are by and large of having constantly to
encourage—and sometimes even cajole—people to continue to partici-
pate in this process of linguistic documentation in order to meet the
expectations of academic linguists. Those were the days before this
kind of research was governed by Ethics Committees which require
participation to be uncoerced. The advice that I was given at the time,
when faced with these kinds of situations, was to appeal to people’s sense
of guilt by saying ‘I have a bossy professor who will not allow me to come
back empty-handed’. I cannot remember now if I ever made use of that
speciWc piece of advice, but it is advice that should never be given to a Weld
linguist today.
While we often assume that people all over the world are proud of their
language and the distinct identity that the language expresses, the reality
is sometimes a little more complex. Some people view a knowledge of the
local language as a sign of unsophistication and ignorance, and Wnd
it embarrassing that a foreigner might want to study it, and this can
make it diYcult for a project of salvage Weldwork to get oV the ground.
Lyle Campbell (pers. comm.) reports that it is quite common right
across Mexico and Central America for indigenous people of many
diVerent groups to have been made ashamed of their local language by
the Spanish-speaking majority and many conceal a knowledge of the
language because of this.
Many try very hard not to be identiWed as Indians, particularly when
away from their communities. People often talk about Indios revestidos
‘re-dressed Indians’, where Indio is extremely pejorative, as bad as—or
worse than—the N-word in English. The image expressed in this phrase is
that members of indigenous groups will dress in western clothing to try to
look un-Indian and thus not suVer the slurs and prejudice. If outsiders
come to a village and ask someone if they speak the local language, the
answer will commonly be ‘no’ even when they do. This is partly because
of the sense of shame that being ‘Indian’ brings, and partly from
a distrust of outsiders who have a history of coming in and cheating
native peoples out of their land and deceiving them in many other ways.
7.2 Working with a limited number of speakers
One of the major problems that we face when carrying out salvage
linguistic Weldwork is the very small number of people who speak the
language. In extreme cases, there may be just a single remaining speaker
of a language. Quite a number of linguistic descriptions in Australia
have been written on the basis of information drawn exclusively, or
almost exclusively, from a single speaker. In fact, my own published
salvage studies of Yaygir, Bandjalang, and Mpakwithi in Australia arose
out of material that was gathered in this way from single speakers of each
language.
7.2.1 Diminishing speaker populations
Earlier in this volume, I pointed to the danger of attempting to write an

account of a living language on the basis of a single speaker who has been
dislocated from his or her language community. Whatever limitations
there are on the validity of a grammar which has been written without
access to observations of spontaneous usage would presumably be
equally applicable to salvage Weldwork that has been carried out under
these circumstances. This means that my own accounts of Yaygir,
Bandjalang, and Mpakwithi quite possibly misrepresented the true nature
of the language in certain respects. The problem is that we will now never
know the extent to which this is true, and what sorts of defects there may
have been in my accounts of those languages, because in each case, the
Wnal speaker who I worked with in the 1970s has since died.
Of course, a language can have more than just a single speaker and still
be moribund. It is probably more likely that there will be at least several
people who can speak a language, or maybe as many as ten or twenty.
Whatever the precise number of speakers, when you are attempting to
describe a moribund language, the small size of the remaining population
of speakers means that you are often not going to be able to choose your
language-helpers in the same way that you can when you are working in
a vibrant language community. This means that you will simply have to
make the best use of whoever is available (and willing) to work with you.
Because of the age of speakers, it is quite possible that somebody’s
physical frailty may be a real factor in limiting their ability to work with
you. I know of one linguist who had been invited by somebody’s family to
record something of one language but it turned out that their elderly
relative was so frail that he could not get up from his bed. In this case, the
only decent response was to explain to the family that this kind of work
would be an intolerable burden on the old man and to promise to come
back when he was feeling better. In the end, that was not possible and the
old man died, but at least he was able to spend his Wnal months with his
family rather than with an intrusive linguist.
In other cases, people may have put you in contact with somebody who
may be physically quite strong but who may be suVering from some kind
of dementia. This can make it diYcult to elicit meaningful and reliable
responses to questions, so opportunities for grammatical and lexical work
may be limited. On one occasion, I was asked to record a story from one
old man and he produced a long account of a historically signiWcant event
that he remembered from his youth. However, when I played the story
back to other people who could understand the same language, they
threw up their hands in despair. They basically declared the story to
be untranscribable because the old man was rambling, sometimes
incoherently. To this day, that story has not been transcribed.
In some cases, you may Wnd that you have to deal with an elderly
person’s insuYciently clear articulation. I have already indicated that an
older person may fail to grasp the notion of the paradigm, thereby
preventing you from acquiring a full range of allomorphic variation.
This may not be a problem with a language that has speakers spanning
all generations, as you can probably expect to Wnd somebody else who
can help you with this kind of elicitation.
Further reading: Evans (2001).
7.2.2 Diminishing structures
When a language has become moribund, it will not only become restricted
to a small group of relatively elderly people, it will also become
restricted in the range of contexts in which it is used. Sometimes, the
remaining speakers will end up living in quite diVerent locations, in which
case the language may be used on only those very rare occasions when
the remaining speakers get together in a single location. Even then, the
opportunity to use the language may be very limited, as the language that
has become the dominant language of the community is likely to intrude.
With few opportunities to use the moribund language, it is only natural
that people’s command of the language will begin to suVer.
There have been a number of studies of the eVects that this kind
of functional and demographic restriction can have on the lexicon, the
phonology, and the grammar of a language. It is very likely that a speaker
of a moribund language will exhibit all kinds of unsystematic variation
between the moribund language and whatever happens to be the domin-
ant language of the community, and possibly also in the direction of the
language of elicitation.
When I was working on Bandjalang in the 1970s, for example, I found
it diYcult to elicit spontaneous textual data, so my description was based
for the most part on data translated from English prompts. It turned out
that the word order in the Bandjalang data that I collected as a result was
predominantly SVO, just like my English prompts. This, however, is
somewhat suspicious, given that Australian languages tend either to
have SOV word order, or they have completely free word order. There
are other facts about Bandjalang which make it look as if the speaker who
I was recording data from was being inXuenced in her word order by the
sentences that I was giving her in English.
There are other situations involving moribund languages where
a speaker is clearly mixing in vocabulary items, and even grammatical
constructions, that belong to the major actively spoken language in the
area. There are only about half a dozen speakers of the Ura language of
Vanuatu, all of whom also speak the Sye language. Whenever I tried to
record Ura data, there was always a component of the data that looked
suspiciously like Sye rather than Ura, and it became necessary to ‘weed’
this data out. It also turned out that some individuals were more likely
to produce ‘contaminated’ data than others, so it was necessary to work
out who were the more reliable speakers, and who were not, and to take
everybody’s data into account when producing the Wnal description.
The grammar of Ura that I eventually produced was therefore not simply
an analysis of my corpus, as I had to weed out certain data Wrst that
I had deemed to be ‘contamination’ from the other language.
When a language becomes functionally and demographically restricted

in these sorts of ways, sometimes the last speakers no longer have
anything like a full command of the language. Such speakers are often
referred to in salvage linguistic studies as partial speakers. Situations can
vary quite considerably as to how impaired any individual’s capability in
a language might be.
It is quite possible that somebody may not realize until you come along
and start eliciting lexical and grammatical data that they do not
know nearly as much of the language as they—or other people in the
community—thought they did. It could be quite harmful for a linguist to
come along and eVectively show a respected elderly member of the com-
munity up as somebody who knows very little. You must remember that
one of the overriding ethical obligations of the researcher is to bring about
no harm to your research subjects.
It is essential, therefore, to be constantly on guard when working with
partial speakers for possible limits to their knowledge. You can expect
that in the vocabulary, less common and more speciWc kinds of terms
are less likely to be remembered than more common words. If you ask
for the words for highly prominent body parts such as head, nose, eye,
and mouth, you stand a good chance of getting a positive response.
However, you should avoid asking for words for fontanelle or molar
teeth in your Wrst elicitation session because people can sometimes feel
embarrassed if they are repeatedly made to demonstrate a lack of
knowledge.
It is more diYcult to predict what sorts of limits there may be on
people’s grammatical knowledge, but the same kind of problem can
arise. When I was working with one of the last speakers of Bandjalang
in northern New South Wales, it became apparent that nouns could be
marked for plural, but that the precise morphological means by which
this was achieved varied according to the semantic type of the noun.
Thus, nouns referring to trees tended to behave one way, while nouns
referring to birds tended to behave in another way, and so on. It also
became apparent that there were quite a lot of exceptions to any of these
generalizations, so it was important for me to test every single noun to
obtain its plural form. It soon became apparent that I was not going to be
able to obtain a plural form for every noun. My language-helper
answered conWdently when she was able to, but at other times looked
somewhat awkward. Eventually, she said ‘Oh, the old people never told
me about that sort of thing’, which I took as a signal that I had possibly
already probed too far for her comfort.
It is important, therefore, for the Weldworker always to show gratitude
for and interest in the data that you have been able to record, and never to
show any kind of disappointment if you have not been able to record as
much as you thought you would be able to. I remember speaking to an
old man in the 1970s who, I had been told, was the last person who knew
anything of the Nganyaywana language of northern New South Wales.
This was a language that linguists were particularly interested in at the
time, as Nganyaywana was thought to be a linguistic isolate, i.e. it was
completely unrelated to all of the other Australian languages. When
I spoke to the old man, he was only able to remember the name of the
language and one or two other words. A decade or so before another
linguist had recorded quite a number of additional words. By the time
I came along, however, his memory had clearly faded and this was
disappointing. However, it was important to express gratitude for the
knowledge that he had been able to share.
While I have heard many stories of linguists having arrived too late to be
able to record information that an elderly last speaker once knew, things
do not necessarily always work out as badly as this. Evans (2001: 262–3)
reports on several diVerent situations involving work on moribund
languages in Australia where speakers’ knowledge of the language showed
signs of improving as elicitation with a linguist progressed. Thus, what
originally may look like an unpromising situation may turn out to oVer
rather more to a patient linguist.
7.3 Legitimate linguistic authority
When a language can claim only a handful of elderly speakers, you might
assume that all of these speakers will exhibit the same kind of interest in
seeing their speech recorded as a basis for your description, and that all
speakers can be equally expected to be able to provide information.
Perhaps a little surprisingly, this is not always the case.
I am currently working towards the documentation of the Tape langu-
age of Malakula in Vanuatu. This is a language which currently has
between ten and Wfteen speakers, with the most conWdent speakers now
in their seventies and eighties. People in this community are very keen for
this language to be documented, and my activities as a linguist are being

strongly supported. People have arranged for me to record a number of
texts from the older members of the community. I found the task of
transcription and further elicitation arising out of these texts quite diY-
cult because the old man who was regarded as the best speaker articulates
his speech in a way that I Wnd quite diYcult to hear. He is also somewhat
hard of hearing, which means that I have to speak quite loudly in order to
be heard. Finally, he Wnds any of my attempts at paradigmatic elicitation
quite frustrating, as he feels that I am simply asking the same word over
and over again. I began to despair at one point of ever succeeding in
producing a description of the language.
It turned out that the son of the old man, who is now in his Wfties, is
also a speaker of the language. He is married to somebody who speaks
a diVerent language and nobody in his household speaks even a single
word of Tape. He considers his own knowledge of Tape to be inferior to
his father’s knowledge, as his father spent his childhood living in an
entirely Tape-speaking community in their original homeland. His father
was part of a migration out of that homeland more than Wfty years ago
into the area where members of this community currently live amongst
speakers of a quite diVerent language.
When I found out that the old man’s son could speak the language,
I asked if he would be willing to help me to transcribe his father’s stories.
He agreed, but when I arrived at his house at the appointed hour to start
work, he was nowhere to be found. I eventually found out that he felt that
his command of the language would be inadequate to the task and that he
was really unwilling to oVer his services because he saw his father
as representing the ‘true’ language. His disappearance to go to his
food-garden was an excuse to avoid disappointing me.
When I explained to him later that all I wanted him to do was help me
to write his father’s words, he agreed to give it a try. It turned out that his
knowledge of the language was more than adequate to the task and we
completed the task together without diYculty. Understanding this man’s
sensitivities, I somewhat gingerly approached the possibility of him record-
ing a story for me in Tape as well. Now that he realized that his know-
ledge was not nearly as inadequate as he had originally thought, he agreed
to give this a try. Eventually, he volunteered several stories, and then
sought out other speakers younger than his father who he thought could
record in the language as well.
There is a second reason why people may express reluctance to be recorded

which relates not to their perceived inadequate command of the language,
but to their lack of a legitimate genealogical connection to members of the
moribund speech community. There is another old man who speaks Tape.
He originates from another language community but as a young man, he
spent a lot of time with speakers of Tape and learned their language. The
result is that he is now one of that very small number of elderly people who
can still conWdently carry out a conversation and tell a story in the language.
However, while people from the Tape community have made various
arrangements for me to record stories from a number of diVerent people,
that old man has never Wgured in those arrangements, nor has he himself
made any request to me to be allowed to tell a story. I once hinted at the
possibility that he be asked to tell a story, but I quickly found people
changing the topic of conversation, seemingly to avoid following up my
suggestion.
I suspect that I will never have the opportunity to record his speech
because, although he speaks Tape as well as anybody else, he is not of
Tape ancestry himself. A signiWcant motivation for community members
in supporting my project to document the Tape language appears to
involve the role that the language plays in demonstrating a connection
with ancestral Tape lands. Because this Tape speaker does not play
a legitimate part in that task, his knowledge of the language appears to
be of little interest to direct descendants of the original speakers of Tape.
7.4 Saving the language?
There can be something of a problem involving salvage Weldwork relating

to the expectations of the community. In a salvage situation, people often
interpret your interest in documenting the language as an implicit
promise to ‘revive’ the language. While in practical terms all you can
do is document the language for future generations, a diVerent
understanding of your intentions (and capabilities) can sometimes
develop within the local community. If you were to fail to deliver on
what might have been seen as some kind of a promise, this has the
potential to result in disappointment, and even anger.
I sometimes Wnd myself dismayed when I encounter stories in the mass
media which claim that some overinXated linguist who claims to be an
international expert on language preservation is about to embark on

a project to ‘save’ some particular dying language, when in fact all that
person is going to do is document the language. It is more than unhelpful
when such linguists claim to be doing precisely what we are not in
a position to do; it is, in fact, unethical.
Any responsible linguist needs to make it absolutely clear to members
of the local community what their intentions are. They have an ethical
obligation to explain to members of the threatened language community
that while salvage descriptions may be of considerable historical interest
to members of the local community, as well as being of interest to
historical linguists and linguistic typologists, we are in no real sense
‘saving’ the language.
This kind of linguistic documentation does nothing more than
conserve data in a language purely for the historical record, eVectively
‘museumizing’ the language. Obviously, museums have their importance
in terms of documenting earlier cultural practices and in conserving
the artefacts associated with those cultures. However, there is no
sense in which such institutions ‘preserve’ cultures or languages as living
systems. While nobody would seriously propose bulldozing Stonehenge
because it derives from a culture that is no longer practised, very few
people would seriously argue (one hopes) that recording something of
a dying language is a completely pointless exercise. Such historical
information is important for local community members, for linguists,
and also for humanity in general—but we must not allow ourselves to
be misled into believing that we are in any way ‘saving’ the language.
Saving a language is a much more complex and much more demanding
task than this.
Some members of the Erromangan community of Vanuatu responded
to the appearance of my description of the moribund Ura language of
their island (Crowley 1999) as allowing for its revival as a spoken
language. However, it is practically impossible to imagine any scenario
by which this could realistically be expected to be achieved. It is unimagin-
able that people on the island could successfully teach themselves the
language from my book, or indeed from any book, given the levels of
functional literacy in the community. Apart from the fact that the
account of the language itself is less than comprehensive, language learning
obviously requires a considerable amount of direct contact with a speaker
(or, ideally, many speakers) of a language.
The most conWdent speaker of Ura is now very old and physically frail,
so any attempt at instituting something like an immersion programme for
young children would be impossibly demanding on this person. Material
resources on the island are extremely limited and the kinds of well-trained
and committed people who would be needed to run relevant community
education projects are extremely thin on the ground. This has meant that
even vocationally oriented programmes that might be seen as giving
people access to means of earning money such as carpentry and
dress-making have often proved diYcult to set up and maintain. It is
diYcult, therefore, to see how people on this island might be able
successfully to implement a language revival programme, especially in
the absence of any kind of local linguistic expertise, along with the
absence of any institutional back-up for such a programme in the local
linguistic ecology.
Of course, the kinds of materials that you end up producing may be
taken up by members of the local community for some kind of language
programme. My publications relating to some of the languages that
I documented in the 1970s from northern New South Wales and far
north Queensland have since been incorporated into locally run language
programmes geared towards members of the local communities. It is
unlikely that these languages will ever again be used in spontaneous
conversation, but the languages do have an important symbolic function
for these communities as expressions of local identity. A knowledge
of these languages also signiWes a historical link to the past, and in
Australian law, this kind of link can be a signiWcant fact in demonstrating
a legal right to owning areas of land.
Further reading: Himmelmann (1998).

References
Austin, Peter. 2006. ‘Data and language documentation’. In Jost Gippert,

Nikolaus P. Himmelmann, and Ulrike Mosel (eds.), Essentials of Language
Documentation, 87–112. Berlin: Mouton De Gruyter.
—— and Terry Crowley. 1995. ‘Interpreting old spelling’. In N. Thieberger
(ed.), Paper and Talk: A Manual for Reconstituting Materials in Australian
Indigenous Languages from Historical Sources, 53–101. Canberra: Aboriginal
Studies Press.
Barley, Nigel. 1983. The Innocent Anthropologist: Notes from a Mud Hut.
Harmondsworth: Penguin Books Ltd.
Boas, Franz. 1911. ‘Introduction’. In Handbook of American Indian Languages.
Bureau of American Ethnology, Bulletin 40, i. 1–83. Washington, DC:
Government Printing OYce (Smithsonian Institute).
Bouquiaux, Luc, and Jacqueline M. C. Thomas. 1992. Studying and Describing
Unwritten Languages. Dallas: Summer Institute of Linguistics.
Bradley, David. 1998. ‘Minority language policy and endangered languages in
China and Southeast Asia’. In Kazuto Matsumura (ed.), Studies in Endangered
Languages, 49–83. International Clearing House for Endangered Languages
Linguistic Studies, 1. Tokyo: Hituzi Syobo.
Bradshaw, Joel. 1994. Review of A Dictionary of Paamese. Oceanic Linguistics
33(1): 257–62.
Chelliah, Shobhana L. 2001. ‘The role of text collection and elicitation in
linguistic Weldwork’. In Paul Newman and Martha RatliV (eds.), Linguistic
Fieldwork, 152–65. Cambridge: Cambridge University Press.
Comrie, Bernard. 1981. Language Universals and Linguistic Typology, 2nd edn.
Chicago: The University of Chicago Press.
Crowley, Terry. 1978. The Middle Clarence Dialects of Bandjalang. Canberra:
Australian Institute of Aboriginal Studies.
—— 1979. ‘Yaygir’. In R. M. W. Dixon and Barry J. Blake (eds.), Handbook
of Australian Languages, i. 363–84. Canberra: The Australian National
University Press.
—— 1981. ‘The Mpakwithi dialect of Anguthimri’. In R. M. W. Dixon and
Barry J. Blake (eds.), Handbook of Australian Languages, ii. 146–94. Canberra:
The Australian National University Press.
—— 1982. The Paamese Language of Vanuatu. Canberra: PaciWc Linguistics.
—— 1983. ‘Uradhi’. In R. M. W. Dixon and Barry J. Blake (eds.), Handbook
of Australian Languages, iii. 306–428. Canberra: The Australian National
University Press.
—— 1992. A Dictionary of Paamese. Canberra: PaciWc Linguistics.
192 references
Crowley, Terry. 1993. ‘Tasmanian Aboriginal language: old and new iden-
tities’. In Michael Walsh and Colin Yallop (eds.), Language and Culture in
Aboriginal Australia, 51–71. Canberra: Aboriginal Studies Press.
—— 1995. A New Bislama Dictionary. Suva: Institute of PaciWc Studies and
PaciWc Languages Unit (University of the South PaciWc).
—— 1998. ‘A salvage sketch of Nāti (Southwest Malakula, Vanuatu)’. In Darrell
Tryon (ed.), Papers in Austronesian Linguistics No.5, 101–48. Canberra: PaciWc
Linguistics.
—— 1998. An Erromangan (Sye) Grammar. Honolulu: University of Hawai’i
Press.
—— 1999. Ura: A Disappearing Language of Southern Vanuatu. Canberra:
PaciWc Linguistics.
—— 2000. An Erromangan (Sye) Dictionary. Canberra: PaciWc Linguistics.
—— 2002. ‘Vinmavis’. In John Lynch, Malcolm Ross, and Terry Crowley, The
Oceanic Languages, 638–49. London: Curzon Press.
Crystal, David. 2000. Language Death. Cambridge: Cambridge University
Press.
Deacon, A. B. 1934 [1970]. Malekula: A Vanishing People in the New Hebrides,
ed. Camilla H. Wedgwood. Oosterhout: Anthropological Publications.
Dimmendaal, Gerrit J. 2001. ‘Places and people: Weld sites and informants’.
In Paul Newman and Martha RatliV (eds.), Linguistic Fieldwork, 55–75.
Cambridge: Cambridge University Press.
Dixon, R. M. W. 1997. The Rise and Fall of Languages. Cambridge: Cambridge
University Press.
—— 2001. ‘A program for linguists’. Unpublished paper presented at the
Fifteenth International Conference in Historical Linguistics. La Trobe University,
Melbourne, 13–17 August 2001.
Donaldson, Tamsin. 1980. Ngiyambaa: The Language of the Wangaaybuwan of
New South Wales. Cambridge: Cambridge University Press.
Dorian, Nancy C. 2001. ‘Surprises in Sutherland: linguistic variability amidst social
uniformity’. In Paul Newman and Martha RatliV (eds.), Linguistic Fieldwork,
133–51. Cambridge: Cambridge University Press.
Duranti, Alessandro. 1994. From Grammar to Politics: Linguistic Anthro-
pology in a Western Samoan Village. Berkeley and Los Angeles: University of
California Press.
—— 1997. Linguistic Anthropology. Cambridge: Cambridge University Press.
Dwyer, Arienne. 2006. ‘Ethics and practicalities of cooperative Weldwork
and analysis’. In Jost Gippert, Nikolaus P. Himmelmann, and Ulrike
Mosel. (eds.), Essentials of Language Documentation, 31–66. Berlin: Mouton
De Gruyter.
Eades, Diana. 1979. The Dharawal and Dhurga languages of the New South
Wales South Coast. Canberra: Australian Institute of Aboriginal Studies.
references 193
Evans, Nicholas. 2001. ‘The last speaker is dead—long live the last speaker!’
Everett, Daniel L. 2001. ‘Monolingual Weld research’. In Paul Newman and
Martha RatliV (eds.), Linguistic Fieldwork, 166–88. Cambridge: Cambridge
University Press.
Faraclas, Nicholas. 1996. ‘New developments in literacy in Papua New
Guinea languages’. In John Lynch and Fa’afo Pat (eds.), Oceanic Studies:
Proceedings of the First International Conference on Oceanic Linguistics,
353–65. Canberra: PaciWc Linguistics.
Foley, William A. (ed.). 1993. The Role of Theory in Language Description.
Berlin: Mouton de Gruyter.
Freeman, Derek. 1983. Margaret Mead and Samoa: The Making and Unmaking
of an Anthropological Myth. Cambridge, Mass.: Harvard University Press.
Gardiner, Margaret. 1984. Footprints on Malekula: A Memoir of Bernard
Deacon. Edinburgh: Salamander Press.
Gerdts, Donna B. 1998. ‘Beyond expertise: the role of the linguist in language
revitalization’. In Nicholas Ostler (ed.), Endangered languages: What Role for
the Specialist?, 13–22. Bath: The Foundation for Endangered Languages.
Gil, David. 2001. ‘Escaping Eurocentrism: Weldwork as a process of unlearning’.
Greenberg, Joseph H. 1966. ‘Some universals of grammar with particular
reference to the order of meaningful elements’. In Joseph H. Greenberg (ed.),
Universals of Language, 2nd edn., 73–113. Cambridge, Mass.: The MIT Press.
—— 1971. ‘The Indo-PaciWc hypothesis’. In Thomas A. Sebeok (ed.), Current
Trends in Linguistics (Vol. 8 of Linguistics in Oceania), 807–71. The Hague:
Mouton.
Hale, Ken. 2001. ‘Ulwa (Southern Sumu): the beginnings of a language research
project’. In Paul Newman and Martha RatliV (eds.), Linguistic Fieldwork,
76–101. Cambridge: Cambridge University Press.
Hercus, Luise. 1969. The Languages of Victoria: A Late Survey (Parts 1 and 2).
Canberra: Australian Institute of Aboriginal Studies.
Himmelmann, Nikolaus P. 1998. ‘Documentary and descriptive linguistics’.
Linguistics, 36: 161–95.
Holmes, Lowell D. 1987. The Quest for the Real Samoa: The Mead/Freeman
Controversy and Beyond. South Hadley, Mass.: Bergin and Garvey Publishers, Inc.
Hyman, Larry M. 2001. ‘Fieldwork as a state of mind’. In Paul Newman and
Martha RatliV (eds.), Linguistic Fieldwork, 15–33. Cambridge: Cambridge
University Press.
Kaplan, Robert B., and Richard B. Baldauf Jr. 2002. Language and
Language-in-Education Planning in the PaciWc Basin. Dordrecht: Kluwer.
194 references
Keesing, Roger. 1975. Kwaio Dictionary. Canberra: PaciWc Linguistics.

King, Michael. 2003. The Penguin History of New Zealand. Auckland: Penguin
Books.
Krauss, Michael. 1992. ‘The world’s languages in crisis’. Language, 68: 4–10.
Kulick, Don. 1995. ‘The sexual life of anthropologists: erotic subjectivity and
ethnographic work’. In Don Kulick and Margaret Willson (eds.), Taboo: Sex,
Identity and Erotic Subjectivity in Anthropological Fieldwork, 219–50. London:
Routledge.
Labov, William. 1972. ‘The logic of nonstandard English’. In Pier Paolo Giglioli
(ed.), Language and Social Context, 179–215. Harmondsworth: Penguin Books
Ltd.
Ladefoged, Peter. 2003. Phonetic Data Analysis: An Introduction to Fieldwork
and Instrumental Techniques. Malden, Mass.: Blackwell Publishing.
Landau, Sidney I. 1989. Dictionaries: The Art and Craft of Lexicography.
Lynch, John. 1998. PaciWc Languages: An Introduction. Honolulu: University of
Hawai’i Press.
—— and Terry Crowley. 2001. Languages of Vanuatu: A New Survey and
Bibliography. Canberra: PaciWc Linguistics.
McConvell, Patrick., and Nicholas Thieberger. 2001, State of Indigenous
languages in Australia—2001, Australia State of the Environment Second
Technical Paper Series (Natural and Cultural Heritage), Department of the
Environment and Heritage, Canberra.
Maddieson, Ian. 2001. ‘Phonetic Weldwork’. In Paul Newman and Martha RatliV
(eds.), Linguistic Fieldwork, 211–29. Cambridge: Cambridge University Press.
Malinowski, Bronislaw. 1929. The Sexual Life of Savages. New York:
Harcourt, Brace and Company.
Mead, Margaret. 1939. ‘Native languages as Weld-work tools’. American
Anthropologist 41: 189–205.
Milroy, Lesley. 1987. Observing and Analysing Natural Language: A Critical
Account of Sociolinguistic Method. Oxford: Oxford University Press.
Mithun, Marianne. 2001. ‘Who shapes the record: the speaker and the linguist’.
Moreno, Eva. 1995. ‘Rape in the Weld: reXections from a survivor’. In Don
Kulick and Margaret Wilson (eds.), Taboo: Sex, Identity and Erotic Subjectivity
in Anthropological Fieldwork, 219–50. London: Routledge.
Mosel, Ulrike, and Even Hovdhaugen. 1992. Samoan Reference Grammar.
Oslo: Scandinavian University Press.
Nettle, Daniel, and Suzanne Romaine. 2000. Vanishing Voices: The Extinc-
tion of the World’s Languages. Oxford: Oxford University Press.
references 195
Newman, Paul, and Martha Ratliff. 2001. ‘Introduction’. In Paul Newman

and Martha RatliV (eds.), Linguistic Fieldwork, 1–14. Cambridge: Cambridge
University Press.
Obeyesekere, Ganarath. 1992. The Apotheosis of Captain Cook: European
Mythmaking in the PaciWc. Princeton: Princeton University Press.
Okrand, Marc. 1992. The Klingon Dictionary: English/Klingon, Klingon/English.
New York: Pocket Books.
Ostler, Nicholas. 1998. ‘What role for the specialist?’ In Nicholas Ostler
(ed.), Endangered Languages: What Role for the Specialist?, 11–12. Bath: The
Foundation for Endangered Languages.
Payne, Thomas E. 1997. Describing Morphosyntax: A Guide for Field Linguists.
Rae-Ellis, Vivienne. 1981. Trucanini: Queen or Traitor? Canberra: Australian
Institute of Aboriginal Studies.
—— 1988. Black Robinson: Protector of Aborigines. Melbourne: Melbourne
University Press.
Rice, Keren. 2001. ‘Learning as one goes’. In Paul Newman and Martha RatliV
(eds.), Linguistic Fieldwork, 230–49. Cambridge: Cambridge University Press.
—— 2005. ‘The linguist’s responsibilities to the community of speakers’.
Conference on Language Documentation: Theory, Practice, and Values.
LSA Linguistic Institute, 9–10 July 2005.
—— to appear. ‘Ethical issues in linguistic Weldwork: an overview’. To appear in
W. van den Hoonaard (ed.), The Ethics Trapeze. Vancouver: University of British
Columbia Press.
Ryan, Lyndall. 1996. The Aboriginal Tasmanians, 2nd edn. Crows Nest (NSW):
Allen & Unwin.
Samarin, William J. 1967. Field Linguistics: A Guide to Linguistic Field Work.
New York: Holt, Rinehart & Winston.
Schieffelin, Bambi B., and Elinor Ochs (eds.). 1986. Language Socialization
across Cultures. Cambridge: Cambridge University Press.
Shopen, Timothy (ed.) 1985a. Language Typology and Syntactic Description,
i: Clause Structure. Cambridge: Cambridge University Press.
—— (ed.) 1985b. Language Typology and Syntactic Description, ii: Complex
Constructions. Cambridge: Cambridge University Press.
—— (ed.) 1985c. Language Typology and Syntactic Description, iii: Grammatical
Categories and the Lexicon. Cambridge: Cambridge University Press.
Sutton, Peter, and Michael Walsh. 1979. Revised Linguistic Fieldwork
Manual for Australia. Canberra: Australian Institute of Aboriginal Studies.
Thieberger, Nicholas. 2004. ‘Topics in the grammar and documentation of
South Efate, an Oceanic language of Central Vanuatu’. Ph.D. dissertation,
University of Melbourne.
196 references
Vaux, Bert., and Justin Cooper. 1999. Introduction to Linguistic Field Methods.
Munich: Lincom.
Whaley, Linday J. 1997. Introduction to Typology: The Unity and Diversity of
Language. Thousand Oaks, Calif.: SAGE Publications.
Williams, C. J. 1980. A Grammar of Yuwaalaraay. Canberra: PaciWc Linguistics.
Index
accessibility 38, 40, 112 background reading 59, see 3.3.1, 101
accommodation 69, 152–3, 173 Bahasa Indonesian (language) 58
Africa 58, 93, 105 Baldauf Jr., Richard B. 48
Amazonia 8, 11 Bandjalang (language) 9, 13, 39, 129,
Ambrym 177 181, 183–4
Americas (the) 8–9, 12, 19, 58, 180 Bantu languages 59
Amerindian languages 58, 121 Barley, Nigel 73
Amery, Rob 50 Bislama (language) 35–7, 43, 46, 64–5,
animacy hierarchy 118 124, 155, 157–62; see also
anthropology 20, 33, 112, 133, Melanesian Pidgin
141–2, 171 BloomWeld, Leonard 12
ethnographer 19, 141, 171 Boas, Franz 19, 121
Weldwork 19 body parts 96–7, 105
linguistic anthropologists 17 borrowings 50, 90, 100, 125
archive 29, 53, 74–6, 80, 101–2, Bouquiaux, Luc 32, 85–6, 101,
see 4.5, 124 104–5, 111
access 119 Bradley, David 48
DELAMAN 120 Bradshaw, Joel 40
OLAC 120 Brazil 58
Asia 59 bureaucracy 70–3
Austin, Peter 64, 76, 85, 117, 120
Australia 6, 8–9, 11–12, 41, 80, 88, Campbell, Lyle 27, 180
105, 181, 189 Canada 8, 41
Australian Aborigines 2–3, 179 Caribbean 8
Australian languages 6, 8, 91, 101, case markers 7
179, 183, 185 Caucasus (the) 11
New South Wales languages 8, 39, Caucasus (the) languages 59
179, 185, 189 Chatham Islands 6
Queensland languages 189 Chelliah, Shobhana L. 131
South Australian languages 50 China 58
Tasmanian languages 2–6, 122 Chomsky, Noam 12
Victorian languages 8 clauses 100
Austronesian languages 58–60, 86 clothes 77–8; see also equipment
Avava (language) 54 commodiWcation of research 35, 51
198 index
community assumptions 28, 83, 153, group sessions 106–7

165, 187 monolingual 94, 155
community attitudes 159–60, 179, prompts 108, 183
180, 187 texts 15, 104, 108, 121, 129
community responsibility 34, 128, Deacon, A. Bernard 166–9, 171–2
138, 173, 188 descriptive adequacy 10–11
Comrie, Bernard 12 descriptive methodology 19–21
consent 28–9 dictionary 10, 36–7, see 2.2.2, 53, 104,
context 103, 130 111, 129, 151
Cooper, Justin 32, 85–6, 111, 176 audience 40–4
Cornish (language) 5 bilingual 38, 41, 43
corpus 17–18, 121 citation forms 39, 42
corpus-based research 18 illustrative sentences 41
Crowley, Terry 3, 8, 9, 11, 13, layout 39, 42
37, 39–40, 43, 63–4, 80, 118, lexical entry 111
178, 188 monolingual 43
Crystal, David 8 problems 38–40
culture 20–1, 95–6, 99, 105–6, 128–9, digital database 42
163–4, 168, 170, 172, 175 Dimmendaal, Gerrit J. 30, 32
culture shock 161–4 discourse 109, 128
Dixon, R. M. W. 12, 16–17, 56
data Donaldson, Tamsin 8
analysis 54, 75, 95, 129, 135, see 6.1 Dorian, Nancy C. 18
audio 17 Dravidian languages 59
grammatical 97, 100, 113 Duranti, Alessandro 17
lexical 113–16 Dwyer, Arienne 56
natural 13, 17 dying languages 9, 70, 132, 155 see
organization 66–7, see 4.4 Ch 7; see also language loss,
primary 53–4 minority languages
reliability 117, 133–5, 139, 142,
144–6, 183 Eades, Diana 8
sensitive 25, 119 electricity 74, 77
storage 54, 118 elicitation, see data collection
video 17 English (language) 4, 52, 93, 100,
data collection 18, 66, see 4.2 106–7, 122, 124, 127–8, 133, 146,
and 4.3, 141 156, 160, 183
bilingual 93 equipment 94, 107, 131, 138, 154; see
dialogue 15 also recording; clothes; food;
elicitation 15, 17, 87, 90–2, see 4.2 microphone
and 4.3, 103–4, 112, 127–9, 152, Erromangan (language) 63, 81, 89,
182, 186 129, 147–8, 156–7, 160
index 199
Erromango (island) 31, 35–7, 54, Goldsmith, Michael 19

78–9, 81, 83, 89, 148, 156–7, 164, graduate student 65, 81, 91, 165,
173, 188 177–8; see also PhD dissertation
ethics see Ch. 2, 70, 118, 120, 126, 130, grammar 10, 18, 53, 63, 104, see 2.2.1,
180, 184, 188 122, 129, 177
Ethiopia 30 grammaticality judgement
Evans, Nicholas 182, 185 102–4, 136
Everett, Daniel L. 154 grants & funding 24, see 3.5, 119
examples 17 Greenberg, Joseph H. 6–7
Gumbaynggir (language) 179
Faraclas, Nicholas 49
Weld linguist Hale, Ken 80, 104, 111, 155
behaviour see 6.5 health 68, 78–9
gender 90, 164 Hercus, Luise 8
types of 12–13, 57–8, 178 Himmelmann, Nikolaus P. 189
Field Methods course 13, 59 Hindi (language) 58
Weld notes see 4.2, 112–13, 117, 137 Hobart 1, 3
errors see 5.2 Holmes, Lowell D. 19
Weldsite 59–61, 78, 80, 84 Hovdhaugen, Even 17
Weldwork 13, 15, 57 human language 6–7, 11–12
experience 161–4 Hyman, Larry M. 16
tradition 12
goals 18, 65–6 idioms 109
preparation 14, 53, 64, see 3.4, 95, India 11, 58, 71
122, 152, 162 Indo-Aryan languages 59
single speaker 14–15, 63, 152 Indonesia 11, 58, 71
types of 14, 16 informant, see language-helper
Fijian (language) 50 interpreter 93–4
Foley, William A. 13 intonation 103
food 77–8; see also equipment Iranian languages 59
Freeman, Derek 19
French (language) 7, 58, 64–5 Kaplan, Robert B. 48
Kaurna (language) 50–1
Gaelic (language) 18 Keesing, Roger 39
Gardiner, Margaret 168, 171 Khoesan languages 59
Gela (language) 36 King, Michael 6
gender roles 164 Klingon (language) 6
Gerdts, Donna B. 179 Krauss, Michael 9
gesture 131 Kulick, Don 172
Gil, David 13, 154 Kwaio (language) 39
200 index
Labov, William 144–7 Melanesia 93

Ladefoged, Peter 18 Melanesian Pidgin (language) 58, 93;
land rights 9, 189 see also Bislama
Landau, Sidney I. 151 metadata 112, 124, 126, 135
language and identity 180–1, 189 Mexico 58, 180
language documentation 10, 16, 52 microphone 73, 75–6, 123; see also
language family 58–9 equipment; recording
language-helper see 4.1, 95, 98, 101, Milroy, Lesley 18
104, 110 minimal pairs 97, 110, 147
ability 184–5 minority languages 49; see also
choice of 89–91, 129, 134, 138–9, language loss, dying languages
164, 181–2 Mithun, Marianne 104
reactions 87–8, 180, 185–7 money 31, 50, 165–6; see also payment
language learning 19–20, 60, 64–5, Mongolic languages 59
83–4, 94, see 6.2 Mon-Khmer languages 58
language loss 1, 5, 8–9, 18–19, 51; see also Moreno, Eva 90
dying languages, minority languages Moriori (language) 6
language maintenance 48–51, 56 morphology 98
language planning 48, 50–1 Mosel, Ulrike 17
language policy 48–9 Mpakwithi (language) 9, 72, 181
language relatedness 6, 101
language revival 187–9 Naman (language) 127, 137
language shift 51 Nāti (language) 80, 178
lingua franca 64, 93–4, 96, 105, 112, native-speaker intuitions 41, 140
134, 155, 158, 160 native-speaker linguist 55–6
linguistic training 58 natural speech, see spontaneous
literacy 26, 28, 44, 49, 80, 122, 140 speech
Lynch, John 11, 51, 118 Nauruan (language) 86
negative 98
Madagascar 59 Nettle, Daniel 8
Maddieson, Ian 18 Neve‘ei (language) 45–6, 63, 158–9,
Malakula (island) 30, 80, 89, 106, 166–8
122–3, 127, 137, 158–9, 166, 168, Newman, Paul 86
170–1, 174, 179, 185 New Caledonia 65
Malinowski, Bronislaw 19 New South Wales 9, 152, 184
Manchu-Tungusic languages 59 New York City 144
Mandarin (language) 58 New Zealand 6, 32, 41, 121, 164
Maori (language) 6, 121–2 Nganyaywana (language) 185
McConvell, Patrick 8–9 North American languages 8, 59
Mead, Margaret 19–20, 133, 141 noun classes 7
index 201
Obeyesekere, G. 171 publication 35, 53, 67, 104, 117,

Oceanic languages 60, 86, 93, 101 120, 166
Ochs, Elinor 18 publisher 37, 39, 59
Okrand, Marc 7
oral tradition 26, 36, 47, 80, 121, 128 Queensland 9, 72, 88
Orkon (language) 177–8
orthography 42; see also spelling Rae-Ellis, Vivienne 3
systems RatliV, Martha 86
Ostler, Nicholas 48 recording 53–4, see 3.7.1, see 5.1.2,
130, 142; see also equipment;
Paama (island) 65, 67, 71, 81–3, 115, microphone
157–8, 162–3 audio 73–4, 101–2, 119, 131
Paamese (language) 13, 40, 44–5, 50, digital 119 122
60, 81, 100, 118, 125, 127, 132, environment 124, 143
136, 155–62 format 119
Palawa Kani (language) 4 video 73, 131
Palawa languages 3–5 relationships
Papua New Guinea 5, 11, 32–3, 50, 55, colleagues 52
58, 64, 88, 176 community 34, 38, 71, 83–4, 105–6,
Papuan languages 58–9 118, 124, 127, 154, 163, 165, 169,
paradigms 88, 91–2, 97–8, 128, 182, 186 172, 175
participant observation see 5.4, 165, native speakers 32, 89, 96, 105, 125,
171, 175 172, 184–5
Observer’s Paradox 144, 146 research outputs 34, 44
payment 30–1, 32, 69, 72, 89, 173, 179; Rice, Keren 56, 86, 154
see also money Romaine, Suzanne 8
Payne, Thomas E. 21, 151 Russia 58
PhD dissertation 13, 65–7, 129, 162, Russian (language) 58
177–8; see also graduate student Ryan, Lyndall 2
phonemics 96–7, 110
phonetics 63, 131, 137 Samarin, William J. 32, 68, 84–5, 95,
phonology 96, 183 101, 104, 111, 113, 117, 129, 131,
phonotactics 96, 110 133, 137, 157, 161
Pijin (language) 39 Samoa 19–20, 133, 141
Pike, Kenneth 155 Samoan (language) 17, 19
politics 83–4 Sapir, Edward 12, 19
polysemy 106–7, 109 SchieVelin, Bambi B. 18
Portuguese (language) 58 semantics 99–100, 136, 184
possession 97, 101 semantic Welds 98, 105, 107–9, 114–15
pragmatics 15, 63 sentences 99–100, 104, 128
pronouns 98 Shopen, Timothy 21, 151
202 index
Sino-Tibetan languages 59 Tok Pisin 64–5

Solomon Islands 36, 39, 58 transcription 53, 92, 104, 109, 113,
South-East Asian languages 58 117, 126, 129–30, see 5.3
Spanish (language) 7, 58, 64, 93, 180 errors 135, 137
spelling systems 44–5; see also translation equivalents 107, 136
orthography travel 69, 78–9
spontaneous speech 90–1, 109, 121, Truganini 1–5, 8
127–9, 147, 153
dialogues 130–1, 155 United States of America 8, 41, 144–5
formal speech 130 Ura (language) 35–7, 51, 139, 183,
monologues 130 188–9
varieties 143, 182 Uradhi (language) 9, 72
stationery 76–8, 94, 113, 115, 135 Utaha (language) 54
stories 25–7, 36, 47, 90–2, 109, 124–8,
138–9, 182 Vanuatu 10–11, 13, 18, 26, 29, 30, 34–6,
topics 126–7, 129 44–6, 54–5, 58, 60, 64–5, 67, 71, 88,
Summer Institute of Linguistics 118, 122–4, 126, 132, 139, 141, 147,
(SIL) 46, 81 153, 156, 163–4, 168, 171, 173–4,
Sutton, Peter 105 178, 183, 185, 188
Swahili (language) 93 Vanuatu Cultural Centre 24, 36, 89
Sye (language) 35–7, 183 Vanuatu languages 10, 18
Vaux, Bert 32, 85–6, 111, 176
TAM (tense/aspect/mood) 97–8 verbs 97–8, 128, 152
Tape (language) 185–7 serial verb constructions 127–8, 132
Tasmania 1, 3–4, 8, 57 Vinmavis (village) 158–9, 168–72
teaching material 9, 44, 46, 80 vocabulary 92, 107, 109–11, 183–4;
technology 17, 42, 73–4, 117, 122 see also wordlist
digital 74
Toolbox 113 Walsh, Michael 105
terminology 17 Welsh (language) 4
text compilations 38, 47–8, 53, 109, West African languages 58
117–18, 128, 137, 140 Whaley, Linday J. 21, 151
theory 13 Williams, Corinne J. 8
Basic Linguistic theory 17 word order 6–7, 99, 183
theoretical bias 17 wordlist 64, 95–8, 105–6, 135; see also
theoretical framework 16–17 vocabulary
transformational model 16 Wurm, Stephen 155
structuralism 16
Thieberger, Nicholas 8–9, 17 Yaygir (language) 9, 181
Thomas, Jacqueline M. C. 32, 85–6,
101, 104–5, 111 Zapotec (language) 93

Field Linguistics

Uploaded by

Copyright:

Available Formats

Field Linguistics

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Field Linguistics

Uploaded by

Copyright:

Available Formats

What topics does the book cover?

What topics does the book cover?

What are some of the challenges of conducting fieldwork?

What are some of the challenges of conducting fieldwork?

Field Linguistics

This page intentionally left blank

ISBN 978-019-928434-4 (hbk.)

1 Field Linguistics: Why Bother? 1

4 Gathering your Data 85

4.3.5 When is the dictionary done? 111

5 Beyond Elicitation 121

6 Problems and Pitfalls 151

7 Salvage Fieldwork 177

A huge number of the world’s languages remain poorly described, or even

relationships in the Weld while sidestepping crucial issues of data-gather-

failures of others for as long as I have been committing errors of my own,

Terry Crowley’s death in early 2005 left behind a number of works in

The publishers wish to express their gratitude to Dr Nicholas Thieberger

1.1 The threat to diversity

No linguist can fail to be aware that during the twentieth century

1.1.1 Some lessons from Truganini

Many of my readers have possibly never have heard of Truganini. Some

ethnic separation—perhaps we should call this early apartheid or ethnic

while a historically legitimate vocabulary can be reconstructed to some

1.1.2 The value of records

It didn’t have to be this way, of course. If only a selection of diVerent

puq legh yaS

1.1.3 Diversity still undocumented

Tragic as the circumstances are surrounding the loss of Truganini’s

The task of fully documenting even a single language is enormous. In

1.2 The ‘Big Questions’

Linguistics could not be linguistics without linguistic descriptions. If we

1.2.1 Armchairs and dirty feet

Armchair linguists often publish quite extensively without necessarily

his Wrst academic appointment there in 1970. In Australian universities, the

1.2.2 Fieldwork at home and Weldwork in The Field

There is a kind of halfway house in linguistic Weldwork between sitting in

While this is clearly a kind of Weldwork, it is not really the kind of

I washed my clothes inst soap.

With third person singular inanimate noun phrases, the pronoun is

I washed my clothes inst.

This is the soap that I washed inst my clothes.

This construction would probably never appear in sentences translated

This is the soap that I wash my clothes inst.

varied set of data. A more pure-minded Weldworker might also want to

Further reading: Hyman (2001).

1.3 What kind of linguistics?

Some descriptive linguists deWne basic linguistic documentation as

The ideal apprenticeship [for a linguist] is to undertake Weldwork on some

There are probably as many diVerent ways of describing the phonology,

The most readable—and arguably, therefore, the most valuable—

account—From Grammar to Politics—implies this priority. Similarly,

1.4 Anthropological and linguistic fieldwork

The threat to the world’s linguistic diversity to which I have already

In North America in the late nineteenth century, people realized that

Margaret Mead wrote of the importance of a knowledge of the local

new headings in order to accommodate particular features of the

2.1 Linguists and speakers