Professional Documents
Culture Documents
Series Editors
VOLUME 6
By
Nancy C. Dorian
LEIDEN | BOSTON
Dorian, Nancy C.
Small-language fates and prospects : lessons of persistence and change from endangered languages :
collected essays / By Nancy C. Dorian.
pages cm. — (Brill’s Studies in Language, Cognition and Culture ; 6)
Includes index.
ISBN 978-90-04-23051-4 (hardback : alk. paper) — ISBN 978-90-04-26193-8 (e-book) 1. Endangered
languages. 2. Scottish Gaelic language—Dialects—Scotland. I. Title.
P40.5.E532S36 2014
491.6’3—dc23
2014002474
This publication has been typeset in the multilingual ‘Brill’ typeface. With over 5,100 characters covering
Latin, ipa, Greek, and Cyrillic, this typeface is especially suitable for use in the humanities.
For more information, please see brill.com/brill-typeface.
issn ����-����
isbn ��� �� �� ����� � (hardback)
isbn ��� �� �� ����� � (e-book)
Foreword ix
Sources xi
Introduction 1
part 1
Language Change in an Obsolescent Language 31
part 2
Speaker Skills and the Speech Community in a Receding
Language Context 135
part 3
Language Shift and Language Maintenance 203
part 4
Language Use 309
part 5
Fieldwork: Methods, Problems, Insights 369
Author Index 445
General Index 449
Nancy Dorian is a star. She is the founder of at least one burgeoning field within lin-
guistics. She has made a lasting impact on a language community, and has published
much of lasting value on many topics to do with the fate of endangered languages and
language shift, and on what can be done about this.
Language endangerment, language obsolescence, shift and loss are among the most
prominent concerns of today’s linguistics, especially sociolinguistics. Nancy Dorian
put them on the map. Her book Language death: the life cycle of a Scottish Gaelic dialect
(University of Pennsylvania Press, 1981) was the first systematic investigation of a lan-
guage on the way out – including intergenerational variation between speakers, their
insecurity, loss of proficiency, and the irreversible changes under the influence of the
majority language (English). This book became an instant classic, and so did Nancy
Dorian. Investigating obsolescence: studies in language contraction and death
(Cambridge: Cambridge University Press, 1990), edited by Nancy Dorian, opened com-
pletely new perspectives on how languages may retreat, and contract, as they gradually
fall into disuse.
Nancy Dorian has had an outstanding career in linguistics. She has published many
books, and several score articles – all of them highly influential, and much quoted. She
has done an immense amount for the Scottish Gaelic community she has been work-
ing with – helping maintain the language, and understand the why and the how of its
dynamics. In 2012, Nancy was the recipient of the prestigious Kenneth Hale award, by
the Linguistic Society of America, for her research on Scots Gaelic ‘that spans a period
of almost fifty years, perhaps the most sustained record of research on any endangered
language, and for her effective advocacy for the cause of endangered language preser-
vation and revitalization’. Nancy Dorian’s voice in support of minority and endangered
languages was one of the earliest, and continues to be one of the most prominent.
Nancy’s illustrious career has in many ways shaped the profile of modern linguistics
in the true sense. As the editor of the section on ‘Small languages and small language
communities’ of the International Journal of the Sociology of Language, Nancy Dorian
helped so many scholars and fieldworkers to openly talk about the issues faced by the
communities they work in. Through her inspiring work, and her warm and engaging
personality, Nancy Dorian managed to bring together scholars from different parts of
the world, getting them to make joint discoveries and work on similar topics, making
their research richer and more interesting. Nancy Dorian has served as a source of
inspiration, and as an informal mentor to many – including us.
To have a special volume of Nancy Dorian’s papers – many of them classics, but not
all of them easy to locate – has been something we have always wished for. This vol-
ume is it. Here we have twenty three papers, and an introduction, by Nancy, which
brings them all together, as the culmination of her life’s work. Each of these is a para-
mount achievement in the areas of endangered languages, language variation, lan-
guage shift and maintenance, and the methodology and practice of linguistic fieldwork.
It is plainly an honour for us to have this volume within our monograph series.
Few if any linguists have displayed such insight, and brilliance as Nancy Dorian. She
remains an admirable role model, and the source of motivation and encouragement
for all real linguists. Colleagues like Nancy make our discipline worthwhile.
Alexandra Y. Aikhenvald
R.M.W. Dixon
N.J. Enfield
Chapters 1–23 have been previously published. The current versions have undergone
varying amounts of revision.
Chapter 3, ‘Making do with Less: Some Surprises along the Language Death Proficiency
Continuum’ was published in Applied Psycholinguistics 7: 257–276, 1986, used with
permission.
Chapter 5, ‘The Problem of the Semi-Speaker in Language Death’ was published in the
International Journal of the Sociology of Language 12: 23–32, 1977, used with permission.
Chapter 7, ‘Defining the Speech Community to Include its Working Margins’ was
published in Susanne Romaine (ed.), Sociolinguistic Variation in Speech Communities,
London: Edward Arnold, 1982, pp. 24–33, used with permission.
Chapter 9, ‘Age and Speaker Skills in Receding Language Communities: How Far do
Community Evaluations and Linguists’ Evaluations Agree?’ was published in the
International Journal of the Sociology of Language 200: 11–25, 2009, used with
permission.
Chapter 10, ‘Linguistic Lag as an Ethnic Marker’ was published in Language in Society
9: 33–41, 1980, used with permission.
Chapter 11, ‘Language Loss and Maintenance in Language Contact Situations’ was
published in Richard Lambert and Barbara F. Freed (eds.), The Loss of Language Skills,
Rowley, Mass: Newbury House, 1982, pp. 44–59, used with permission.
Chapter 12, ‘The Value of Language-Maintenance Efforts which are Unlikely to Succeed’
was published in the International Journal of the Sociology of Language 68: 57–67, 1987,
used with permission.
Chapter 14, ‘Purism vs. Compromise in Language Revitalization and Language Revival’
was published in Language and Society 23, 4: 479–94, 1994, used with permission.
Chapter 16, ‘Bi- and Multilingualism in Minority and Endangered Languages’ was
published in Handbook of Bilingualism, Tej K. Bhatia and Wm. C. Ritchie (eds.), Oxford:
Blackwell Publishers, 2004, pp. 437–459, used with permission.
Chapter 18, ‘Telling the Monolinguals from the Bilinguals: Unrealistic Code Choices in
Direct Quotation Within Scottish Gaelic Narratives’ was published in the International
Journal of Bilingualism 1: 441–54, 1997, used with permission.
Chapter 19, ‘Celebrations: In Praise of the Particular Voices of Languages at Risk’ was
published in Ogmios. Newsletter of the Foundation for Endangered Languages 12: 4–14,
1999, used with permission.
Chapter 20, ‘Gathering Language Data in Terminal Speech Communities’ was published
in The Fergusonian Impact. Volume 2. Sociolinguistics and the Sociology of Language,
Joshua A. Fishman et al. (eds.). Berlin: Mouton de Gruyter, pp. 555–575, 1986, used with
permission.
Chapter 23, ‘The Private and the Public in Language Documentation and Revitalization’
was published in New Perspectives on Endangered Languages, José Antonio Flores
Farfán and Fernando Ramallo (eds.), Amsterdam: John Benjamins, 2010, 29–47.
1 Early Days
at-risk language for many years; yet in a more general sense they also reflect
the development of what has meantime become known as the field of lan-
guage endangerment. Linguists, sociologists, linguistic anthropologists, and
missionary linguists were by the second half of the twentieth century increas-
ingly engaged with the many speech forms in various parts of the world that
had few or no young speakers and so could be viewed as potential candidates
for language loss. As a result, a body of research dealing with the condition of
those speech forms, the consequences of the likely loss of some (or even many)
of them, the prospects of maintaining or revitalizing them, and the feasibility
of documenting them had begun to emerge. The overall and individual volume
introductions to Austin and McGill’s four-volume Endangered Languages col-
lection offer a good account of this development, a development so recent that
a preponderance of the papers included date only from the previous twenty
years (Austin and McGill 2011).
During this same period my own East Sutherland fieldwork interests
expanded to encompass a number of perspectives well beyond those of my
original Survey-related work. I continued to work with East Sutherland Gaelic
periodically during the rest of the 1960s and all of the 1970s, and although
there followed a health-related hiatus of a little more than a decade, fieldwork
resumed in 1990 and continued until 2010. (It continues still, for that matter, but
to an extent sadly limited by the loss of nearly all local speakers.) The resumed
“fieldwork” was made possible by tape recordings and long-distance telephone
calls and also in a number of happy instances through visits paid me in the U.S.
by East Sutherland Gaelic speakers. The descriptive and dialectological focuses
of the earliest work were soon joined by two more focuses that came to the fore
in the 1970s. The first of these was a focus on oral history and traditional lore, a
line of inquiry that arose from a need to understand the circumstances of fish-
erfolk life and the particular cultural environment of these communities. The
second was a focus on the linguistic changes that could be detected as the role
of Gaelic altered within the former fishing communities, passing from domi-
nant language among the oldest of the bilinguals to only partially controlled
language among the youngest. Evidence of age-related proficiency differences
among speakers led in turn to interest in the ways in which the least proficient
speakers made use of their language skills and continued to function as mem-
bers of the bilingual community.
In the 1990s another two focuses emerged. One was the documenting of dis-
tinctive forms of East Sutherland Gaelic language use, something that came to
seem ever more important as fluent command of this speech form declined
and the prospects for its future continued to dim. The other was recognizing
siblings to share a house with one another. All of these arrangements were to
be met with among the people who became my sources, and close consan-
guinial and affinal kin ties produced visiting patterns that often added drop-in
Gaelic visitors to fieldwork sessions (ch. 21). As time went on I was exposed to
the usage of a relatively large number of speakers, which in turn led to a grow-
ing awareness on my part of differences that emerged both in responses to
elicitation tasks and in ordinary conversation and corresponded by and large
to the ages of the speakers. Here, it seemed, was a ready-made environment for
examining linguistic change in progress where it had not often been examined
up to that time, namely in a speech form that was no longer being transmitted
and so was likely soon to pass out of use.
By the 1960s, there was nothing left of the fishing industry that had been
a major presence on the east coast of Sutherland in the 19th and early 20th
centuries. Between the two World Wars fishing had gone into decline in Brora,
Golspie, and Embo, and after the second World War it disappeared. Brora and
Golspie still had distinct residential areas that had previously housed only
fisherfolk and still housed many fisherfolk descendants (Lower Brora in the
former, the West End in the latter), and Embo – though with a much reduced
population as compared with the early years of the twentieth century – still
retained the fishing village layout it had previously had, with a series of straight
parallel streets of two-family houses running down to the sea. Fishing had kept
the bilinguals of the three villages poor, separate, and stigmatized. The circum-
stances that brought each of these characteristics into being are discussed in
a number of the studies included here, so for present purposes it can suffice
to say that Gaelic-speaking parents increasingly, and often quite consciously,
declined to speak Gaelic with their children or grandchildren, even though
most of them continued to speak Gaelic habitually with their contemporaries
and their elders (Dorian 1981: 83, 104–06; ch. 16). That is, Gaelic was fading from
the scene not by disuse among fluent speakers, but by fluent speakers’ fail-
ure to use Gaelic regularly with their children or require Gaelic from them.
Remarkably, some of those children acquired a degree of Gaelic all the same,
through their own interest and effort. Although they did not reach their par-
ents’ level of proficiency, or in some cases their older siblings’ level of profi-
ciency, this resistance to abandonment of a stigmatized language from within
a stigmatized group was itself a matter of research interest (ch. 3, ch. 6; Dorian
1981: 109,112).
In a language-shift setting such as this, it seemed that any number of ques-
tions might be raised about age-related proficiency differences in the local
Gaelic. For example, how systematic were they? Could an age-based proficiency
3 Assembling a Database
For me as a fieldworker interested in these questions and others like them, the
most immediate challenge lay in assembling a body of data that could offer
answers to some of them. My original evidence that such questions might be
raised and answered in these villages came both from elicitation (translation
tasks posed to speakers of various ages) and from exposure to conversation
in multi-speaker environments. What I most needed in order to gain a better
perspective on these matters was a large body of directly comparable mate-
rial, something that could not be achieved in this setting through freely spoken
material alone. (There were for example no origin myths, traditional stories,
disaster accounts, or the like that a variety of speakers might be expected to
reproduce along very similar lines.) The best prospect for achieving compa-
rability was to present the same set of sentences for translation to an age-
differentiated sample of speakers, and from 1970 onwards I set about doing
just that (while of course continuing to gather freely spoken material as well).
In 1974 and 1976 in particular, I tried to establish the parameters of discernible
language change via translation tasks presented to four older fluent speakers
from all three villages, four younger fluent speakers from Embo (where more
subtle age-related differences were in evidence than in Brora and Golspie), one
formerly fluent speaker from Embo (1974 only), and from 6 to 8 demonstrably
imperfect speakers from Brora and Embo. The 120 sentences of the 1974 transla-
tion battery and the 150 sentences of the 1976 translation battery were designed
to elicit structures that would set elements of ongoing change off against ele-
ments of stability in the local Gaelic (chs. 1–4; chs. 5–9; Dorian 1981: 117–21).
In this work I had two major advantages. One was that by 1974 I had been
working with East Sutherland fisherfolk Gaelic for over a decade and spoke the
local dialect myself, although with some mixture of Golspie and Embo forms
(and also, of course, with some anglicisms and other mistakes). This made it
possible for me to design my elicitation sentences with an eye to highlight-
ing change. The other advantage was that the speakers I worked with were
highly adept translators. This came about in good part because local prac-
tice in reported speech was to use the language expected by the interlocu-
tor rather than the language in which a remark was originally made (ch. 18).
With language shift underway, kin networks in these villages included both
older people who were bilingual and younger people who were monolingual
in English. When individual speakers passed along in conversation any remark
they thought would interest someone else, they would frame the remark in
whichever language they normally used with that particular conversation part-
ner, even if the original remark had actually been made in the other language.
Of course this very frequent conversational translation was more spontaneous
and less self-conscious than formal translation tasks, but the general effect of
these reported speech habits was an ease with translation that made elicita-
tion unproblematic for a good many bilingual speakers.
There was a great advantage in being aware in advance of certain character-
istics of local speech when designing elicitation tasks. Knowing, for example,
that diminutives were exceptionally freely used in the local Gaelic made it pos-
sible to add an important check for nominal gender to the elicitation batteries.
A single test sentence such as “The wee glass fell and it broke” could provide a
triple check for feminine gender: the initial consonant of the noun (which dif-
fered after the article according to gender in certain phonological classes), the
replacement of the pronoun in the second clause, and the diminutive suffix
added to the noun ‘glass’. For conservative older speakers, the same sentence
with a traditionally masculine noun such as ‘bowl’ would show differences in
all three respects. As in the earlier dialect-descriptive work, the availability
of a relatively large speaker sample was essential, and I enlarged my pool of
sources accordingly; but in this subsequent work it was the inclusion of low-
proficiency speakers that was particularly important.
compare with that of 20 older native speakers. He described the weak Breton
of “terminal” speakers as pidginized and the only slightly less weak Breton of
“preterminal” speakers as creolized, but in studying their Breton output he was
able to identify certain word-formation processes that appeared particularly
likely to be lost by imperfect speakers. In a study of East Sutherland fisherfolk
Gaelic, also including imperfect speakers and likewise undertaken in the early
1970s (ch. 2), I was able to compare the morphologically rich plural and gerund
formations produced by fully fluent speakers with those produced by imper-
fect speakers and determine that while simplification appeared in imperfect-
speaker forms, it was very much less extreme than might have been anticipated
(and far short of the morphological simplification typical of pidgins). On the
basis of responses to the elicitation batteries of 1974 and 1976, in fact, I con-
cluded that East Sutherland fisherfolk Gaelic could be said to be “dying . . . with
its morphological boots on” (ch. 2); Thomason (2001: 236) later identified a
similar and still more striking case involving Montana Salish.
Although an age-and-proficiency continuum proved identifiable in East
Sutherland fisherfolk Gaelic, what factors other than age as such were involved,
especially in the lower reaches of the continuum, was difficult to determine.
What was clear was that there were some less than fully fluent speakers within
these enclaved communities of Gaelic-English bilinguals, capable of speaking
Gaelic and viewed by themselves and others as members of the fisherfolk pop-
ulation, who produced a Gaelic that was not considered the equal of the Gaelic
spoken by fully fluent members of the speech community (see the articles in
section II). Some of the imperfect speakers (those identifiable as “formerly flu-
ent speakers”) had been fully fluent speakers into adolescence, but they had
married English monolinguals and lived elsewhere for long periods, seldom
having the opportunity to use Gaelic. At the time when I worked with them,
several of them were in various stages of recovering their Gaelic. Other imper-
fect speakers had lived elsewhere for briefer periods and then returned to their
home villages; some of these speakers were also married to monolinguals. A
few had lived locally all along yet also spoke at less than fluent levels in the
1960s and ’70s; some of these were younger children in families where older
children had introduced more and more school-learned English into the home.
Fully fluent speakers in Embo explicitly identified the imperfect speak-
ers in their midst (apart from the “formerly fluent” group, who seemed to be
exempted as the lapsed fluent speakers they were) as people who made mis-
takes in their Gaelic, and I coined the metaphorical term “semi-speaker” to
designate them (ch. 5). Whether contemporary semi-speakers had once been
fully fluent child speakers of Gaelic and had lost proficiency (one Brora case
was clearly of this sort – see Dorian 1981: 81 – and probably two others as well),
or whether their acquisition had been partial and they had never achieved
full fluency (a history likely for at least one Brora semi-speaker and three of
the Embo semi-speakers), was difficult to determine. But since the one Brora
woman who had indisputably been a fluent child speaker at school entry age
was a particularly low-proficiency semi-speaker as a woman in her late fifties, it
seemed that in Sutherland a history of full-fluency acquisition in childhood did
not bear directly on proficiency later in life. This outcome differs from Sasse’s
very clear-cut findings in the Arvanitika-speaking village of Boeotia in Greece
(Sasse 1992: 62). There he found that all speakers identifiable as semi-speakers
had had parents who had not talked to them in Arvanitika; instead they had
acquired their Arvanitika by listening to fluent older speakers and having occa-
sional interchanges with them. Fluent Arvanitika speakers, in sharp contrast,
were raised by adults who spoke to them in that language and had shown a
positive attitude toward the language. In another departure from the expec-
tations the Boeotian case might suggest, I found in work with multi-sibling
bilingual families both in East Sutherland and in Berks County, Pennsylvania,
where Pennsylvania German was still spoken by some members of local secular
(non-Anabaptist) families, that parental policy on using the heritage language
at home had less effect on acquisition than birth order did (ch. 6). The effect
of various acquisition histories remains something that invites more study.
The importance of imperfect speakers to following the patterns of mainte-
nance, shift, and linguistic change in East Sutherland fisherfolk Gaelic would
be hard to overstate. Their Gaelic, when set directly beside that of fully flu-
ent speakers, highlighted (morpho)phonological, lexical, and grammatical
features of East Sutherland Gaelic that were subject to change as age and
proficiency dropped. Their continuing use of Gaelic, in spite of grammatical
and lexical deficiencies evident to their elders (and criticized by those elders),
highlighted factors of personal history and local social structure that supported
language loyalty (ch. 3, ch. 6, ch. 11; Dorian 1981: 107–10). The fact that most of
them were women reflected the generally better integration into home-village
kin networks of younger women as compared with younger men. The striking
integrity of their receptive skills contrasted – vividly in a number of cases –
with the shortcomings of their productive skills, and their excellent knowledge
of sociolinguistic norms contrasted at times all too obviously with the imper-
fect knowledge (and so the greater likelihood of committing social offenses)
of the guest linguist who had learned to speak their language. Their general
acceptance as members of the bilingual community, despite considerable fail-
ings in their Gaelic, indicated that the conservative linguistic completeness
and invariance especially valued by descriptive linguists was not valued to a
comparable degree among community members (ch. 7, ch. 9).
Determining the differences between the Gaelic spoken by the oldest and
most fully fluent of fisherfolk Gaelic speakers and the youngest and least fully
fluent was key to learning what changes were taking place in this speech form
as transmission became less routine and conditions for acquisition less ideal.
At the end of §2 above, a rich array of questions was raised to which it seemed
possible that developments in East Sutherland fisherfolk Gaelic might offer
some answers. Not one of those questions was likely to be asked by anyone
but a linguist, however, and certainly none of them by members of the local
speaker population. For the most part researchers working with small, reced-
ing languages have had a set of objectives that privilege Western scientific
modes and topics of inquiry and their objectives and are often – indeed usu-
ally – not shared by the speaker community (Hill 2002; Dorian 2002).
In the 1970s it was common to speak of “dying languages” and “language
death” in connection with languages that were not being transmitted to chil-
dren. The biological metaphor was controversial from the outset, however,
given that language is primarily a social rather than a biological phenomenon;
and in addition, many language communities found it offensive to have the
label “dying language” attached to their ancestral speech form by outsiders
who had no direct connection to the ethnic community and often no personal
experience of the language at all (see especially Hill 2002). Both objections
were valid enough, and there has been a decline in the use of “language death”
terminology in the rhetoric of outsiders with professional linguistic expertise
but without personal experience of the languages they are referring to, much
less ethnic connection to them. But anthropologist Bernard Perley, a mem-
ber of a Canadian First Nation, considers that all of the metaphorical label-
ing characteristic of expert rhetoric has been detrimental to small indigenous
communities’ sense of vitality and restorative possibility, where language is
concerned (Perley 2012: 135, 141). In addition, he asserts, simplistic notions of
“saving” languages by documenting them have been counterproductive, pro-
ducing more attention to the artifacts of documentation than to speakers
(op. cit. 134).
One problem inherent in working with receding languages is that such lan-
guages commonly show changes directly related to reduced use. Comparison
between forms used by earlier populations, if such records are available, or
by older speakers as opposed to younger speakers, frequently demonstrate
shrinkage of one or many sorts: reduction of phonological inventory, loss of
grammatical distinctions, decline in lexical options. These are not generally
perceived as positive developments, either by ethnic community members or
by descriptive linguists, although not necessarily for the same reasons. Perhaps
inevitably, a process of pejoration has appeared where the description of these
change processes is concerned. This has happened for example in connec-
tion with the term “semi-speaker”, to which negative overtones have become
attached. I was at pains when I introduced the term to celebrate the remark-
able way in which semi-speakers put limited skills to maximum use (ch. 3) and
when they speak fluently, or to act as sources for such a language (Evans 2001:
250–261). For at least one part of Australia Eades points to special difficulties
in establishing when an Aboriginal language ceases to be spoken: “. . . for many
Aborigines the use of a language-group label to identify a person remains long
after that language is no longer spoken. In fact it is often difficult to say when a
language is no longer spoken, since some Aboriginal groups continue to mix in
some words from Aboriginal languages when they are speaking what is essen-
tially Standard English” (Eades 1981: 13). In southeastern Australia, Aboriginal
languages believed extinct were found in common use “in at least rudimentary
form” as secret languages designed to confound the police (Wurm 1991: 15).
Reporting on South America, where languages believed to be extinct are
rediscovered from time to time (he instances Cholón in the Peruvian Andes),
Adelaar recommends that linguists refrain from statements that a native lan-
guage is no longer spoken: “Categoric statements by linguists to the effect that
there are no speakers left of a particular language often do more harm than
good as they may keep other linguists from continuing the search” (Adelaar
1991: 51). In one recognized type of language retreat, “bottom-to-top death”,
everyday use of the local language disappears, but some ritual use is still
preserved for ceremonial purposes. Campbell and Muntzel (1989: 185) tell of
a source who provided a religious text in Chiapanec (an Otomanguean lan-
guage of Mexico) that he reserved for performance but spoke no Chiapanec
otherwise except for a small number of isolated “remembered” vocabulary
items. Fluent learners from outside the ethnic group would not normally be
included in tallies of remaining speakers, perhaps, but how is a fluent outsider
who becomes a transmitter of the language to be reckoned? This is the sur-
prising role that linguist William Shipley took on late in life, transmitting the
Mountain Maidu language of northern California to a grandson of the Maidu
speaker who had contributed most to making Shipley himself a fluent speaker
(Shipley 2000). Even languages that have no active speakers remaining may
be considered potentially viable, especially by members of the ethnic group
in question, if documentation is full enough to allow them to mount a well-
substantiated reclamation effort, as has happened in the case of the “sleeping”
language Miami (Leonard 2008).
Indeterminacies about what constitutes a speaker and what level and fre-
quency of use justifies the claim that a language is still spoken are encountered
in discussions of language revitalization, too. Irish has long been a key case in
this regard. If a substantial number of Irish citizens have a second-language
knowledge of Irish (enough to take an Irish-language newspaper or listen to
broadcasts in Irish, say) but do not actively use Irish on any regular basis, is their
existence a sign of the success of Irish “revival” efforts or of their failure (ch. 13)?
were altogether different from his: “Simply put, they were interested in words
not grammar” (Collins 1998: 260). Kroskrity, similarly, found that the linguis-
tic focus of Arizona Tewa speakers’ very long-standing puristic ideology was
strictly lexicographical, with no comparable attention paid to instances of
grammatical convergence resulting from the very same history of language
contact (Kroskrity 1998: 110).
Yet another tacit ideological point of view, one that generally went unac-
knowledged among early researchers working with endangered languages, was
identified by Susan Gal: writings about the receding languages of small and
mostly rural communities tended to be perfused with a pastoralist perspec-
tive that looked back to a more intact past, linguistic as well as social; this was
reflected in the use of the metaphor “language death” itself and in an emphasis
on structural loss and current-speaker inadequacy (Gal 1989: 316). The focus
was on the disappearance of conservative structures used by the oldest speak-
ers, while innovative aspects of younger speakers’ speech, such as those she
identified in Oberwart Hungarian, were less likely to be acknowledged (Gal
1989: 315–16).
Particularly pervasive in the Western world has been an assumption that
linguistic homogeneity is somehow fundamental, with monolingualism the
normal condition and the ideal nation a bounded territory in which a single
language is spoken (Irvine and Gal 2000: 63, 76). Most of our accounts of bi-
and multilingualism appear in the writings of Western scholars who present
these states as unusual or problematic conditions, settings for imbalance in
language dominance and strength (e.g. “Bilingualism . . . is a natural setting
for the unraveling of native language abilities”, Seliger and Vago 1991: 3) and
sustainable only so long as practical necessity requires them (“People will not
indefinitely maintain two languages when one will serve across all domains”,
Edwards 1994: 110). Influenced by such widespread but mostly unacknowledged
assumptions, Westerners, including linguists, can find it deeply surprising to
encounter well-established multilingual populations who unproblematically
maintain fluency in several languages, even when one of those languages is
a regional lingua franca that potentially serves all practical needs (ch. 16).
Remarkable, too, to researchers of Western European cultural background, are
accounts of multilinguals who in old age take pleasure in adding to the reper-
tory of languages they already speak (ch. 16; see also the Aboriginal Australian
role of the “polylingual specialist” and “linguistic virtuoso”, described in the
sources cited in Brandl and Walsh 1982). Intolerance of bilingualism, partic-
ularly pronounced in the U.S., can leave us surprised, too, to learn of larger
populations who matter-of-factly accept the long-term presence of a minority
population speaking an allophone language in their midst (ch. 16). Relatively
Despite some urgent statements in the early 1990s (Dixon 1991, Hale et al. 1992),
the need for a massive effort to document the many speech forms likely to pass
out of active use in the foreseeable future was slow to gain any real traction
among linguists. In the 1970s and ’80s, it was hard for linguists engaged with
endangered languages to get their colleagues, let alone the media and the read-
ing public, to pay attention to the oppressive fates and dim prospects of many
small language communities scattered around the globe. This is scarcely the
case now, however, and in an ironic turn of events the ubiquity and the dra-
matic character of warnings about the “endangerment crisis” are now them-
selves the subject of critical commentary (e.g. Moore 2006, Muehlmann 2011).
Since the late 1990s (e.g. Himmelmann 1998), discussions about documen-
tation have come to center on the advantages of (cross-disciplinary) team
We can predict more securely that future linguists will value the efforts
of today’s documenters than that indigenous groups will, but even there we
should not be too confident of our success. For one thing, linguists are still
drawn, almost irresistibly it seems, to the most unique features of any language
they study and are inclined to overstudy these and neglect less glamorous fea-
tures (ch. 19; Berge 2010: 57–58). For another, research priorities change, and
so do descriptive and analytic directions. Our future successors are sure to
fault us in our turn for gaps and skewings in our work. Suggestively, in this
regard, Woolard reports that participants in the discussion group prefiguring
an important 1998 volume on language ideologies “were struck by the appar-
ent absurdity of nineteenth-century philology’s relentless reading of spiritual
qualities from linguistic structures”, but then wondered whether “the single-
minded reading of power into and out of communicative practices that has
characterized our own late-twentieth-century sociolinguistics will look as ludi-
crously obsessive in another century’s retrospect” (Woolard 1998: 28).
Already uncommon, multi-decade engagements with a single minority
language are likely to become rarer still as concern grows over the number of
languages still to be documented. An engagement of that extent offers time to
review and expand or correct earlier observations and analyses, a luxury many
endangered language researchers never have and even fewer will have in the
future. In the case of my work with East Sutherland fisherfolk Gaelic, two expe-
riences in particular have led me to greater appreciation of the advantages of
multi-faceted and long-term work with the speech forms of a single area.
As a result of my involuntary switch from on-site fieldwork to telephone
fieldwork in the 1990s, the data that I recorded changed character considerably.
Field linguists have traditionally been more inclined to record certain kinds
of material than others (Foley 2003; Berge 2011: 56), and I was no exception:
my early recordings ran heavily to stories, autobiographical narratives, inter-
views, and some folk tradition. Although I asked questions fairly frequently,
these interchanges remained primarily informational. By contrast, the mate-
rial recorded later over the phone was first and foremost conversational. My
sources generously continued to do a good deal of translation-task work over
the phone with me, focused at that stage on personal pattern variation (ch. 21),
but most of the time we talked back and forth and the Gaelic on these tape
recordings was very different in style from what I had recorded in the 1960s and
’70s. (For example, they were much less revealing of puristic English loanword
avoidance and much more revealing of increasing use of English discourse
particles, adverbs, and conjunctions.) Because what gets recorded is affected
by so very many factors – e.g., how well source(s) and researchers know one
another or like one another, how many people are present and listening, how
much factionalism or leadership competition exist within the population the
researcher would like to record, what the recording medium is, how forthcom-
ing the sources are (unknowns such as the behavior or even aptitude of previ-
ous researchers can be major influences; see Grinevald 2005), the race, gender,
and/or age of the researchers and the sources, even the season of the year
and the weather in some settings – it seems inevitable that any documentary
record we produce will be skewed in ways and directions that we recognize
poorly or fail to recognize at all. This does not of course mean that we should
abandon the effort to document, only that we should be very cautious about
supposing that a language for which we have copious audio and video record-
ings can be considered well or fully documented.
The second notable experience concerned the documentation of that very
personal pattern variation that by the 1990s I was especially interested in.
Variation of various kinds was overwhelmingly abundant in the local fisherfolk
Gaelic, and only once a very substantial corpus was available from a consider-
able number of speakers could anyone hope to determine what part of the
variability represented geographical variation and what part personally pat-
terned variation (or for that matter stylistic variation). In one case, for example,
only the chance fact that a lexical item was both a geographical variant and a
personally patterned variant made it ciear to me (belatedly at that) that two
different kinds of variation were simultaneously involved. I had tracked the
word airgiod ‘money, silver’ as a geographic variable (its Embo form differed
from its Brora and Golspie form) long before I had acquired and sorted the
larger number of instances from Brora and Golspie speakers that were needed
to establish that it was also a personal pattern variant in those two villages,
with different sets of speakers using a slightly different first-syllable vowel. It
took years to untangle such complexities of variation, and since personally
patterned variation turned out to be intimately linked to the social organiza-
tion of these villages, despite the fact that the variables as such were not social
group markers, no documentation effort short of a multiple-year engagement
with the language would have made that untangling possible.
The second of these fieldwork experiences is to me more disconcerting of
the two. In unwritten languages without external norming, individual variation
is a much more frequent phenomenon than has previously been recognized
(see the literature cited in Dorian 2010, chapter 8; also Goddard 2010, Chatsis,
Miyashita, and Cole 2013, and Aikhenvald forthcoming). In many endangered
language settings it would be not just easy, but even natural, to take the profu-
sion of variant forms for some sort of late-stage laxity or for the dissolution of
linguistic competence rather than the distinct linguistic phenomenon it is, and
so to expunge it from a record that was unconsciously aimed at the invariance
associated with standardized languages. This mistake would leave the linguis-
tic record incomplete in much the same way that eliminating young speakers
and imperfect speakers as sources once did.
Long-term fieldwork does not of course guarantee that documentation will
be ideally complete, let alone superior in some particular fashion. It will how-
ever be different, because there will be opportunity to review early-acquired
data from a more broadly based late-fieldwork perspective and because the
possibility of new late-stage information increases. In my own case personal
circumstances set me on this path, rather than original intention, but in retro-
spect I take it to have been a useful path to follow.
References
Adelaar, Willem F.H. 1991. The endangered languages problem: South America. In R.H.
Robins and E.M. Uhlenbeck, eds, Endangered languages, 45–91. Oxford: Berg.
Aikhenvald, Alexandra Y. 2001. Language awareness and correct speech among the
Tariana of Northwest Amazonia. Anthropological Linguistics 43(4):411–30.
———. 2003. Teaching Tariana, an endangered language from Northwest Amazonia.
International Journal of the Sociology of Language 161:125–39.
———. 2013. Shifting attitudes in north-west Amazonia. International Journal of the
Sociology of Language 222:195–216.
———. Forthcoming. Language contact and language blend: Kumandene Tariana of
north-west Amazonia.
Austin, Peter K., and Stuart McGill, eds. 2011. Endangered languages. New York:
Routledge.
Berge, Anna. 2010. Adequacy in documentation. In Lenore A. Grenoble and N. Louanna
Furbee, eds, Language documentation: Practice and values, 51–66. Amsterdam: John
Benjamins.
Bloomfield, Leonard. 1964 [1927]. Literate and illiterate speech. In Dell Hymes, ed.,
Language in culture and society: A reader in linguistics and anthropology, 391–96.
Cambridge: Cambridge University Press.
Bradley, David. 2007. What elicitation misses: Dominant language, dominant seman-
tics. In Peter K. Austin, ed., Language Documentation and Description 4:136–44.
Brandl, M.M., and M. Walsh. 1982. Speakers of many tongues: Toward understanding
multilingualism among Aboriginal Australians. International Journal of the Sociology
of Language 36:71–81.
Campbell, Lyle, and Martha C. Muntzel. 1989. The structural consequences of language
death. In Nancy C. Dorian, ed., Investigating obsolescence: Studies in language con-
traction and death, 181–96. Cambridge: Cambridge University Press.
Chatsis, Annabelle, Mizuki Miyashita, and Debbie Cole. 2013. A documentary ethnog-
raphy of a Blackfoot language course: Patterns of variationism and standard in the
‘organization of diversity’. In Shannon Bischoff, Deborah Cole, Amy Fountain, and
Mizuki Miyashita, eds, The persistence of language: Constructing and confronting the
past and present in the voices of Jane H. Hill. Amsterdam: John Benjamins.
Childs, G. Tucker. 2009. How to pretend you speak a dying language when you don’t
really know how to speak it: Methodological worries in documenting dying lan-
guages. In M. Brenzinger and A.M. Fehn, eds, Proceedings of the 6th World Congress
of African Linguistics, 45–56. Cologne: Köppe.
Collins, James. 1998. Our ideologies and theirs. In Bambi B. Schieffelin, Kathryn A.
Woolard, and Paul V. Kroskrity, eds, Language ideologies: Practice and theory, 256–
70. New York: Oxford University Press.
Debenport, Erin. 2002. Comparative accounts of linguistic fieldwork as ethical exer-
cises. International Journal of the Sociology of Language 206: 227–44.
Dixon, R.M.W. 1991. The endangered languages of Australia, Indonesia and Oceania. In
Robins, R.H., and E.M. Uhlenbeck, eds, Endangered languages, 229–55. Oxford:
Berg.
Dorian, Nancy C. 1978. East Sutherland Gaelic: The dialect of the Brora, Golspie and
Embo fishing communities. Dublin: Institute for Advanced Studies.
———. 1981. Language death: The life cycle of a Scottish Gaelic dialect. Philadelphia:
University of Pennsylvania Press.
———. 1993. A response to Ladefoged’s other view of endangered languages. Language
69(3):575–79.
———. 1995. Sharing expertise and experience in support of small languages.
International Journal of the Sociology of Language 114:129–37.
———. 2002. Commentary: Broadening the rhetorical and descriptive horizons in
endangered-language linguistics. Journal of Linguistic Anthropology 12(2):134–40.
———. 2010. Investigating variation: The effects of social organization and social set-
ting. New York: Oxford University Press.
Dressler, Wolfgang U. 1972. On the phonology of language death. Proceedings of the
Chicago Linguistic Society 8:448–57.
Eades, Diana. 1981. ‘That’s our way of talking’: Aborigines in southeast Queensland.
Social Alternatives 2(2):11–14.
Edwards, John. 1994. Multilingualism. London and New York: Routledge.
Evans, Nicholas. 2001. The last speaker is dead. Long live the last speaker! In Paul
Newman and Martha Ratliff, eds, Linguistic fieldwork, 250–81. Cambridge: Cambridge
University Press.
Flores Farfán, José Antonio. 2001. Culture and language revitalization, maintenance,
and development in Mexico: The Nahua Alto Balsas communities. International
Journal of the Sociology of Language 152:185–97.
Foley, William A. 2003. Genre, register and language documentation in literate and pre-
literate communities. In Peter K. Austin, ed., Language Documentation and
Description 1:85–98.
Gal, Susan. 1989. Lexical innovation and loss: The use and value of restricted Hungarian.
In Nancy C. Dorian, ed., Investigating obsolescence: Studies in language contraction
and death, 313–31. Cambridge: Cambridge University Press.
Goddard, Ives. 2010. Linguistic variation in a small community: The personal dialects
of Moraviantown Delaware. Anthropological Linguistics 52(1):1–48.
Grinevald, Colette. 2005. Why the tiger language and not Rama Cay Creole? Language
revitalization made harder. In Peter K. Austin, ed., Language Documentation and
Description 3:196–224.
Haas, Mary R. 1941. Tunica (Extract from Handbook of American Indian Languages,
Vol. IV). New York: J.J. Augustin.
Hale, Ken, Michael Krauss, Lucille J. Watahomigie, Akira Y. Yamamoto, Colette Craig,
LaVerne Masayesva Jeanne, and Nora C. England. 1992. Endangered languages.
Language 68(1):1–42.
Hill, Jane H. 1973. Subordinate clause density and language function. In C. Corum, T.C.
Smith-Stark, and A. Weiser, eds, You take the high node and I’ll take the low node:
Papers from the comparative syntax festival, 33–52. Chicago: Chicago Linguistic
Society.
———. 2001. Dimensions of attrition in language death. In Luisa Maffi, ed., On biocul-
tural diversity, 175–89. Washington D.C.: Smithsonian Institution Press.
———. 2002. “Expert rhetorics” in advocacy for endangered languages: Who is listen-
ing, and what do they hear? Journal of Linguistic Anthropology 12(2): 119–33.
Himmelmann, Nikolaus P. 1998. Documentary and descriptive linguistics. Linguistics
36:161–95.
———. 2008. Reproduction and preservation of linguistic knowledge: Linguistics’
response to language endangerment. Annual Review of Anthropology 37:337–50.
Irvine, Judith T., and Susan Gal. 2000. Language ideology and linguistic differentiation.
In Paul V. Kroskrity, ed., Regimes of language: Ideologies, polities, and identities,
35–83. Santa Fe: School of American Research Press.
Kroskrity, Paul V. 1998. Arizona Tewa speech as a manifestation of a dominant language
ideology. In Bambi B. Schieffelin, Kathryn A. Woolard, and Paul V. Kroskrity, eds,
Language ideologies: Practice and theory, 103–22. New York: Oxford University Press.
———. 2002. Language renewal and the technologies of literacy and postliteracy:
Reflections from Western Mono. In William Frawley, Kenneth C. Hill, and Pamela
Munro, eds, Making dictionaries: Preserving indigenous languages of the Americas,
171–92. Berkeley: University of California Press.
Kuter, Lois. 1989. Breton vs. French: Language and the opposition of political, economic,
social, and cultural values. In Nancy C. Dorian, ed., Investigating obsolescence:
Studies in language contraction and death, 75–89. Cambridge: Cambridge University
Press.
Ladefoged, Peter. 1992. Another view of endangered languages. Language 68(4):
809–11.
Leonard, Wesley Y. 2008. When is an “extinct language” not extinct? Miami, a formerly
sleeping language. In Kendall A. King, Natalie Schilling-Estes, Lyn Fogle, Jia Jackie
Lou, and Barbara Soukup, eds, Sustaining linguistic diversity: Endangered and minor-
ity languages and language varieties, 23–33. Georgetown: Georgetown University
Press.
Long, Michael H. 1993. Second language acquisition as a function of age: Research find-
ings and methodological issues. In Kenneth Hyltenstam and Åke Viberg, eds,
Progression and regression in language: Sociocultural, neuropsychological and lin-
guistic perspectives, 196–221. Cambridge: Cambridge University Press.
McEwan-Fujita, Emily. 2010. Ideologies and experiences of literacy in interaction
between adult Gaelic learners and first-language Gaelic speakers in Scotland.
Scottish Gaelic Studies 26:87–114.
McLendon, Sally. 1978. How languages die: A social history of unstable bilingualism
among the Eastern Pomo. In Margaret Langdon, Shirley Silver, and Kathryn Klar,
eds, American Indian and Indo-European Studies, 137–50. The Hague: Mouton.
Miller, Elmer S. 1995. Nurturing doubt: From Mennonite missionary to anthropologist in
the Argentine Chaco. Urbana and Chicago: University of Illinois Press.
Milroy, James. 1992. Linguistic variation and change: On the historical sociolinguistics of
English. Oxford: Blackwell.
———. 1999. The consequences of standardization in descriptive linguistics. In Tony
Bex and Richard J. Watts, eds, Standard English: The widening debate, 16–35. London:
Routledge.
Milroy, Jim. 2000. Historical description and the ideology of the standard language. In
Laura Wright, ed., The development of Standard English, 1300–1800, 11–28. Cambridge:
Cambridge University Press.
Moore, Robert E. 2006. Disappearing, Inc.: Glimpsing the sublime in the politics of
access to endangered languages. Language and Communication 26:296–315.
Muehlmann, Shaylih. 2011. Von Humboldt’s parrot and the countdown of last speakers.
Language and Communication 32:160–68.
Newman, Paul, and Martha Ratliff, eds. 2001. Linguistic fieldwork. Cambridge:
Cambridge University Press.
Ó Dochartaigh, Cathair, ed. 1997. Survey of the Gaelic dialects of Scotland: Questionnaire
materials collected for the Linguistic Survey of Scotland, Vol. 1. Dublin: Dublin
Institute for Advanced Studies, School of Celtic Studies.
Peacock, John Hunt. 2010. The good news and the bad news about reviving my Spirit
Lake Dakota nation’s endangered language. In Hywel Glyn Lewis and Nicholas
Ostler, eds, Reversing language shift: How to re-awaken a language tradition, 132–39.
(Proceedings of Conference FEL XIV) Bath: Foundation for Endangered Languages.
Perley, Bernard C. 2012. Zombie linguistics: Experts, endangered languages, and the
curse of the undead voices. Anthropological Forum 2(2):133–49.
Polinsky, Maria. 2011. Reanalysis in adult heritage language: New evidence in support of
attrition. Studies in Second Language Acquisition 33:305–28.
Sasse, Hans-Jürgen. 1992. Language decay and contact-induced change. In Matthias
Brenzinger, ed., Language death: Factual and theoretical explorations with special
reference to East Africa, 59–80. Berlin: Mouton de Gruyter.
Schmid, Monika S. 2011. Language attrition. Cambridge: Cambridge University Press.
Schmidt, Annette. 1985. Young people’s Dyirbal: An example of language death from
Australia. Cambridge: Cambridge University Press.
Seliger, Herbert W., and Robert A. Vago. 1991. The study of first language attrition: An
overview. In Herbert W. Seliger and Robert A. Vago, eds, First language attrition, 1–15.
Cambridge: Cambridge University Press.
Shipley, Bill. 2000. The Maidu story. Sandy Chung, Jim McCloskey, and Nathan Sanders,
eds, The Jorge Hankamer Web Fest. Babel.ucsc.edu/jorge/, accessed 16 Mar. 2013.
Stebbins, Tonya. 2003. Fighting language endangerment: Community directed research
on Sm’algyax (Coast Tsimshian). Osaka: Endangered Languages of the Pacific Rim.
Thomason, Sarah G. 2001. Language contact: An introduction. Washington D.C.:
Georgetown University Press.
Tsitsipis, Lukas. 1981. Language change and language death in Albanian speech com-
munities in Greece: A sociolinguistic study (Unpublished Ph.D. dissertation,
University of Wisconsin, Madison).
———. 1983. Language shift among the Albanian speakers of Greece. Anthropological
Linguistics 25(3):288–308.
———. 1984. Functional restriction and grammatical reduction in Albanian language
in Greece. Zeitschrift für Balkanologie XX(1): 122–31.
Will, Vanessa Katharina Angela. 2012. Why Kenny Can’t Can: The language socialization
experiences of Gaelic-medium educated children in Scotland. Unpublished University
of Michigan dissertation.
Woodbury, Tony. 2003. Defining documentary linguistics. Language Documentation
and Description 1 :35–51.
Woolard, Kathryn A. 1998. Introduction: Language ideology as a field of inquiry. In
Bambi B. Schieffelin, Kathryn A. Woolard, and Paul V. Kroskrity, eds, Language ide-
ologies: Practice and Theory, 3–47. New York: Oxford University Press.
Wurm, Steven A. 1991. Language death and disappearance: Causes and circumstances.
In R. H. Robins and E. M. Uhlenbeck, eds, Endangered languages, 1–18. Oxford: Berg.
1 Introduction
1 This chapter is based on research begun in East Sutherland during a year’s residence in
1963–64, and continued at frequent intervals up to the present. A grant from the American
Council of Learned Societies made possible a full summer’s work on the project in 1970, and
the School of Scottish Studies was generous with space and facilities during that same period
and others. I would like to thank Howard Hoffman for help with the graphs which appear in
this paper.
there was nothing to attract new population to the village. As a result, Gaelic-
English bilinguals continued to appear in Embo for about 25 years after active
bilingualism ceased to be the norm for fisher-descendants in the two larger vil-
lages. The youngest fluent bilinguals in Embo are currently in their early 40’s.
These are people who spoke Gaelic regularly at home and on the school play-
ground in their childhood years, and who now explicitly claim Gaelic as their
mother tongue, even when they acknowledge that they are more competent in
English. In Brora and Golspie, a group with so consistently Gaelic a background
can be found only among the 70- and 80-year-olds. The current bilinguals of
Brora and Golspie had parents who were very nearly monolingual in Gaelic, for
the most part. So did the older bilinguals in Embo. But the younger Embo bilin-
guals grew up in a community which lacked a monolingual Gaelic norm. These
younger Embo bilinguals express doubt about the “correctness” of their Gaelic,
and often remark that their Gaelic is inferior to that of their parents and grand-
parents. Explicit comment on the decline in the quality of their Gaelic focuses
almost entirely on the lexicon, however: the younger speakers feel sure their
elders had many more ‘words for things’ than they have themselves. There is
a much lower awareness of only one instance of ongoing analogical leveling
in the morphology, and some sporadic note is taken of certain phonological
developments; but there is no awareness at all in the community of develop-
ments currently underway in the grammar of the so-called ‘initial mutations’.
clearly in a terminal phase in the area, it was easy to suppose that this motley
picture reflected a sporadically disrupted stage in the decay of the local form
of the Gaelic language. It was only the gradual development of an ever wider
age-range of informants that brought the situation into clearer focus. Some of
my original and oldest informants died, and the speakers who replaced them
in my sample were inevitably a little younger. In Embo, where a younger group
of speakers is statistically preponderant, I originally aimed to work only with
older speakers, i.e. people of 60 or more. But eventually I developed a younger
group of informants as well, through sheer statistical availability and the
friendly helpfulness of younger friends and acquaintances, speakers between
40 and 60. As the sample of speakers broadened, and younger speakers either
replaced or supplemented older speakers, I gradually became aware that the
apparent “mistakes” of the vanishing oldest speakers were somewhat com-
moner among the slightly younger speakers, and extremely common among
the youngest Embo speakers.
In ESG the mutational system may be said to be the same for all age groups in
the following senses: every speaker has retained an active use of initial con-
sonant mutation as a syntactic device; and every speaker has a repertory of
2 Since the herring fishing began to die out more than 50 years ago, and the last of the Gaelic-
English bilinguals of the agricultural population died out during the same period, the East
Sutherland fisherfolk and their descendants have been increasingly isolated linguistically.
But even during the heyday of the herring industry, when both men and women travelled
extensively around the coasts following the fishing, the aberrance of ESG made interaction
with other Gaels difficult. This same aberrance makes contact with the standard language –
in writing, on radio, or in church services – of relatively little reinforcement value to East
Sutherlanders. No surviving East Sutherland Gaelic speaker is actively literate to the point
where he or she could write a letter wholly in Gaelic, and this again reflects the distance
between the standard language and the unwritten dialect. [Correction, 2014: G3 subse-
quently proved able to do so.] Passive literacy exists in a small number of cases, but it is
almost exclusively confined to a limited ability to read the Bible, especially the metrical ver-
sion of the Psalms.
initial consonantal choices which includes (to use the traditional terminology)
those of “lenition”,3 those of “nasalization”,4 and those of the unmutated radi-
cal initial. Given this basic system in common, and given a group of older, more
conservative speakers who represent a mutational norm, the younger speak-
ers’ departures from that norm might consist in (1) substituting the opposite
mutation for the one their elders use; (2) using a mutation where their elders
use none; (3) using either mutation interchangeably where their elders use
only the one or the other; and (4) using no mutation at all where their elders
obligatorily use one or the other. All these deviations do in fact occur, but not
all of them in any one syntactic environment. Each group or sub-group of
mutational environments shows deviant developments peculiar to itself, and
consequently each must be examined separately with regard to the pattern
of change involved. This paper will concentrate on two striking instances of
change in progress, and contrast these with environments where little or no
change appears.
3 In most cases lenition requires the replacement of an initial obstruent by a spirant. But in
the case of initials which are themselves already spirants, and also the initial consonants of
certain clusters, the replacement may be another spirant or zero. The lenition replacements
in ESG are as follows (V = vowel, C = consonant, I = front vowel, and U = back vowel): Lph =
[f]; LthV, LsV, LchI, LsI = [h]; LthC = zero; LčhU, LšU, L khy, LkI = [ç]; LkhU = [x] (Embo and Brora
also have Lt, LkU = [x], and Golspie has Lt, LkU = [γ]); Lf = zero; Lmy, LmI, Lpl, Lpy, Lpr, LpV =
[v]; Lpw = zero; Lč, LkI = [y]; Lky = zero; Lsn, Lsn´ = [r ~ n], Lstr = [r].
4 In most cases, nasalization in ESG requires the replacement of an initial voiceless consonant
by a voiced consonant – usually, although not always, the voiced equivalent of the original
radical initial. The nasalization replacements in ESG are: Nph, Np = [b]; Nth, Nt, NsV = [d]; Nčh,
Nč, Nš, Ny = [dž]; Nkh, Nk = [g]; Nstr, Nsn, Nsn´ = [dr].
5 A considerable isogloss bundle separates Embo village from Brora and Golspie, with phono
logical isoglosses predominant, but lexical, morphological, and mutational isoglosses repre
sented as well. A very much smaller number of isoglosses runs between Brora and Golspie
villages, and there are even rare isoglosses which link Brora and Embo as opposed to Golspie,
or Golspie and Embo as opposed to Brora.
6 I.e., younger speakers may say /ən ǰo:ɫ ən ǰo:ɫədar/ [ən džo:ɫ ən džo:ɫədar] ‘Will the sailor
sail?’, while older speakers are much more likely to say /ən šo:ɫ ən ǰo:ɫədar/ [ən šo:ɫ ən
džo:ɫədar].
7 This represents the development which Kiparsky (1968: 200) predicts in the movement of
rules, so that the rule for the nasalization of the groove sibilant now affects more items than
it did previously.
8 Imperfect bilinguals who misuse the verbal mutations exist; they include Brora and Golspie
‘semi-speakers’ under 60 and Embo ‘semi-speakers’ under 40. They are excluded from discus
sion in this chapter by definition, since they are not fluent speakers. Their mutational usages
reflect not genuine changes in the mutational system, but imperfect control of Gaelic in
general, as revealed by many other irregularities in their speech.
sity of handling the full verbal apparatus. Most of the highest frequency verbs
are extremely irregular and do have mutatable initials. If we can assume early
mastery of this system, and inordinately heavy reinforcement of its mutational
phenomena because of the full mutational systems of the highest frequency
verbs, then we may after all have at least a clue to the conservatism of the ver-
bal grammatical system, including mutations.
Although the term ‘passive’ suggests a verbal structure to the English speaker,
the passive in ESG is essentially a nominal construction,10 centered on a pos-
sessive pronoun modifying a gerund (traditionally called the verbal noun in
Gaelic grammar). It is the possessive pronouns in this construction which
cause the mutation, and it is within the mutational system of the possessives
that change is taking place. The choice of the passive as the construction in
which to study the mutational changes involving the possessives is dictated
by the fact that the passive is the only construction in which the possessive
pronouns survive as a productive system in ESG.
Even the full paradigm of possessive pronouns is difficult to elicit in ES,
although the paradigm, once obtained, proves to parallel that of the standard
grammar very closely. The possessive pronouns which cause mutation in con-
servative ESG are precisely the ones which produce mutation in the standard
language, and the mutations produced are also the same:
sg pl.
10 The truly verbal passive, in which the category is expressed by the suffixal morphology of
the verb itself, survives only as a fossil in the expression rugadh mi /rugu mi/ ‘I was born’
and possibly in one or two other less common expressions.
11 The phonological shape of this possessive (1st and 2nd plural) is peculiar to ESG and other
northern dialects, but the lack of consonantal mutation after it is characteristic of the
standard language as well.
Except for their appearance in the passive, all but two of these possessives are
little-used in ESG, with the plurals rarest of all. The full range of possessives,
with appropriate mutations or lack of mutation, can be elicited from older
speakers, and even occasionally from some younger speakers, in only one envi-
ronment other than the passive: fossilized fixed expressions involving kinship
terms (‘when our father was alive’), body parts (‘at her back’, i.e. ‘behind her’),
or other inalienables (‘they lost their lives’). Outside these fixed expressions,
the pronouns of the 1st and 2nd singular occur freely;12 but the other singulars
typically, and the plurals invariably, are replaced by an alternative possessive
structure involving the definite article and the preposition aig /ig/ ‘at’, which
conjugates for person. Thus with a clothing term, e.g., we would typically get a
syntactically suppletive paradigm:
12 I.e., they occur freely within the semantic range open to possessives in ESG, chiefly
inalienables and things that can be worn or carried on the person. If the noun modified
does not fall into this semantic range, the alternative aig-construction will be used.
13 /bel/ ‘is’ (interrogative) shows nasalization in Embo, lenition in Golspie and Brora. The
phonological form given for paitean ‘sweater’ is also the Embo version; Golspie and Brora
have a close [e] in the same word. Since most of the informants whose speech is cited in
this study are from Embo village, Embo forms will be used throughout unless otherwise
noted.
14 Both are to be found in use in western dialects and in the written standard, but grammars
of Gaelic, whether prescriptive or descriptive, often ignore them. Neither is mentioned
in, e.g., Maclaren’s Gaelic self-taught (n.d.), or in Oftedal’s relatively full treatment of a
Hebridean dialect (1956), although examples of both appear in the sample texts Oftedal
provides.
‘go’; this construction will be referred to as the dol-passive. The full construc-
tion is made up of a finite verb followed by a subject which consists of a verbal
noun modified by a possessive pronoun:
The possessive pronouns of the 3rd singular and plural usually appear
homophonously as /ə/ or, after a preceding word-final vowel, as zero; hence
only the mutations resulting from the possessive can specify the subject of this
construction for the 3rd person, if the subject is a pronoun rather than a noun:
{Lga əL khumal ə stε} /xa xumal ə stε/ ‘He was kept in’.
{Lga ə khumal ə stε} /xa khumal ə stε/ ‘She was kept in.’
{Lga əN khumal ə stε} /xa gumal ə stε/ ‘They were kept in.’
The second passive construction takes its finite verb from the irregular verb
bith ‘be’, and will be referred to hereafter as the bith-passive. The bith-passive
consists of finite verb plus subject pronoun, followed by a prepositional phrase
in which the possessive once again modifies a verbal noun:
In this construction, too, the possessive pronouns of the 3rd persons usually
appear homophonously as /ə/ or zero, but here the presence of a preceding
subject pronoun makes the information provided by the mutation redundant.
15 Standard Gaelic chaidh ‘went’ is not the immediate source of East Sutherland /γa/
(Golspie) ~ /xa/ (Brora and Embo). In order to account for the voiced Golspie version, the
underlying base form would have to be either {Lta} or {Lka}; I have arbitrarily chosen the
latter to represent the base here. Note that the symbol a, in phonetic as well as phonemic
and morphophonemic transcriptions, is used for typographical convenience to represent
a back unrounded vowel.
frequent, throughout all persons. Embo speakers in their 40’s and 50’s,16 for
example, produced the utterances
This occasional absence of mutation in the 3pl., though rare, has an obvious
model in the normal absence of mutation in the other persons of the plural
and in the 3sg. feminine.
{Lga aǰ ənN pa:u s ə ɫɔx} /xa aǰ ə ba:u s ə ɫɔx/ ‘They got drowned in the
loch.’
16 E17 and E11 respectively in a chronological numbering of informants which makes E1 the
oldest Embo informant, E2 the second-oldest, and so forth through E19.
But once the subject pronoun is present, the younger speakers are much more
likely to extend lenition to all persons, exactly as they do for the most part with
the bith-passives. Thus E16 produced a 3pl. with lenition in the sentence
{Lga aǰ əL khur a max} /xa aǰ ə xur ə max/ ‘They were put out’
80’s E1 9 – –
9 (100%) 0 0 Totals
70’s E2 6 – –
E3 3 2 –
E4 2 1 1
E5 4 3 –
15 (68%) 6 (27%) 1 (5%) Totals
60’s E6 – 1 2
E7 3 – 1
E8 7 – 1
E9 1 1 4
11 (52%) 2 (10%) 8 (38%) Totals
50’s E10 1 3 2
E11 – – 5
E12 7 – 5
E13 1 2 3
E14 2 3 7
11 (27%) 8 (19%) 22 (53.5%) Totals
40’s E16 – 4 12
E17 – 1 3
E18 – 3 9
E19 – – 5
0 8 (22%) 29 (78%) Totals
We have already seen that the analogical remodeling of the dol-passive after
the pattern of the bith-passive allows variability in initial consonants to appear
in the verbal noun of the dol-passive as in that of the bith-passive. Table 2
shows the number of actual instances of passives produced with appropri-
ate vs. inappropriate17 initial consonant for both passives in the interview
17 ‘Inappropriate’ here means either lenition or no mutation at all, since the environments
are the 3sg. feminine and the 3rd plural.
100
80
60
40
20
0
80 70 60 50 40
Figure 1.1. Percentage of conservative (non-analogical) dol-passives plotted according to the
age-group of the speakers.
18 The interview called for Gaelic translations of English originals. Past experience had
shown that the bith-passives were normally produced as translation equivalents of the
English be-passive, whereas I was especially interested in developments in the dol-passive;
hence the English models given were all get-passives. Among older speakers this proved
to produce a bias against the bith-passive, which I allowed to go uncorrected because
I already had ample information from early fieldwork on the conservative bith-passive.
Younger speakers tended to use one of the two finite verbs to the exclusion of the other,
more often bith than dol, and in such cases I instructed them to “begin that sentence with
/xa/” or “begin that sentence with /va/” in order to test for the structure of both passive
constructions. These procedures account for the over-all preponderance of dol-passives
in the data.
Table 1.2 Incidence of appropriate and inappropriate initials in the 3 sg.f. and 3 pl. of both
passives
80’s E1 – – 9 –
9 0 Totals
70’s E2 2 – 6 –
E3 – – 3 –
E4 5 1 3 1
E5 3 1 5 2
10 2 17 3 Totals
60’s E6 – – 1 2
E7 – – 4 –
E8 – – 8 –
E9 1 – 5 1
1 0 18 3 Totals
50’s E10 8 3 4 2
E11 3 2 2 3
E12 – – 10 2
E13 3 2 4 2
E14 – 1 6 6
14 8 26 15 Totals
40’s E16 1 1 9 7
E17 5 10 1 4
E18 – 1 1 11
E19 – 2 – 5
6 14 11 27 Totals
The results of Table 2 are presented in graph form in Figure 2, where the
percentage of forms with appropriate initial consonant (taking both passives
together) is plotted in terms of the speakers’ age-groups. Except for a slight
rise between the 70 to 60 decade-groups, the function exhibits a clear down-
ward movement from 100% for the octogenarian E1 to 29% for the youngest
decade-group.
120
100
80
60
40
20
0
80 70 60 50 40
Figure 1.2 Percentage of inappropriate initial consonants in both passives plotted according to
the age-group of the speakers
19 It should be noted that no variation occurs in dol-passives unless the subject pronoun is
intruded. Subjectless dol-passives are mutationally perfectly orthodox, even in ESG.
20 By considering only the noun accompanied by the article, we deliberately exclude the
vocative from consideration. Aside from the vocative, the noun without article shows no
initial mutations, and the final mutations of the case system do not survive for either the
definite or the indefinite noun. As for the vocative, lenition of the initial consonant is the
sole sign of its occurrence, in nearly all ESG nouns. (The preceding vocative particle a of
the standard language is as good as non-existent in the dialect, and the word-terminal
morphology of the vocative exists as a relic in at most five or six words, optionally at that.)
The only mutational “change” apparent in ESG vocatives is occasional unmutated initials
produced by the very youngest Embo speakers. These speakers are the same ones who
sometimes fail to mutate after {kle:L} or {ta:L}, and the failures are equally rare. These two
cases of mutational failure might seem to be functionally very dissimilar, since mutation
after {kle:L} and {ta:L} is totally redundant informationally, whereas mutation with the
vocative is normally the sole marker of the grammatical category. But in fact the voca
tive is so strongly marked by phonological signals (e.g. non-final contour with a sharp
preceding rise in pitch and a drop in pitch for the vocative noun itself) that loss of the
mutation produces no loss of information: phonological markers simply take over for
the morphophonological marker.
21 Feminine definite nouns in second position show no trace of the unlenited forms of the
standard grammar; instead, a lenited nominative always occurs in ESG.
22 It is possible to supply an ESG pronunciation for the missing genitives in these cases
because the genitive singular coincides with the plural, and ESG does preserve the pala
talized plurals.
23 Nominative and accusative are not distinguished. The term nominative will be used to
cover the general nominative/accusative form.
24 An unmutated feminine genitive singular in the definite noun does not survive in ESG,
even in fossilized phrases.
25 This rule is invariable, but the class “feminine” is itself variable. Gender-class membership
may vary from village to village, so that a noun which is masculine in one village (e.g. Golspie
/pu:/ ‘shop’) is feminine in the others (Brora and Embo /pu:/); but such geographically-
based differences are rare. A much commoner source of fluctuating gender-class mem
bership is the gradual transfer of part of the feminine noun-class into the masculine class.
Even older speakers show an occasional idiosyncratic masculine which is feminine for
their peers, but among younger speakers the transfer can occur on a fairly large scale.
26 Reasons 1 and 5 do not hold for s-initial masculines; 3 and 5 do not hold for m-initial
masculines; 4 and 5 do not hold for f-initial masculines. There is no other phonologically-
constituted and phonetically coherent sub-group of type 2.
My early attempt (1968) in the tracking down of the scale and limits of muta-
tional variability involved the use of a vast battery of nominative and dative
test-sentences for nouns of both genders and all mutatable initial consonants.
Subsequent work on mutational variability drew heavily on the results of this
earlier testing, and went on to test further the especially rich class of /kh/-
initial masculines, which includes more extremely common nouns than does
any other. Table 3 shows the results of the earliest testing for the nominative
case of definite nouns in 15 /kh/-initial masculines. The respondents were four
Golspie speakers, all around 70 years of age; one Brora speaker near 80; and
four Embo speakers, one over 60, one in his 50’s, and two in their 40’s. However,
there was never a full complement of nine responses because married couples
insisted on working together, and only the first response from such a pair was
taken as valid – i.e. unbiased – whichever spouse it came from. The table shows
a near-monolithic solidarity in favor of nasalization in the nominative after
the definite article. The sole instance of lenition after the article came from an
Embo speaker in his 40’s.
Largely because I was convinced that in daily speech I had met with more
variability than these tests showed, I decided to check the apparent solidarity
of this response from another angle. After an interval of from two weeks to
two years, the same sentences were read back to five of the original respon-
dents, under the pretense of checking to see whether they were recorded cor-
rectly; but in the read-back, the mutation other than the one originally given
was deliberately substituted. On this test the results became less unanimous.
In 27 cases with /kh/-initial masculines, older speakers insisted on their own
original nasalization; younger speakers insisted on their original nasalization
in only 13 cases. Older speakers accepted a contrary lenited version in just 6
cases, whereas younger speakers accepted a lenited version in 16 cases. That
is, older speakers rejected lenition of these masculines in the nominative more
than 80% of the time; younger speakers accepted lenition for the same mascu-
lines better than 50% of the time.
This evidence of some weakening of nasalization in the nominative among
younger speakers was confirmed by a later (1970) test which called for a Gaelic
translation of the English sentence ‘Who threw the stone that killed the stone-
mason?’ This sentence was simultaneously a test for the preservation of gender
and for the nominative mutation of the masculine noun. The Gaelic /khɫɔx/
‘stone’ is a reliable feminine for older speakers, while /khɫɔxar/ ‘stonemason’ is
a reliably masculine agent-noun derived from it. Three Golspie and two Brora
speakers, all over 70, lenited the feminine /khɫɔx/ according to the invariable
rule requiring lenition of definite feminines in the nominative (unless in initial
/th t čh č š/), and also nasalized the masculine /khɫɔxar/, again properly from
the point of view of conservative usage. The results for Embo speakers are pre-
sented in Table 4. Older Embo speakers did exactly as their Golspie and Brora
counterparts had done; but younger Embo speakers showed a weak tendency
to treat the feminine /khɫɔx/ as a masculine, and the youngest group of Embo
speakers showed a deviation toward a lenited masculine nominative after the
definite article. Once again the mother-and-son pairs E4 and E19 and E6 and
E18 were represented, and the difference in their usages highlighted the depar-
ture of the younger generation from the conservative norm, namely nasaliza-
tion after the article in the nominative.27
27 The symbol < represents an eventual self-correction in the direction the symbol indicates.
No such symbol appears between E16’s lenited and nasalized versions of /khɫɔxar/; she
gave them both as equally valid alternatives.
80’s E1 E1
70’s E2 E2
E4 E4
E5 E5
60’s E6 E6
E7 E7
E8 E8
E9 E9
50’s E10 < E10 E10
E11 E11
E12 E12
E14 E14
40’s E16 E16 E16
E17 < E17 E17
E18 E18
E19 E19
28 The prepositions themselves may or may not be mutating elements which cause leni
tion in a following indefinite noun; in ESG, for example, the prepositions {tə(L)} ‘to, for’
and {kən(L)} ‘without’ lenite an immediately following noun optionally. But the mutating
properties of the preposition have nothing to do with the mutation of the definite noun
after the preposition, which is a product of the case system alone. Thus the preposition
/ri/ ‘to’ does not itself lenite; witness the popular Embo proverb {LkaL ses phokh fɔɫi ri pɔɫ}
/xa hes phokh fɔɫi ri pɔɫ/ ‘An empty purse won’t stand against (to) a wall.’ But a preposi-
tional phrase with the definite noun, ‘against (to) the wall’, would be regularly lenited by
conservative speakers: {riš əL pɔɫ} /riš ə vɔɫ/.
Table 1.5 Responses to preposition plus article plus /kh/-initial noun (1968)
29 There are only 14 nouns in the prepositional tests; /khath/ ‘cat’ was inadvertently omitted.
E1 N L N L – – L
E2 N L N L N L
E3 N L N N1 L2 N L
E4 N L N N1 L2 N L
E5 L L L L N L
E6 N N N L N N
E9 N L N L N N
older speakers
younger speakers
E10 N L N N N L
E11 N N N N N N
E12 N N N L N N
E13 N L N L N L
E14 L L N N N L
E15 N L N N N N
E16 N L L L N N
E17 N L L L N N
E18 N N N N N N
E19 N L N N N N
norm in which /ɔrn/ governs the nominative rather than the dative case. It
is interesting to note that, in the original 1968 battery of prepositional tests,
only /er/ ‘on’, one of the component elements of /ɔrn/, also showed a weak
nominative-governing pattern (more speakers nasalized than lenited in three
of nine test cases). Otherwise all the other prepositions tested followed /as/
in favoring the dative. A reflection of this can be seen in the overall favoring
of lenition, by younger speakers as well as older, in the responses of Table 5.
Still, older speakers produced less than a third as many nasalizations as leni-
tions in Table 5, despite the high number of /ɔrn/-sentences, whereas younger
speakers in the same sentences produced nasalizations in better than half the
cases. And the age-grouped responses of Table 6 show clearly that the actual
change in progress in this part of the case system is a movement away from the
dative and toward the nominative. With the exception of the idiosyncratic E5,
who uniquely used the standard Gaelic form of the preposition /ɔrn/ and twice
lenited after it, no older speakers substituted lenition for nasalization after
/ɔrn/. On the other hand, older speakers gave a total of five nasalizations after
/as/, two of them as first-named variants but the other three as sole versions.
Younger speakers, for their part, showed three lenitions after /ɔrn/, but the
trend ran more than five times as strongly in the other direction: there were 16
instances of nasalization after /as/. The existence of a preposition which ‘gov-
erned the nominative’ in conservative ESG usage must have provided a ready
model for nasalization of definite masculines in the prepositional environ-
ment, and current younger-speaker usage is increasingly adopting that model.
Tape-recorded materials confirm the picture provided by Table 6,33 at the same
time offering evidence that a lenited dative is in fact the conservative norm
away from which current usage is moving. Older and younger speakers alike
show absolutely invariant lenition of the very highest-frequency masculines
in the dialect after every preposition but /ɔrn/. The masculines /pal/ ‘village’,
/khɫɔdax/ ‘shore’, /phɔrsth/ ‘harbor, ferry’, and /khũãn/ ‘sea’ never nasalize after
any preposition but /ɔrn/, regardless of the age of the speaker, and here again
we see a conservative norm reflected by the strong resistance to prepositional
nasalization with the highest-frequency masculines available. On the other
hand, leaving aside this special group of high-frequency nouns, the instance of
prepositional nasalization is much higher on tape for younger speakers than
for older. To contrast three Embo speakers well-represented speaking freely on
tape and widely separated in age: the octogenarian El uses the nominative after
the preposition /l´eš/ ‘with’ in addition to /ɔrn/, but otherwise uses the dative;
33 Lengthy tape-recorded reminiscences and stories were collected from as many informants
as were capable of speaking comfortably with a recorder going. Most of the tapes were
made in the form of interviews between the investigator and the informants, in which
questions were asked and answered. The exceptionally interesting tape from which the
data on E14 and E16 prepositional usage is taken was made in the absence of the investiga
tor (but for her benefit) and represents a husband reminding a wife eight years his junior
of how things used to be when they were young and Embo was still very much a fish
ing village. The tape from which the scanty material on E5 and E13 is taken was a letter-
tape made by those two speakers and the investigator for an emigrant sister in Australia
who had been home to visit the year before. The tape was therefore in no way made for
research purposes, but by permission of the two Embo speakers a copy was made to that
end before the original was mailed.
the husband-and-wife team E14 and E16 not only nasalize four-to-one after /
l´eš/, but they also nasalize all but the highest-frequency masculines after /ig/
‘at’, and they show sporadic nasalization after /kus ~ kəs/ ‘to’, /as/ ‘in’, /er/ ‘on’,
and /thrε̃/ ‘through’. To take another case, the brother-sister pair E5 and E13 are
poorly represented on tape; yet even in the meager material available, their age
difference is reflected in the fact that E5 lenites two of three masculines, while
E13 nasalizes four of five.
Masc Fem.
Nom./Acc. N L
Gen. L L
Dat. L L
Masc Fem.
def. art. N L
Masc Fem.
Nom. N ~ L L
Gen. N ~ L L
Dat. N ~ L L
phrase; and it, too, has lost the (rare) word-final palatalizations with internal
vowel change which one declension shows in the standard language. Case as
a surface-structure phenomenon has virtually disappeared in ESG. Second,
under these circumstances, the definite article ceases to take on varying mor-
phological shapes within a case system, and moves toward the status of so
many other morphemes in the language: it becomes a paradigmatically iso-
lated morpheme, much like {kle:L}, {ta:L} etc. Grammatically it is not redun-
dant, however, since it signals the gender of the following noun.
Once again we find that the loss or threatened loss of certain mutational
interrelationships causes no informational loss to the language. In a suggestive
article on Gaelic grammar, Borgström notes (1968: 13), speaking of the fuller
system of western dialects: “The intricate morphology of [Gaelic] cases . . . is
only an intellectual play without much informative value”. As in the verbal sys-
tem, nouns with unmutatable initials occur plentifully, and there is sufficient
grammatical apparatus available outside the mutational system to make the
syntactic role of any given noun abundantly clear. ESG replaces the genitive
of definite-noun attributives by prepositional phrases, and the prepositions
themselves make the “dative” lenition of masculine nouns informationally
unnecessary.
The striking thing about the mutations of the ‘case system’ in ESG is not
that syntactic expansion of the nominative is giving rise to a single mutated
definite-noun form dependent on gender, but that the expanding nominative
preserves a gender-distinctive mutation in a dialect in which gender is itself a
rapidly weakening category (see fn. 25).
his silence on the subject, it seems safe to assume that Breatnach 1964 also
found no loss of mutation in his search for terminal characteristics in areas of
relic Irish Gaelic speech. Perhaps the relative independence of initial muta-
tion from specific grammatical categories has even in the end contributed to
its longevity. At any event, it is undeniable that it has survived far better in ESG
than the word-final consonant mutations (chiefly palatalizations) that were so
intimately linked with the genitive and dative cases; even the coincidence of
some of these final mutations with the final mutations of irregular plural for-
mation did not operate to preserve the phenomenon outside the plural itself.
nor the Brora Gaelic speakers enjoyed a period of social and linguistic isolation
comparable to Embo’s, and, perhaps as a consequence, the full-scale changes
traceable in Embo are only embryonically present in Golspie and Brora. In
fact, precisely because of the lack of a younger generation of fluent bilinguals,
a description of the ESG dialect in Golspie and Brora is conspicuously more
difficult, because less obviously and clearly patterned, than is a description of
ESG based primarily on Embo materials.
Studies of western Gaelic dialects have generally aimed at codifying a pho-
nological and grammatical norm for the communities concerned, which has
meant that they were by definition unlikely to deal extensively with variability
in usage. Detailed studies which focus on variation and change are needed for
western dialects before the seeming rapidity of change in ESG can be properly
assessed.
References
Borgström, Carl Hj. 1968. Notes on Gaelic grammar. Celtic Studies, ed. by James Carney
and David Greene, 12–21. London: Routledge & Kegan Paul.
Breatnach, R. B. 1964. Characteristics of Irish dialects in process of extinction.
Communications et rapports du Premier Congrès de Dialectologie Générale, 141–5.
Louvain: Centre International de Dialectologie Générale.
Fishman, Joshua A. 1968. Readings in the sociology of language. The Hague: Mouton.
Gauchat, Louis. 1905. L’unité phonétique dans le patois d’une commune. Aus romani-
schen Sprachen und Literaturen: Festschrift Heinrich Morf, 175–232. Halle: Max
Niemeyer.
Holmer, Nils. 1938. Studies on Argyllshire Gaelic. Uppsala: Almqvist & Wiksell.
Hymes, Dell H. 1962. The ethnography of speaking. Anthropology and human behavior,
ed. by T. Gladwin and William C. Sturtevant, 13–53. Washington, D.C.: Anthropo-
logical Society of Washington. Reprinted in Fishman, 99–138.
Kirparsky, Paul. 1968. Linguistic universals and linguistic change. Universals in linguis-
tic theory, ed. by Emmon Bach and Robert T. Harms, 171–204. New York: Holt,
Rinehart & Winston.
Labov, William. 1966. The social stratification of English in New York City. Washington,
D.C.: Center for Applied Linguistics.
Lehmann, Winfred P. 1964. Historical linguistics. New York: Holt, Rinehart & Winston.
Maclaren, James. n.d. Gaelic self-taught. Third edition. Glasgow: Alex MacLaren.
Oftedal, Magne. 1956. The Gaelic of Leurbost, Isle of Lewis. (A linguistic survey of the
Gaelic dialects of Scotland, 3; Norsk Tidsskrift for Sprogvidenskap, supplementary
vol. 4.) Oslo: Aschehoug.
Stevick, Robert D. 1963. The biological model and historical linguistics. Language 39.
159–69.
Zengel, Marjorie S. 1962. Literacy as a factor in language change. American
Anthropologist 64.132–9. Reprinted in Fishman, 296–304.
1 Introduction
In recent years, increasing attention has been focused on pidgin and creole
languages, in part as a kind of proving ground for both linguistic and sociolin-
guistic theory. Not least among the interests of recent writers have been issues
of simplification and convergence as linguistic processes; witness the very sub-
stantial third section of Hymes 1971. In its preface, Hymes writes of four
“ ‘moments’ which a theory of pidgin and creole languages must integrate”. Of
these, the first two are: “(1) the universal tendencies to adapt speech, and vari-
eties of a language, by simplification in some circumstances, expansion in oth-
ers; (2) the occurrence of these tendencies in situations of language contact, so
as to give rise to partial confluence of linguistic traditions”.
While fully acknowledging the great value of pidginization and creolization
studies in the investigation of simplification (and/or elaboration) and conflu-
ence in language use, I submit that the study of language death has much to
offer in these same areas of investigation, and that so far it has been much too
little tapped as a source of information in these matters. This is not to say that
simplification and confluence appear in language death in the same degree, at
the same points – or for the same reasons – that they do, say, in pidginization.
Indeed, I hope to show in this paper that they need not. But this, I think, only
makes it the more important that we include the special case of language
death when we venture on the topics of simplification and confluence.
It has, of course, long been recognized that dying languages characteristi-
cally show reduction of one kind or another – or, most often, of many kinds at
once. Early reports tended to be quite general and to be impressionistically
rendered. Thus Bloomfield 1927 characterized White-Thunder’s Menomini as
“atrocious”: “His vocabulary is small; his inflections are often barbarous; he
constructs sentences of a few threadbare models.” Krauss 1963–70 offers many
comments on failings in the Eyak texts he collected from a last few speakers,
but most of them are general rather than specific (“inappropriate here”, “dis-
torted”, “confused towards end”, all from p. 44); and they are not systematized
to show in what ways the language is suffering changes in its patterns. Miller
1 The gerund is normally known as the ‘verbal noun’ in the study of Gaelic; but ‘gerund’ is
adopted here for brevity, and for the sake of maximum terminological differentiation of the
two structures under discussion.
s imilarly with gerunds. How far this is from the actual case in the dying dialect
at hand will appear in due course.
The terminal speech form which provides the data for this study is the East
Sutherland dialect of Scottish Gaelic. Because it exists in considerable isola-
tion from all other Gaelic dialects, and because it is succumbing not in compe-
tition with some more prestigious or normalized variety of Gaelic, but rather
in competition with an entirely different language (namely English), East
Sutherland Gaelic (ESG) can, for our purposes, be termed a dying language.
That is, the struggle for dominance, and currently for survival, involves two
languages, Gaelic and English. Each is in fact represented by a regional variety
of a language of greater national currency. While East Sutherland English will
be called simply English throughout, since its structure is essentially that of
standard English in the formation of noun plurals and gerunds, East Sutherland
Gaelic will be identified by means of the local label because it differs in some
of the particulars discussed here from the standard Gaelic of textbooks.
ESG is spoken by fewer than 150 people, all of them bilingual in English, on
the east coast of the county of Sutherland in the extreme north of mainland
Scotland. In the two largest East Sutherland villages, Brora and Golspie (pop.
1200–1300 in each case), less than 5% of the population speaks Gaelic. Fluent
speakers in these villages are between 70 and 80 years of age. Between the ages
of about 45 and 65, a small number of speakers can make themselves under-
stood in Gaelic; but their Gaelic, in terms of the norms of the older group, is
imperfect in many ways. These I have called semi-speakers (Dorian 1973: 417).
In the little village of Embo (pop. ca. 275), the same two groups exist, but the
semi-speakers here are between about 35 and 45; transitional between the
older fluent speakers and the semi-speakers is a group of younger fluent speak-
ers whose Gaelic is very good, yet shows certain departures from the conserva-
tive norms of the older fluent speakers (Dorian 1973). That Embo should have
developed a wider spectrum of Gaelic proficiency probably reflects its greater
isolation, smaller size, and lower number of resident English speakers. Perhaps
half the adult population of Embo speaks Gaelic, and the death of the lan-
guage will be slightly more protracted there. But Gaelic is dying as surely in
Embo as in the other two villages – since, as in Brora and Golspie, no young
people are acquiring the language.
In order to determine whether changes were appearing in the complex mor-
phology of noun plurals and gerunds in ESG, a large number of test sentences
designed to elicit those structures were presented in English, for translation
into Gaelic, to members of the three groups:2 older fluent speakers (o.f.s.),
2 A distinction between o.f.s. and y.f.s. is not made by the communities themselves; it is estab-
lished on the basis of differences in grammatical usage between age-groups, as reported in
younger fluent speakers (y.f.s.), and semi-speakers (s.s.) The sample consisted
of 4 o.f.s., 4 y.f.s., and from 5 to 8 s.s. The two fluent-speaker groups remained
constant throughout the study; a core of 5 s.s. also remained constant, while
three others were or were not represented according to their availability when
the tests were administered – and, in the most extreme case, according to the
speaker’s ability to produce the material. (That is, one “s.s.” participated in only
the briefest and simplest subset of test sentences.)
For fluent speakers, testing which requires translation of English stimulus
sentences into Gaelic poses no problem. All are accomplished bilinguals, and
the position of Gaelic in local life is such that translation between Gaelic and
English is a commonplace activity, since even kin-linked interaction groups are
likely to include both monolinguals and bilinguals. S.s., on the whole, find the
translation of relatively simple sentences from English to Gaelic – and only
relatively simple sentences can be used, if responses are to be expected across
all three groups – to be the most congenial of the three tasks I have thus far
asked of them (these were production of isolated Gaelic words from the lexico-
statistical list of core vocabulary, translation of simple English sentences into
Gaelic, and free Gaelic conversation in interviewing). Because of the restricted
role of Gaelic in the lives of the s.s., no elicitation of Gaelic from them could be
without stress. They are not regular speakers of Gaelic, and even the best of
them use Gaelic only in brief bursts, interrupted by longer stretches of English
in their interactions outside test situations. From my considerable interaction
with some of the s.s., I believe that their test results in these translation tests
are closer to the upper end of their proficiency spectrum in Gaelic than to the
lower end; e.g., one s.s. steadfastly professes not to be a speaker of Gaelic at all,
and expresses amazement that she can regularly produce the sentences I cajole
her into attempting. Free conversation in Gaelic would be less likely to produce
good results for such an informant than the translation of single, relatively
simple, set sentences.
It will be useful first to consider the position of the noun plural and the gerund
in the Scottish Gaelic sentence. The ESG noun must be expressed in the plural
in all places where a noun would be plural in English, except that the singular
Dorian 1973. The communities do show some explicit recognition of a group corresponding
to the s.s. here, and these community judgments of less-than-fluent proficiency prove to have
demonstrable linguistic correlates. Very fine discriminations can be made by the community
and confirmed by linguistic testing (Dorian 1977b).
appears after certain numerals (2, 20, and any multiple of 20; the numbers 3
through 5 for certain high-frequency nouns). Syntactically there is no agree-
ment between number in the noun and number in the verb, because the verb
does not show number. The definite article shows number, but adjectives and
demonstratives do not; if the noun phrase includes the definite article, then
number will be expressed twice, once in the article and once in the noun itself:
Pronoun replacements for the noun also show number, e.g. ‘the window’ –
‘she, it’;3 ‘the windows’ – ‘they’.
Far and away the commonest use of the Scottish Gaelic gerund, and the only
use tested for this study, is formation of the progressive verb phrase. Progressive
aspect is much more prominent in Gaelic than in English. It applies to verbs
which would not normally permit progressive aspect in English (‘see’, ‘hear’,
‘understand’, ‘believe’); and it is often used with the inflected verb ‘to be’ in
periphrasis, in lieu of inflected forms of the lexical verb itself:
These are all derived from the root /priš/ ‘break’ (with gerund /prišu/).
The non-Celtic reader will have no trouble in accepting the prominence of
the noun plural in any normal stretch of Gaelic speech. Because of the extreme
preference for progressive verb phrases, the gerund achieves a prominence
very nearly as great. Consequently there is no difficulty whatever in devising
3 ‘Window’ is grammatically feminine in ESG, but feminine pronoun replacements are giving
way to masculine in the singular (the plural pronoun does not express gender). This seems in
fact to be a case of confluence; ESG /a/ ‘he, it’ is associated with English it and extended to all
inanimates.
simple stimulus sentences which call for noun plurals and gerunds; and where
either plural or gerund is difficult for the informant to produce, it is strictly the
token, and not the type, which causes the difficulty.
Table 2.1 below presents the morphological devices which operate in the for-
mation of the noun plural and the gerund, as elicited from o.f.s. (taken as rep-
resenting a conservative norm of sorts) in the test sentences.
One device which may require a word of explanation is final mutation. In
Scottish Gaelic, this is the substitution of a different final consonant for the
one in the root of the word. ESG often substitutes a palatal equivalent of the
original, sometimes with a concurrent change (from back to front) in
the immediately preceding (and in that case always lightly stressed) vowel. Or
the reverse may occur: a non-palatal equivalent is substituted for a palatal pho-
neme. The set of substitutions is limited and fixed, and occurs with a restricted
set of words; i.e., it is not productive. The consonant substitutions which are
typically word-final may occasionally become internal through the addition of
a suffix; since the same pattern of substitutions appears, the term final muta-
tion will be stretched to cover such cases.
Table 2.1 shows some overlap in morphological devices for the formation of
the noun plural and gerund: four of the first six devices are common to both. In
noun plural formation, only syncope (in type X) is wholly unique; in gerund
formation, only subtraction (in types VI–X) and the zero formation (in type XI).
But vowel alternation and quantity change, which appear independently as
devices in the formation of noun plurals, appear only in combination with
other devices in the formation of gerunds. With regard to suffixation, the num-
ber of different suffixes which are used as the sole device in the two structures
also proved to be remarkably similar (though the forms are almost entirely
distinct): 9 different suffixes appeared independently in forming the noun plu-
ral, and 8 in forming the gerund (though neither figure represents a maximum
number of suffixes for the structure). Among the exemplars of Table 2.1, two of
the gerund formations elicited are idiosyncratic, supplied by only one o.f.s. in
the entire sample; moreover, they are idiosyncratic even in the much larger
sample which I have used for other work on ESG. Thus, for most speakers, sub-
traction actually combines only with vowel alternation plus suffixation (ger-
und type IX) and with suffixation (gerund type X).
Table 2.1
Noun Plural
Singular Plural
Gerund
Root Gerund
Table 2.1 (Continued)
Gerund
Root Gerund
VII. subtraction
+ quantity change /e:riç/ /eri/b ‘rising’
VIII. subtraction
+ final mutation
+ suffixation /kɫu:əškh/b /kɫu:əsth/ ‘moving’
IX. subtraction
+ vowel alternation /khrəǰ/ or
+ suffixation /khrεǰ/ /khresčən/ ‘believing
X. subtraction
+ suffixation /e:šiç/ /e:šnax/ ‘listening’
XI. zeroc /khrεkh/ /khrεkh/ ‘selling’
a Final mutation in gerund formation was normally accompanied by vowel alternation in the
most conservative ESG usage of recent years (that of speakers now dead), in the presence of
the suffix /-u/. Final mutation without accompanying vowel alternation is now commoner,
even among o.f.s., before /-u/.
b This is an idiosyncratic form supplied by only one o.f.s. in the sample.
c ‘Zero’ means no morphological change; the gerund is identical with the root.
Although both noun plural formation and gerund formation are morphologi-
cally rich, the number and prominence of nouns or verbs which can use one or
another of these formations differs markedly. Furthermore, the productivity of
the formations is anything but uniform, even within the conservative norm
provided by o.f.s.
Many noun plurals and gerunds are formed by suffixation, but of the 8 or 9
suffixes available in each case, only one or two are productive for fluent speak-
ers in the sense that they attach readily to loanwords or serve as analogical
alternants for rarer forms. No formations other than suffixation are productive
in these senses, either for nouns or verbs. Some non-productive formations
nonetheless apply to large numbers of nouns or verbs. Quantity change is a
very common noun plural formation (type XI) for fluent ESG speakers, and the
zero model is very common in conservative gerund formation (type XI). On the
other hand, the various plurals with vowel alternation (types VI–X) involve
relatively few nouns; but like their analogs in English (the foot/feet group), they
include some of the highest-frequency nouns in the language, which makes
the device of vowel alternation feel more important than it is statistically – and
similarly with suppletion among the gerunds (type III).
Consequently, if we find statistical advances in the use of the productive
suffixes, it will scarcely be surprising; and it is at least conceivable that we
might find some statistical advance in the use of quantity change in noun plu-
rals and the zero formation in gerunds. However, a reduction in vowel alterna-
tion in plural formation or suppletion in gerund formation could represent a
striking change in the dialect without looming very large statistically, because
of the prominence of the words concerned.
If we now look in detail at the noun plural formations produced by the three
groups of speakers, and then in similar detail at the gerund formations of the
three groups, we will be able to follow certain developments in this rich area of
terminal ESG morphology.
Table 2.2 gives the noun plural formations of all three groups of speakers. The
percentage of all noun plural formations within the test sentences is given for
the formation type, and beside it the number of actual occurrences. The table
is arranged in descending order of frequency according to the usage of the
o.f.s., who represent a sort of conservative norm.
Table 2.2
Table 2.2 (Continued)
Table 2.3
SPEAKER GROUP # %
o.f.s. 134 50
y.f.s. 118 44
s.s. 292 63.5
Table 2.4
TABLE 2.5
Type II 28 10 24 9 23 5
Type VII 24 9 28 10.5 11 2.5
Type IX 3 1 4 1.5 7 1.5
Type V 2 1 3 1 6 1
Vowel alternation and final mutation can in fact be seen to have declined in
popularity as pluralization devices in general among s.s. – separately and in
Table 2.6
long 34 87 29 88 8 35
half-long 4 10 4 12 5 22
short 1 2.5 – – 10 43
The fate of this group of plurals is in one sense different from that of all the
others: its decline rests on a phonological development among s.s. – a very
heavy loss of the phenomenon of consonant length in general, and a consider-
able weakening of the phenomenon of vowel length. Phonologically, the s.s.
show a marked tendency to lose phonemes and prosodemes of Gaelic which
are not shared by English; thus they generally fail to use the ‘dark’ velarized /ɫ/
or the palatal fricative /ç/, and vowel nasalization as well as vowel and conso-
nant length become sporadic or disappear. Where quantity change is con-
cerned, this means that, despite the frequency of the device in noun
pluralization among fluent speakers, there can be no question of any increase
in the phenomenon among s.s. However, there are at least two ways in which
this group of nouns might be treated, for pluralization purposes, by the s.s.:
addition of a suffix compatible with their phonotactic shape, or loss of all overt
plural markings. The latter is the usual development, and brings us to the next
morphological device.
4.3. Zero plurals, which are unknown among o.f.s. and highly exceptional
among y.f.s., constitute 9% of all s.s. plurals – a striking development. Of the
42 zero plurals supplied by the s.s., 18 (or 43%) derive from the loss of final
consonant length in plurals where such length would be the sole pluralization
marker among conservative speakers (type XI). Back-formation of a singular
identical to the /-(V)n(:)/ plural, with loss of final consonant length, accounts
for three more instances.4 The rest of the s.s. zero plurals (50%) are indepen-
dent of developments in consonant length.
4.4. While certain rarer o.f.s. suffixes are totally unrepresented among y.f.s.
and s.s., none of the pluralization devices as such (or combinations) in use
among o.f.s. is completely absent in the other two groups – not even those
constituting less than 5% of the o.f.s. plurals, like suppletion or final mutation
plus /-εn/.
4.5. Each of the three groups of speakers has some idiosyncratic plural for-
mations (all cases are of idiosyncratic suffixes, except the three y.f.s. instances
of vowel shortening plus /-in(:)/); but the s.s. have proportionately slightly
more. The percentages for the three groups respectively, in the usual order, are
2, 2, 4.5.
5 The Gerund
4 Fluent speakers, e.g., have sg. /khu:rsti/, pl. /khu:rstin:/; two s.s. have sg. and p1. /khu:rstin/.
Table 2.7
/-u/ (I)a 41 26 38 24 62 27
zero (XI) 25 16 27 17 42 18.5
final mutation + vowel 16 10 20 12.5 4 2
alternation (II)
/-al/ (I) 14 9 16 10 58 25.5
vowel shortening + 9 6 4 2.5 3 1
/-u/ (IV)
vowel shortening + 8 5 9 5.5 11 5
/-al/ (IV)
/-tən/ (I) 8 5 11 7 6 2.5
subtraction (VI) 7 4.5 9 5.5 11 5
/-šən/ (I) 4 2.5 1 .5 – –
suppletion (III) 4 2.5 4 2.5 7 3
/-əm/ (I) 4 2.5 4 2.5 10 4.5
final mutation + 3 2 4 2.5 – –
suffixation (V)
vowel alternation + 3 2 4 2.5 – –
subtraction +
/-sčən/ (IX)
/-i/ (I) 2 1 3 2 3 1
/-ən/ (I) 2 1 – – 2 1
/-d/ (I) 2 1 – – – –
vowel shortening + 1 .5 1 .5 – –
/-tən/ (IV)
subtraction + 1 .5 3 2 1 .5
/-nax/ (X)
vowel shortening + 1 .5 – – – –
subtraction (VII)
subtraction + 1 .5 – – – –
final mutation
+ /-th/ (VIII)
vowel alternation – – – – 3 1
/-kən/ (I) – – – – 1 .5
(Continued)
Table 2.7 (Continued)
/-iç/ (I) – – – – 1 .5
/-in/ (I) – – – – 1 .5
/-x/ (I) – – – – 1 .5
infixation – – – – 1 .5
5.1. The percentage of all gerunds formed by simple suffixation (suffixation alone,
type I) has risen among s.s. as compared with the other two groups of speakers,
as shown in Table 2.8. Virtually all this gain is caused by the increasing use of the
suffix /-al/ among s.s., for whom it has become exceptionally productive.
Table 2.8
SPEAKER GROUP # %
o.f.s. 77 49
y.f.s. 73 46
s.s. 145 63.5
5.2. Among the devices of relative prominence in the formation of the o.f.s.’
gerunds, the one which shows the greatest decline in use is final mutation plus
vowel alternation (type II), which drops off sharply in s.s. use: 10%, 12.5%, 2%.
This same decline appears if we tabulate all occurrences of final mutation,
in combination not only with vowel alternation but also with other devices
(types II, V, VIII), as in Table 2.9.
Table 2.9
SPEAKER GROUP # %
o.f.s. 20 13
y.f.s. 24 15
s.s. 4 2
5.3. Change in vowel quantity in combination with suffixation (type IV) has
declined in gerund formation, as shown in Table 2.10, even when we omit from
consideration the single idiosyncratic o.f.s. case of vowel shortening plus sub-
traction (type VII; cf. Table I, fn. b).
Table 2.10
SPEAKER GROUP # %
o.f.s. 18 11.5
y.f.s. 14 9
s.s 14 6
Table 2.11
It is important to note that the marked increase in the use of simple suffixation
by s.s. is not foreshadowed by any comparable increase among y.f.s. as com-
pared with o.f.s.; on the contrary, y.f.s. use simple suffixation slightly less in
each case. That is, the rise in simple suffixation is not part of a gradual contin-
uum of change within the language, but a phenomenon characteristic of, and
peculiar to, s.s. use of the language. Indeed, the only plural formation which
shows much increase among y.f.s. is type VIII, vowel alternation plus suffix-
ation. Such a shift in favor of a doubly-marked plural is altogether in line with
Kury�owicz’s first ‘law’ of analogical change (1966: 162) – whereas the changes
evinced by s.s., changes thus perhaps characteristic of language extinction in
general, are never in the direction of Kury�owicz’s ‘morphème bipartite’, either
in noun plurals or in gerunds.
Also of interest is the fact that the rise in simple suffixation, relative to all
other formation types, results very largely (indeed almost exclusively) from the
increased use of a single ‘favored’ suffix: /-ən/ in the noun plural, and /-al/ in
the gerund. That is, the sense of plurality in nouns, and gerund formation in
verbs, resides for s.s. increasingly in one strongly productive allomorph. In the
case of /-ən/, the s.s. are ‘over-using’ a suffix which is by all odds the common-
est noun plural suffix in ESG (though not, as it happened, the commonest one
in the test sentences for any but s.s.).5 The case of /-al/ is the more interesting,
because here the s.s. are favoring a suffix which is not normally the commonest
5 The statistical preponderance of the /-ən/ plural allomorph was underplayed in the test sen-
tences, so as to avoid the planting of a built-in model for analogical change in the test itself.
gerund-forming suffix in the language at large (as indeed it is not for any group
of speakers in the test sentences). In this regard, the s.s. again demonstrate
their relative independence of the y.f.s. position, since the y.f.s. group shows
only very small differences from o.f.s. norms for the /-u/ and /-al/ gerund
allomorphs.
Whereas simple suffixation shows a marked gain in use among s.s., the
peculiarly Gaelic device of final mutation shows a decline in use, as shown in
Table 2.12. This is despite the fact that, like simple suffixation but unlike vowel
alternation or subtraction, it plays a modest role, both alone and in combina-
tion with other devices, in forming BOTH noun plurals and gerunds in the usage
of conservative speakers.
Table 2.12
PLURAL GERUND
SPEAKER GROUP # % # %
o.f.s 57 21 20 13
y.f.s 59 22 24 15
s.s. 47 10 4 2
Again it is conspicuous that the drop-off in the use of final mutation as a mor-
phological device is both characteristic of and peculiar to s.s.; it is not conso-
nant with the behavior of y.f.s., who actually show slightly more use of final
mutation than the o.f.s. in each case. The reasons for the lesser use of final
mutation as a gerund-forming device than as a plural-forming device in all
three groups of speakers, when in theory ESG offers more opportunities for
gerund formation by final mutation,6 are complex and interesting, but outside
the scope of this paper. The dramatic falling-off of gerund formation by final
mutation among s.s. vis-à-vis fluent speakers reflects the transfer of many
verbs originally of this class into the suffixation class.
Vowel alternation in the noun plural (and on a much smaller scale in the
gerund) and subtraction in the gerund have also fallen off somewhat in the
usage of s.s., but the drop in the use of subtraction is found only where it is
combined with other devices. The drop in the use of vowel alternation is rather
more marked, however, and again shows the pattern of s.s. decline as a break-
away development, unrelated to y.f.s. usage; see Table 2.13.
6 ESG has more verb roots in final /-iç/, to give potential gerunds in /-ax(u)/, than noun roots
in final /-ax/, to give potential plurals in /-iç/.
Table 2.13a
PLURAL GERUND
SPEAKER GROUP # % # %
o.f.s. 70 26 19 12
y.f.s. 85 31.5 24 15
s.s. 68 15.5 7 3
a Note that there is an overlap here with Table 12, since some vowel alternations occur in
conjunction with final mutation in noun plurals (types VII, IX) and in gerunds (type II).
Table 2.14
PLURAL GERUND
SPEAKER GROUP # % # %
o.f.s. 49 18 19 12
y.f.s. 48 18 14 9
s.s. 27 6 14 6
definite article (13 instances) or where the definite article was called for but
omitted (8 instances). Thus failure to mark the noun plural morphologically is
not wholly explained by a built-in redundancy.7
A second possible explanation is intralinguistic influence from one class of
words to another. A goodly number of morphological devices are common to
both noun plurals and gerunds; perhaps the s.s. are simply extending this over-
lap to yet another device (as with the introduction of vowel alternation alone
into gerunds, noted above). Or, as a third explanation, there is the possibility of
interlinguistic influence. English, of course, has a small class of nouns, largely
the names of edible animals (deer, sheep, fish etc.), which require no plural
marker. The fact that English is the dominant language of all the s.s. may make
it so much the easier to develop a parallel zero plural in Gaelic. As it happened,
four s.s. instances of zero plural involved edible animals (rabbit, lamb); two
others involved a domestic animal (horse). Most probably all three of the fac-
tors suggested operate conjointly to produce the s.s.’ predilection for zero
plurals.
Finally, it is perhaps worthy of comment that suppletion – a device which is
itself used on a very small scale in the formation of both noun plurals and ger-
unds by o.f.s., and does not combine with any other device in either type of
construction – shows no appreciable change across the three groups of speak-
ers in either case. Only two nouns with suppletive plurals in conservative usage
appeared in the test sentences, one a very high-frequency noun, the other
much less so. Most s.s, retained the high-frequency suppletive plural; most lost
the low-frequency suppletive plural. It should be noted, however, that high fre-
quency of the noun is not a guarantee of the preservation of a non-productive
plural formation. Some very high-frequency nouns which have vowel-alterna-
tion plurals for fluent speakers developed suffixal plurals for most s.s., or joined
the zero-plural group. Suppletion is undoubtedly reinforced in gerund forma-
tion by two factors: the verbs which form their gerunds by suppletion are truly
among the very highest in frequency in the language; and their other forms
7 The nouns which appeared as zero plurals for the s.s. ran rather heavily to ones which would
have been consonant-length plurals (type XI) for fluent speakers (16 instances out of 42).
Almost exactly as many were of the vowel alternation type, with or without suffixation (types
VI and VIII, 15 instances). Scarcely any were of the suppletive type (type III; two instances of
the same noun, from the same speaker); and, as one would expect, relatively few were of the
simple suffixation type (type I; 7 instances). The final mutation type, with or without suffix-
ation and vowel alternation (types II, V, IX), supplied only two instances (separate instances
of the same noun, for two different speakers).
(“principal parts”, so to speak) are also suppletive, even more so than the
gerunds.
7 Conclusions
It is clear from §6 that the s.s. group emerges as very much the innovators in
terms of changes in the part of the ESG morphological system which we have
examined. (This is in fact the richest part of that system; no other part of the
inflectional or derivational morphology of ESG shows such complexity.)
Although I have been at pains elsewhere (Dorian 1973) to demonstrate that the
y.f.s. do not always coincide in their usage with the o.f.s., in the complex mor-
phology of noun plurals and gerunds they are seldom far from the o.f.s. norm.
The s.s., however, show many departures from that norm. One might suppose
that the linguistic discontinuity of the s.s. would be a linguistic parallel to a
similar generational discontinuity in social life, i.e. that s.s. would interact rela-
tively little with fluent speakers, forming a group apart, both socially and lin-
guistically. This is not the case. The s.s. do have peer-group ties to one another
in several cases, but their ties to the older generation, especially to the genera-
tion of their grandparents, are so conspicuous, in the linguistic autobiogra-
phies that I have collected from a number of them, as to lead me to propose,
only half-jocularly, a ‘grandmother factor’ in the genesis of the s.s. as a linguis-
tic phenomenon, at least in ESG. Some social explanation is, after all, required
to account for the fact that these younger members of the community, whose
command of Gaelic is weak and imperfect by their own ready admission, con-
tinue to be willing (eager, in a number of cases) to speak a language which is
clearly dying and has extremely negative prestige on the local scene. Repeatedly,
they report a favorite female figure (other than the mother) in the first or sec-
ond ascending generation – usually a grandmother, but sometimes a great-
aunt or a cousin at one or two removes – with whom Gaelic was the sole or
favored language of communication in their early years.8 Most of them main-
tain close ties with older Gaelic-speaking kin. Social discontinuity is, then, not
the explanation here for linguistic discontinuity.
The formal discontinuity which we note does correlate positively with a
functional discontinuity, however, as the very labels ‘fluent speaker’ and ‘semi-
speaker’ imply. Fluent speakers are habitual users of Gaelic. The settings in
which ESG can be used appropriately have been drastically reduced in the past
8 Michael Silverstein (p.c.) reports that this is ‘exactly the case over and over in [his] Wasco-
Wishram Chinookan work as well’.
fifty years (from work, religion, community, and home to only the last in Brora
and Golspie, and only the last two in Embo); but the fluent speakers, even the
youngest, do use Gaelic almost exclusively as a home language. This is true for
none of the s.s., although three of them live in settings where they could achieve
regular home use of Gaelic if they wished it and felt capable of it. The s.s. are
all speakers of English first and foremost, who reserve their use of Gaelic for
interaction with a few highly specific individuals (mothers or other older-gen-
eration kin; in one case a spouse) and a few equally restricted settings (most
notably, settings where Gaelic serves a concealment purpose, such as joint
shopping trips or joint journeys on public transportation).
The functional discontinuity observed for ESG is perhaps reflected here in
the fact that the rich, even over-rich, morphological categories studied prove
resistant to marked change until the lower reaches of the proficiency contin-
uum of speakers. This is particularly striking because the morphological com-
plexity of the two structures concerned would seem almost to invite
simplification via leveling, and certainly to provide the means: simple suffix-
ation, which is clearly attractive to the s.s. as a device, could easily be extended
to account for all noun plurals and gerunds. Y.f.s., who do show departures
from the o.f.s. norm in other aspects of ESG structure, seem almost entirely
impervious to the possibilities for simplification in these complex morphologi-
cal structures. These y.f.s. not only share, both actively and passively, the mor-
phological norms of the o.f.s., they also share their norms for the use of Gaelic.
Fluent speakers, both older and younger, adhere to the norm of habitual home
use, as noted above, while s.s. are deviant in this regard. Evidently the func-
tional factor of habitual home use is sufficient to prevent the onset of morpho-
logical simplification among y.f.s., while the lesser use of Gaelic among s.s.
facilitates it. S.s. are not altogether unaware of fluent-speaker morphological
norms. They often react strongly to a conservative usage heard after complet-
ing test sentences with me, saying ‘That’s what I should have said!’ or ‘That’s
the right thing, isn’t it?’ But they share at most receptive morphological norms
with the fluent speakers, not productive ones.
Other observations can be made about simplification in ESG morphology as
examined here. Simplification certainly exists in the morphological perfor-
mance of the s.s.; the marked rise in the use of simple suffixation (especially a
rise produced by a single “favored” suffix) is a clear case of a movement toward
simplicity in a highly complex morphology. Nonetheless, a great deal of that
complexity remains. Final mutation, vowel alternation, suppletion, quantity
change, subtraction, zero formation – all are still in use among s.s., although to
a lesser extent than among fluent speakers. If one accepts Samarin’s concept
(1971: 119, 123) of “substantive pidginization” and its wide distribution, then one
Table 2.15
9 Cf. Reinecke on Pidgin French in Vietnam (1971: 51–2): “Except for a few isolated set forms,
S[tandard] F[rench] inflection has been dropped”; “the P[idgin] F[rench] verb is
invariable.”
10 Three of the group are relatively strong speakers, as s.s. go; two are of intermediate profi-
ciency; two are quite weak speakers; one is so weak as to have been barely usable and
should not really be termed a “speaker”.
in this chapter. Indeed, ESG might be said to be dying, at least with regard to
noun plurals and gerunds, with its morphological boots on. In the absence of
detailed comparable studies of other dying languages, it is hard to interpret the
resistance of ESG to large-scale morphological simplification. A number of fac-
tors seem to be of possible significance: the relatively protracted obsolescence
of the language, with the number of speakers dwindling only gradually; the
strong inter-generational ties characteristic of this community, with women of
the second ascending generation playing a key role in the linguistic socializa-
tion of the s.s.; and the “integrative” rather than “instrumental” role (see
Lambert 1967: 102) of Gaelic in East Sutherland, where the language is emblem-
atic of membership in a particular subgroup in the population – the fisherfolk
and their descendants – but no longer serves any economic purposes. Pidgins,
by contrast, tend to spring into being rather quickly and to have a function
considerably more “instrumental” than “integrative”; one may suppose that
they figure less often in wide cross-generational linguistic socialization.
The fact that ESG, in its terminal state, behaves so differently from the “typi-
cal” pidgin in terms of morphology bears out my contention that language
death needs to be added to – but not equated with – pidginization as a source
of data on simplification and confluence in language contact. But it is also
clear that we need studies of language extinction as rich and broad as those on
pidginization before we can venture generalizations about the circumstances
under which dying languages will in fact resist wholesale morphological sim-
plification and confluence. Whether all the factors noted above are necessary
to the preservation of morphological complexity, or some one or two of them,
or whether still other factors not evident here are often operative, remains to
be determined.
References
Bloomfield, Leonard. 1927. Literate and illiterate speech. American Speech 2.432–39.
Dorian, Nancy C. 1973. Grammatical change in a dying dialect. Lg. 49.414–38.
———. 1976. Gender in a terminal Gaelic dialect. Scottish Gaelic Studies 12.279–82.
———. 1977a. A hierarchy of morphophonemic decay in Scottish Gaelic language
death: the differential failure of lenition. Celtic Linguistics – 1976 (Word, vol. 28),
96–109.
———. 1977b. The problem of the semi-speaker in language death. International
Journal of the Sociology of Language 12.23–32.
Dressler, Wolfgang. 1972. On the phonology of language death. Papers from the Eighth
Regional Meeting, Chicago Linguistic Society, 448–57.
———. MS. Wortbildung bei Sprachverfall.
Hamp, Eric P.; Fred W. Householder; and Robert Austerlitz (eds.) 1966. Readings in lin-
guistics II. Chicago: University of Chicago Press.
Hill, Jane H. 1973. Subordinate clause density and language function. You take the high
node and I’ll take the low node: Papers from the Comparative Syntax Festival, 33–52.
Chicago: CLS.
———, and Kenneth C. Hill. 1977. Language death and relexification in Tlaxcalan
Nahuatl. International Journal of the Sociology of Language 12.55–70.
Hymes, Dell (ed.) 1971. Pidginization and creolization of languages. Cambridge:
Cambridge University Press.
Krauss, Michael E. 1963–70. Eyak texts. Photocopy.
Kurylowicz, Jerzy. 1949. La nature des procès dit ‘analogiques’. Acta Linguistica 5.121–38.
[Reprinted in Hamp et al., 158–74.]
Lambert, Wallace E. 1967. A social psychology of bilingualism. Journal of Social Issues
23.91–109.
Miller, Wick. 1971. The death of language, or Serendipity among the Shoshoni.
Anthropological Linguistics 13.114–20.
Mühlhäusler, Peter. 1974. Pidginization and simplification of language. (Pacific
Linguistics, B26.) Canberra: Australian National University.
Reinecke, John E. 1971. Tay Boi: notes on the pidgin French of Vietnam. In Hymes,
47–56.
Samarin, William J. 1971. Salient and substantive pidginization. In Hymes, 117–40.
1 Introduction
The kinds of loss, both general and particular, reported by now for obsolescent
speech forms are legion. Phonological, morphophonological, morphological,
syntactic, semantic, and lexical complexity or richness have all been shown
to undergo reduction as languages are used less and/or used by a dwindling
number of speakers. The reductions involved can range from outright disap-
pearance of forms, constructions, categories, and stylistically or semantically
related options, to a drop-off in the types or overall number available (or at
least in use).
Both of these outcomes can result, it seems, from a variety of circumstances.
The number of conversation partners with whom use would be appropriate
may decline to a prohibitive level; audiences for certain types of verbal perfor-
mance may vanish. The structure of the language itself may make the merger
of two previously parallel sets of forms or syntactic constructions likely. The
relative difficulty or rarity of certain forms, constructions, lexical differentia-
tions, and so forth, can lead to partial or total disuse (with or without some
residual fossilization). The existence of similar structures or items in both
superordinate and subordinate language can lead to the retention of just one
of several original options, via convergence, in the subordinate language. In
general it appears difficult to maintain marked complexity such as elaborate
oratorical style, rich allomorphic variety, or minutely differentiated lexical
fields under circumstances of prolonged and intense language contact, partic-
ularly if one of the languages enjoys significantly greater prestige and official
support. It may be difficult even in relative geographical or social isolation if
the economic vitality of a community falls off too sharply or if the number of
speakers falls too low (the two developments perhaps occurring in tandem).
2 Scottish Gaelic
It should be noted at the outset that the fully fluent speakers in this study, no
matter what their numbers in a particular village and no matter how frequent
their opportunities actually to speak Gaelic on a regular daily basis in later
life, had all been raised in households where Gaelic was the first language and
in a social setting where segregation of ESG speakers (residentially, occupa-
tionally, in social intercourse, and in marriage) was still largely the norm. They
had learned Gaelic first, as children, spoke it either exclusively or much better
than English when they began school, continued to use it habitually in their
parents’ homes, and in many cases could still be considered Gaelic-dominant
when included in my study if the criterion for dominance was taken to be pref-
erence for Gaelic under conditions of emotional or physical stress.
Use of Gaelic is above all interlocutor-governed in East Sutherland; certain
individuals trigger the use of Gaelic in certain other individuals on a habitual
basis, whereas matters such as topic or setting, though sometimes relevant,
are unimportant by comparison. One fairly effective measure of Gaelic domi-
nance was the degree to which a fluent speaker used Gaelic with me or in my
presence under conditions of excitement, fatigue, anger, worry, and the like. I
was a speaker, but not ideally fluent, nor could I be counted on to be a perfect
decoder, most especially under emotional or physical circumstances such as
when the native speaker’s tempo picked up, the noise level rose, or the num-
ber of disfluencies sharply increased. If a fluent speaker persisted in Gaelic,
or broke into Gaelic, under such circumstances when I was the interlocutor
or among just a very few interlocutors, the likelihood was very great that that
speaker was simply more comfortable in Gaelic to a degree where a less-than-
ideal conversation partner could not inhibit the preference for use of Gaelic. It
is on the basis of participant-observation over a fifteen-year period (1963–1978)
during which there were a good many such occasions that I venture the opin-
ion that many of the fluent speakers with whom I worked should be consid-
ered Gaelic-dominant, including – very conspicuously – the youngest of them.
A smaller number of individuals could best be considered truly balanced bilin-
guals, almost entirely responsive to the interlocutor factor (or that factor com-
bined with topic and setting factors) regardless of any stress factor. And a few
(including one of the oldest – probably, judging by his remarks, by reason of
emotional reaction to the local bias against the Gaelic-speaking fisherfolk) had
become English-dominant. By contrast, two or three elderly people were so
obviously Gaelic-dominant (more fluent and better able to express themselves
in Gaelic) that no close observation of their behavior under stress conditions
was necessary for assessment of preference or relative abilities.
Since I will be pointing to weaknesses in the Gaelic of the fully fluent here,
it seems important to establish that despite their ever diminishing numbers
and steady movement into the upper reaches of the population pyramid in
East Sutherland, these were speakers of high verbal skill who used their native
language for a quite normal range of purposes. They could narrate, argue, joke,
gossip, tease, discuss health, community, national affairs, and business mat-
ters; they were fully able to exchange news, advice, plans, and opinions on
matters grave and trivial. All could quote Biblical passages and psalms; some –
with very few exceptions, males – controlled the relatively formal language of
prayer. All knew some proverbial lore and children’s rhymes. A relatively small
proportion (again mostly males) could read familiar Biblical Gaelic passages;
a few could read nonreligious material in Gaelic. Very few could write Gaelic
with any accuracy or freedom as to subject matter – probably fewer than half
a dozen. Because of the ease with which English loanwords are taken into ESG
and the remarkable degree to which they are adapted to Gaelic grammar and
morphophonology, even the most up-to-date and technical subjects could
be discussed in Gaelic, and were. Extremes of code-switching were very rare
for this reason, except that certain particular people were notorious for it as
part of their personal style and were strongly condemned for it; such individu-
als were excluded from the sample of the fully fluent in this study, since the
community did not in general regard them as fully fluent in speech behavior,
whether or not they might have been capable of full Gaelic fluency by dint of
greater effort and self-discipline.
One peculiarity of the fully fluent sample merits special mention: On the
whole, the younger half of the fluent sample had more opportunity to use
Gaelic regularly, at the time of the study, than did the older half. This reflects
the simple fact that the younger individuals were more likely to have living
spouses, siblings, and schoolmates available as regular conversation partners,
whereas the older individuals were likely to have lost some of their regular
Gaelic conversation partners to the higher mortality natural for their age
group. There were certain noticeable consequences of this phenomenon,
most particularly in lexical recall. The more elderly sometimes had more dif-
ficulty retrieving vocabulary items, especially in isolation and on demand,
than their younger counterparts, even when the words in question were in fact
well known to them and could be heard in their spontaneous speech. No such
age or isolation factor appeared in grammar, however. The older fully fluent
speakers were for the most part better able to supply conservative structures,
although in this connection personality factors came into play: some individu-
als among the ranks of the fully fluent, older or younger, could be pressed for
the most conservative or elaborate structures in their repertoire more success-
fully than others (that is, without becoming confused or hesitant and with rel-
ish for the challenge to their abilities).
Most of the structures which proved difficult for fully fluent speakers were syn-
tactically somewhat complex and had probably fallen out of common use as
the language receded. One exception was a deficiency in morphological struc-
ture, namely uncertainty about the formation of ordinal numbers as opposed
to cardinal numbers. Gaelic numbers are notoriously difficult compared with
1 There are some extra complications not readily discernible even in an element-for-element
translation. For just one example, 2, 20, and all multiples of 20 are normally followed by the
singular of any noun, while all other numbers take the plural of any accompanying noun. But
nouns which occur with very high frequency after numbers sometimes appear in the singular
after 3, 4, 5, and 10 in ESG: “year” and “time, occurrence” are two such examples.
there are too many elements in such a numeral (“3,” “20,” “8,” and “-teen”), all of
which present themselves at least theoretically as candidates for the obligatory
ordinal-forming /-u/ to individuals who simply do not make use of such high
ordinals – and perhaps never did.
cardinals does. And this despite the fact that context makes the intended
meaning quite clear in both cases.
Most of the structures which proved problematic for fully fluent speakers
involved syntax rather than morphology. Although it seems very likely that the
constraints of a translation task increased the difficulty of the structures, it
has to be kept in mind that all of these speakers were being asked to produce
many other complex structures via translation tasks as well; yet only the ones
discussed here were notable for the rather general and self-conscious difficulty
they caused speakers of quite considerable skill.
Two out of the four problematic syntactic structures shared the feature that
they used essentially the same construction twice in immediate succession.
There seemed to be something inherently confusing about such structures,
since one of the two which proved troublesome is relatively short and simple,
while the other is relatively long and involved. I never heard either one used
in spontaneous speech. Rather, I realized that the structures would exist (in
one case because it was so logically obvious, in the other because a colleague
asked me how something would be expressed in Gaelic and it dawned on me
what the answer would have to be, at least in a conservative form of ESG), and
they would come out a certain way in ESG, given the grammar of the dialect. I
then set out to see whether they could be elicited. I should note, again, that I
went after a few other structures which did not crop up in spontaneous speech
in the same fashion and got responses without hesitation and without faulty
grammar or distorted meaning.2
The simpler of these two problematic structures called for two verb com-
plements in a row, each a gerund. Given that people could build sentences
of the type “I’m trying to hear him” and also of the type “I’m making him go,”
it stood to reason that they should also be able to construct sentences which
combined the two types, to produce sentences such as “I’m trying to make
him go.” Yet when I asked for (among others) the sentence “I’m trying to make
him hear me,” entirely fluent speakers produced defective translations with
various missing elements. The defective productions typically dropped one
2 An example would be “You ought to have a dog,” a nonobvious structure in ESG because there
is no lexical verb “to have” to serve as the complement of “ought.” The translation turned out
to be, roughly, “It’s incumbent on the dog to be at you,” and no one had the slightest trouble
producing it on demand.
complement or another, to give, for example, “I’m making him hear me,” or,
more oddly, “I’m trying him to hear me.” One resourceful fluent speaker made
an end-run around the difficulty by phrasing the sentence altogether differ-
ently: “I’m trying to make that he’ll hear me”; but the structure of the expres-
sion “to make, cause to happen” was itself slightly defective in this version (it
lacked a prepositional element). Another sentence of the same general type
(“Are you trying to make me leave?”) gave rise to similar difficulties; excep-
tional was the similarly structured sentence “He’s trying to make her put a fire
on,” which four out of four speakers asked got right on the first try. By contrast,
four out of five speakers gave defective versions of “I’m trying to make him hear
me,” and four out of eight gave defective versions of “Are you trying to make me
leave?” Two of the four who gave defective versions of the former were tested
on two different occasions separated by at least two years, with the same sen-
tence presented on each occasion; one got it right on one occasion and right
on the second try on the other occasion, while the other speaker never arrived
at a correct version despite many tries on one occasion but produced a correct
version – barring a missing initial consonant change required – on the third
try, on the other occasion.
In my subsequent efforts to fathom the curious difficulty caused by this
double verb-complement structure, I found that passive ability to handle it
was much better than active ability. On a different field trip from the one dur-
ing which I did most of the investigation of active control of this structure, I
presented the sentence “I’m trying to make him hear me” in ESG to eight fluent
speakers and asked for an English translation. Six of the eight produced accu-
rate English translations instantly and easily.3
In view of the difficulties with active control of the double verb-comple-
ment structure, it was not surprising to find that fluent speakers also had
trouble with prepositional possessive relatives of the type “The woman in
whose house it happened died.” Such sentences would require, in conserva-
tive ESG, two identically structured prepositional relative clauses in immedi-
ate succession: literally, “Died the woman at (-whom) was the house in (-which)
happened it” (with the italicized portions representing exactly parallel struc-
tures). Because the verbs in this sort of sentence have to be inflected for
tense, involving not only the infamous initial consonant mutations of Celtic
3 One YFS chose an alternative meaning, namely “show,” for the verb which also translates as
“try,” and then ignored the expression “to make” in favor of a continuation which made sense
with “show”: “I’m showing him how to hear me.” One OFS used “try” in her translation, but
weakened the causative sense of the remainder: “I’m trying for him to hear me.”
(as was also true of the double verb-complement structure) but also particles
and dependent as opposed to independent verb forms, they can be consid-
ered difficult as compared to the double verb-complement sentences. In the
latter, the verb complement is always the gerund, changeless except for the
initial consonant mutation if that consonant is one susceptible to the muta-
tion (always lenition in this construction, a mutation which generally replaces
stops with spirants and spirants with other spirants or zero: /khɫĩ:nčən/ “hear-
ing,” /fiaxən . . . xɫĩ:nčən/ “making . . . hear,” /fa:gal/ “leaving,” /fiaxən . . . a:gal/
“making . . . leave”).
Fluent speakers asked to produce a Gaelic translation of the English sen-
tence “The woman in whose house it happened died” vastly preferred to avoid
the double prepositional structure. Two anomalously began the sentence with
nouns (Gaelic is a Verb-Subject-Object language); they then tried to repair,
producing respectively “The house in which it happened, the woman died” and
“The woman in the house it happened, she died.” With each of these speakers
(both very cooperative and resilient sources, luckily), I asked them to try again,
beginning the sentence with “died,” as would in fact be normal for Gaelic.
The elder of the two then got the structure right; the younger produced “The
woman in the house it happened died.” After several more tries, each close
but slightly off, I said the conservative Gaelic structure myself. He responded,
“That’s right enough.” Mindful of Labov’s experience with nonstandard English
speakers who proved unable to repeat a formulation outside the patterns of
their own dialect even when well motivated (Labov, 1979, pp. 332–334), I then
asked him to repeat the version I had said. He did so with no difficulty what-
ever. Another YFS who thoroughly liked challenges was posed the problem;
he rephrased, but produced a good ESG sentence, namely “The woman who
belonged to the house where it happened died.” Pressed for another version, he
gave “The woman at (-whom) was the house where it happened died.” Asked
to keep this general structure but work in the preposition “in” after the noun
“house,” this man then produced the double prepositional (possessive) relative
on the third try.
A quite different sort of construction which struck me as possibly trouble-
some for the fully fluent even before I began to investigate it closely seemed
predictably difficult both because the conjunction involved was of low fre-
quency even in the positive and also because the negative clauses introduced
by that conjunction became semantically very close to a much higher fre-
quency construction. Thus, a shift to the higher frequency alternative seemed
likely. This was in fact what happened.
Most fluent speakers were familiar enough with the conjunctival phrase
/nas ɫə: na/ “unless,” but even in the positive they were inclined to use “if . . . not”
in its stead.4 Asked to render “He’ll do it unless he’s too tired,” three out of
four speakers initially produced “‘He’ll do it if he isn’t too tired’”; one shifted to
/nas ɫə: na/ when pressed to translate the word “unless” more literally. Asked
for the Gaelic of “I won’t go unless he goes, too,” four out of seven fluent speak-
ers produced the /nas ɫə: na/ construction, but the other three, all OFSs, used
“if . . . not.” One of the three who used “if . . . not” was pressed for “unless” but
was unable to produce it on this occasion, although she did use it in other scat-
tered translation-test sentences on a variety of occasions.
Presented with a Gaelic sentence which used /nas ɫə: na/, eight fluent speak-
ers (out of eight asked) translated it unhesitatingly, uniformly rendering the
conjunction as ‘unless’ in English.
In the negative, when Gaelic-to-English translations were requested, the
switch-over to a form of “if” was more prevalent than in the positive, as pre-
dicted. For example, eight out of eight fluent speakers asked rendered /nas ɫə:
na/ as “unless” in their English translation of the Gaelic sentence “He won’t
sell this house unless he gets another house,” but only four out of seven fluent
speakers rendered /nas ɫə: nax/ as “unless” in the sentence “He won’t get [=
inherit] the shop unless he doesn’t marry her.” In this latter sentence, three flu-
ent speakers used “if” in their translations, all incorrectly, to give “if he doesn’t
marry her.” Among the four who preserved “unless” in their translations, three
also made the “unless” clause positive, reversing the sense of the original. One
caught his own mistake and then supplied a (correct) negative version of the
English “unless” clause, but three others carried on with the positive. One of
them, a YFS, spontaneously translated her English version back into Gaelic,
and in doing so she also made the Gaelic version of the “unless” clause positive,
that is, the reverse of what had been given her to translate.
Semantically either “. . . unless he marries her” or “. . . unless he doesn’t
marry her” is a possible sequel to “He’ll inherit the shop . . .”; this probably
made the translation task more difficult. In the Gaelic-to-English translation
of another sentence, “I’ll see him tomorrow unless he doesn’t come,” shifting
to the positive in the “unless” clause would make no sense, and indeed no one
did that in this instance. Three fluent speakers offered English translations; all
used “unless” without hesitation, and all gave negative versions of the second
clause. (One of the three was a YFS who had made the negative “unless” clause
4 The form of this conjunctival phrase varies slightly across the three ESG-speaking villages. It
is given here unvaryingly in the Embo form simply for consistency’s sake. But when Gaelic
sentences were presented for translation into English I always altered the form, both positive
and negative, to suit the native village of the person to whom the sentences were presented.
we would have been killed,” and “If the rope broke, we would be killed.” In the
first English sentence it is now certain that the rope did not in fact break, and
we were consequently not killed, though we might have been. In the second
English sentence our fate is still undetermined – we may yet be killed, though
it is only speculation rather than prediction (whereas “If the rope breaks, we
will be killed” is predictive). Earlier translation-task experience had indicated
that the direct contrast between the two English sentences was necessary if I
was to get any Gaelic responses at all which reflected the fact that the condi-
tion was past and no longer pertained in one instance but not in the other. The
distinction which appears pluperfect in form in English can be made in ESG,
but it involves the use of a passive state-of-being (the “passive” being more
nominal than verbal in ESG in any case; see Dorian, 1973, for discussion); liter-
ally “If the rope were at/to its breaking . . .,” (that is, “broken”) as opposed to “If
the rope would break. . . .”
Despite the direct contrast in the later testing there were some failures. One
YFS, asked first for the “pluperfect” contrary-to-fact and then for the regular
conditional, gave the conditional in both cases. Realizing that his responses
were identical, he pondered a moment and then said, “It’s exactly the same.
Funny, that, isn’t it?” He proved unable to produce a “pluperfect” sort of
construction.
Another YFS gave a “pluperfect” for the regular conditional, an extremely
unusual mistake, then gave the regular conditional in the “if”-clause of what
should have been the contrary-to-fact with “pluperfect” in her effort to differ-
entiate. In confusion at that point, she got the result clause wrong – she gave
it as a simple past tense rather than any sort of conditional, although she was
an excellent speaker who normally had no trouble whatever with conditionals
as such. A Golspie OFS gave identical renderings for both English sentences
until I objected and pressed strenuously for a “pluperfect,” at which point she
did produce one. One Embo OFS and one Golspie OFS got the contrary-to-
fact completed condition right on the first try, given the direct contrast in the
English stimulus sentences.
On a second immediate-succession-contrast test, the same YFS who had
remarked that his first two sentences were “exactly the same” also gave identi-
cal uncompleted-condition sentences in this case: He produced “If he came,
we would leave” in Gaelic as a translation both for that sentence in English
and also for “If he had come, we would have left.” The other two speakers who
made mistakes in the first test (reversal in the one case, semantic collapse in
the second) now got the completed condition right in the second pair of test
sentences. Evidently one round of direct-contrast testing was enough to allow
them to straighten the confusion out and handle the distinction. But once
again, both the presence of failed attempts scattered throughout my field note-
books over the years and also marginal notations in those notebooks about the
difficulty of eliciting the completed contrary-to-fact condition indicate that
the contrast between a possible condition which still may apply and a possible
condition which was not fulfilled is maintained poorly and is recognized as
confusing and seen as difficult, even by fully fluent speakers.
Apart from these four syntactic environments and the one morphological
one described first, other notable grammatical difficulties manifested by fully
fluent speakers were either idiosyncratic (peculiar to a given speaker or to
one family) or clearly associated with a lexical item or a grammatical category
which had passed out of common use across the board. There were certainly
other deviations from the most conservative grammatical norm (the merger
of the two passive constructions is a case in point; see Dorian, 1973), but the
speakers were still able to express their meaning clearly and to differentiate the
structure concerned from other structures. They showed no recurrent confu-
sion or hesitation about the utterances they offered in such cases.
To a researcher who spent a good deal of time over more than a dozen years
deliberately thinking up posers for a group of speakers of a language clearly
fading from existence, with opportunities for use diminishing along with the
number of available habitual conversation partners, it seems remarkable that
there are so few constructions as this which cause self-conscious difficulty. The
facts that the population was largely illiterate or at best literate only in a very
limited fashion in their mother tongue, and that they got little or no positive
reinforcement for their skills in Gaelic until I appeared on the scene and began
to express a linguist’s admiration and appreciation for their abilities,5 likewise
make the relatively small number of “troublesome” constructions remarkable.
As to the nature of the constructions which caused the self-conscious dif-
ficulties and the reasons for its being these constructions rather than others,
a number of possible factors seem to be involved. Only in one case does it
seem likely that English has in any direct sense driven out Gaelic (although
increased use of English and decreasing use of Gaelic is surely a general fac-
tor in all decline and decay of Gaelic in East Sutherland): As noted earlier, the
formation of higher numerals is notoriously complex in Gaelic, and all over
the Highlands native speakers of Gaelic show a tendency to substitute English
5 My opinions were contrary to the norm: my informants were looked down upon within
the local setting for being Gaelic speakers.
for high Gaelic numerals, both cardinals and ordinals. Also as noted earlier, the
general absence of a “pluperfect” in both Gaelic and East Sutherland English
makes difficulties in contrasting anterior and present time in conditional sen-
tences understandable, since real need for that sort of distinction would be
felt relatively rarely; and the semantic substitutability of “if” and “if . . . not” for
“unless . . . not” and “unless” makes use of the higher frequency conjunction
“if” an attractive alternative in that case. If anything, the retention of “unless”
by any speakers at all becomes a bit hard to explain. It is a low-frequency con-
junctival phrase and not strictly necessary. The most likely explanation for
retention is that /nas ɫə:/ is the comparative of “little” and occurs fairly fre-
quently in the meaning “less,” thus keeping the general expression /nas ɫə:/ in
common use. The connection of less with “unless” in English translation may
be relatively salient for some individuals.
As for the prepositional possessive relative, its complexity contrasts sharply
with the extreme simplicity of subject and object relatives in Gaelic. These
latter use a relative particle undifferentiated as to case, number, or gender
followed by a verb phrase identical with that of a main clause; even a good
number of SSs succeed in forming them. By contrast just the prepositional
relative (let alone the prepositional possessive relative) is a complicated struc-
ture. In its most conservative form it calls for a prepositional structure with a
verb form as object, and the verb (or its accompanying particle) must show
the initial consonant mutation known as nasalization, which typically voices
voiceless consonants and replaces spirants with stops. Furthermore, the verb
or verb phrase must appear in the dependent rather than the independent
form. Since the prepositional relative phrase requires two of these relatively
complex structures in a row, it is not unexpected to find that fluent speakers
have trouble with them, even speakers who can form the single prepositional
relative satisfactorily. It is also the case that there are two possible forms of
the purely prepositional relative (e.g., “That’s the broken chair that I was sit-
ting on”) in ESG, one of which postpones the preposition until the end of the
clause, probably under the influence of English, and permits the substitution
of a simple relative without mutation and without dependent verb form for
the more complicated structure called for by a nonpostponed preposition.
This alternative is not available in the prepositional possessive relative, with
its doubled-up structure; but some speakers may nonetheless be struggling
mentally to locate such an alternative, since it’s common enough (preferred,
even, by some speakers) for the prepositional relative as such.
The sole construction troublesome for fully fluent speakers which seems
particularly unexpected is the double verb-complement structure. It is not
really very difficult; it is both logical and relatively short. Only the end-to-end
linking of almost identical structures makes it confusing, and it was strictly on
that basis that I foresaw the difficulties fluent speakers might have with it and
began to ask for translations of sentences which would require it.
What about the possibly surprising strengths of the Gaelic used by semi-
speakers (SSs)? In a sense, of course, all of their abilities are surprising. I will
take most of their capacity for granted, all the same, and discuss here only
a few of the more unusual aspects of their use of Gaelic. For the SSs, such
unusual characteristics are usually individual rather than group-wide, since
their Gaelic is much less stable and much more idiosyncratic than that of flu-
ent speakers. Their acquisition histories are not uniform: some began as flu-
ent child speakers and lost capacity; some were always imperfect speakers;
one or two were passive but not active childhood bilinguals who somewhat
later made an effort to develop some active skills. Furthermore, some have had
much more sustained interaction with fluent speakers than others: a few have
lived all their lives in households where fluent speakers used Gaelic regularly
within their hearing, while most others got the bulk of their exposure and
active experience from elderly kinfolk who survived for varying periods of the
SSs’ lives. Individuals within the latter (and larger) group have had quite vari-
able opportunities to hear and use Gaelic since the death of the crucial older
person(s) in their kinship networks.
All of the SSs have extraordinary receptive skills in Gaelic – an ability to
decode messages which is dramatically out of line with their ability to encode
messages. However defective their phonology, morphophonology, morphology,
and syntax, they have the stunning ability of the native speaker to understand
virtually everything said in their presence, regardless of noise level, speed, or
faulty articulation. This is the more striking because it contrasts with my own
skilled-learner shortcomings in these circumstances, despite the fact that my
Gaelic is more fluent and more grammatical than theirs. Their decoding skills
are not merely contextual, furthermore, since they can provide actual transla-
tions if asked or if they see that someone is not following.6
When it comes to their active skills, SSs have to make do with very much less
than the resources fluent speakers have available if they want to join in Gaelic
interactions. They can achieve a surprising amount of success by sticking
largely to fixed phrases and very high-frequency vocabulary and grammatical
6 Confirmation of the exceptional decoding skills of very imperfect speakers (and even one
completely passive bilingual) is provided by comprehension tests carried out by Schmidt
(1985) among the youngest users of Dyirbal in North Queensland, Australia.
8 Conjunctions
Despite the fact that conjunctions are a closed class, and a relatively small one,
they represent a trouble spot for SSs. There are a number of reasons for this.
Most conjunctions require choices about mutation or nonmutation of the ini-
tial consonant in the main verb of the clause which they introduce or in a
particle which precedes it; this is sometimes the same as and sometimes in
addition to a choice between a dependent and an independent form of the
verb. Some of the conjunctions are of low frequency and simply do not make
their way into the SSs’ repertoires; a few conjunctions have several different
forms depending on what tense they appear with, or, especially, on whether
the clause is negated or not.7
7 Negation is a special problem because a few conjunctions (and also some particles) have
a negative counterpart which subsumes the negation, so that no other negating element
appears in the clause; the negative form may or may not resemble the positive form in sound.
The test sentences were not actually designed to explore the control of con-
junctions, since conjunctions overall seemed a more complex matter than
most SSs could be expected to handle. Nonetheless a number of the test sen-
tences did contain conjunctions in the natural course of things. Although the
coverage is far less systematic than would have been the case if conjunctions
had been a real focus of the tests, the number and range of such sentences offer
an interesting glimpse of SS ability to deal with relatively challenging material.
Where a SS simply did not know the conjunction in question, several strat-
egies appeared. Rarest by far was borrowing from English. One of the two
strongest SSs adopted /khəs/ (from English because) in the sentence “He didn’t
come because he’s too tired,” but the rest of those asked (none of whom were
familiar with fluent-speaker /wəl/ “because”) generally omitted the conjunc-
tion and left the causal connection to be inferred from juxtaposition of “He
didn’t come” and “He’s too tired.”
It should be noted that “because” is not the commonly used conjunction in
Scottish Gaelic that it is in English; although ESG use of /wəl/ is moderately
high, the form is a local one and many western dialects seem to prefer causal
inference of the type demonstrated by the ESG SSs or else substitution of “and”
(“He went home early and him tired with working in the field”).
Substitution of a more common conjunction, where the general mean-
ing permitted, was a useful strategy among SSs for making the complex more
expressable. A strong SS, who knew and used the conjunction “when” on occa-
sion, nonetheless circumvented it in two cases, once via a shift to “and”: “She
lost her life when she was only five” was rendered as “She lost her life and she
wasn’t but five.” A very weak SS made a similar substitution of “but” for a con-
junctival construction she didn’t know at all: “He understands everything, but
can’t speak,” for “Although he can’t speak, he understands everything.” One
of the intermediate-level SSs made the same sort of substitution of “but” for
“although . . . not” in the same sentence. Omission, with juxtaposition carrying
the burden of semantic/syntactic relatedness, was the strategy employed by
the weakest of all the SSs in this same sentence; she rendered it as “He has no
speaking; he’s understanding everything.” She employed the same juxtaposi-
tional strategy for “if” in several cases, for example, “I won’t see him again; he’s
going away tomorrow” for “I won’t see him again if he goes away tomorrow,”
and “Don’t give (him) £5; you’ll lose your money” for “You’ll lose your money
if you give him £5.” The second of these juxtapositional substitutions for “if”
preserves the sense of the original; the first one does not.
One of the two strongest SSs and one on the intermediate level allowed “if”
to stand in for “unless,” which would have worked if they had used the nega-
tive form of the conjunction. But since they either did not control that form or
9 Merger
also unique, so that their use avoided the ambiguity of merger and allowed
more possibility of interpretation by context and repetition. Unhappily, these
unique forms were also nonce-forms for the corpus, and so the question of
whether they would be used consistently by the SSs in question cannot be
answered. Each of these women showed less merger than might have been
expected: Partial merger was more common, but the overlap was relatively
small, considering how minimal their overall skills were. One of them made
consistent use of /kə/ (normally “that”) to mean “when”; as if to compensate,
she adopted a strategy of deletion (7 out of 8 instances) for “that,” the only
exception being an occurrence with the very high frequency verb “to be.” Three
of the 8 instances of “that” in the test corpus occurred in clauses with the
verb “to be”; but if inconsistencies and exceptions are to appear, this is cer-
tainly likely to be the verb to produce them, since it is highly irregular and has
more than the standard number of distinct forms. So it is not entirely unex-
pected that this speaker produced her one syncretistic use of /kə/ “that” in that
particular environment. She used both /mə/ and /na/ variably for “if”; she also
extended /mə/ to mean “unless,” and, like the other SSs who used this strategy,
she failed to negate and produced a meaning opposite to that of the stimulus
sentence. This speaker made one further use of /mə/: She tried substituting it
for “until,” a conjunction she clearly did not control. She also clearly did not
like the result, since she immediately tried twice again, discarding /mə/, and
produced a unique (and incorrect) form /khən ə/ on each of the subsequent
attempts. Though wrong, this new effort is much closer to the correct /khəs
ə/; she preferred it, and in sticking with it she eliminated the partial merger of
“until” with “if” and “unless.”
Her cousin, the third of the very weak group of SSs (and the one who does
not spontaneously make active use of Gaelic), had some striking successes
with conjunctions. She got “when,” “until,” and – most surprisingly – the nega-
tive “that . . . not” form correct. She missed on “unless,” but did not merge it
with “if”; instead she produced a unique (incorrect) form /kən ə/. Her only
syncretistic use of conjunctions was a very unusual one. She never deleted
“that,” but she used /ə/ (or a variant of it) three times instead of her usual /kə/
(5 instances). This produced partial merger, because she also used /ə/ (or one
of its variants), as well as /mə/ and /na/, for “if.” Furthermore, /ə/ is the normal
positive form of the relative particle, and this speaker does make quite regu-
lar and correct use of that particle; so the syncretism becomes marked when
the relative is taken into consideration as well. (The relative does not need to
be considered for other speakers, since no conjunction takes the form /ə/ for
them.) All the same, there is no conjunction for which this SS used /ə/ (or one
of its variants) as her dominant form, much less her sole form; she does use it
10 Subtle Compensations
8 This SS lives in London, as she has most of her life. She spent her first six years in a grand
mother’s household in Embo, however, and Embo Gaelic was the habitual language of her
parents, to whose home in London she moved after age 6. The parents spoke only Gaelic to
each other and to nearly all adult visitors to the house, although not regularly to this elder
child until she came to insist on it. Thereafter, her mother in particular became a fairly regu
lar Gaelic conversation partner for her in the London environment.
This same weak but eager SS took a notion one summer to correspond with
me in Gaelic – an undertaking which only becomes extraordinary when the fact
that she is utterly illiterate in Gaelic emerges. Despite her lack of any notion of
how to write Gaelic, despite the facts that Gaelic has quite a lot of phonemes
not present in English and that the English orthography therefore doesn’t
provide any obvious method of rendering them, and despite the fact that this
woman’s Gaelic is defective grammatically and morphophonologically, there
were only two words (in a total of 358 words of Gaelic across four letters) which
I was unable to decipher. She managed to devise writings good enough to make
her meanings clear, for the most part, with a bit of ingenuity also on my end
as decoder. In fact it was considerably easier for me to decode her messages
than to adopt her system of encoding so as to be able to write back; again to
her enormous credit, however, she was consistent and clever enough in her
renderings so that I could use her writing system to respond, even though I
had to suppress my own knowledge of Gaelic orthography (incomplete to be
sure – ESG is an unwritten dialect and I have never formally studied Gaelic at
any time) in order to do so. It is hard to think of more dramatic proof of how
much a very weak “speaker” of a dying dialect can do with very little than this
woman’s success in communicating by writing through a language which she
controlled poorly and in which she was illiterate.
11 Concluding Remarks
References
9 I have argued elsewhere (Dorian, 1982) that the abilities of SSs and their ready inclusion in
verbal interactions among fluent speakers warrant a redefinition of the speech community.
1 Introduction
features of the Scottish Gaelic spoken by bilingual fisherfolk and their descen-
dants in the village of Embo, on the east coast of the county of Sutherland in
Highland Scotland. All four grammatical features were showing some change
in progress across a 55-year age-and-proficiency continuum at the time when
I first began to work with this receding dialect in the early 1960s, and I have
tracked their ongoing use among a dwindling number of bilinguals ever since.1
While two of the four features had at least rough parallels in English grammar,
two did not, and the question raised here is whether the features with no paral-
lel in English grammar (unmatched structures hereafter) will show swifter and
more extensive loss in receding Gaelic than the features which have an English
parallel (matched structures hereafter).
The fishing villages of east-coast Sutherland came into being around the begin-
ning of the 19th century. At that time great landlords all over the Highlands,
pursuing a new source of cash income, summarily removed their Gaelic-
speaking tenants in order to create sheep farms. In East Sutherland the inland
evictees were resettled at the coast under conditions that required them to
take up fishing as a livelihood: too little land was allotted to make agricultural
self-support possible, hooks and lines were distributed, and in at least one case
Scots-speaking fisherfolk from the Moray coast well to the south were settled
among the evictees to model fishing skills (Dorian 1981: 29–37). During the
next century and a half the evictees’ descendants became skilled fisherfolk,
but their livelihood was a dangerous and precarious one offering only inter-
mittent financial reward. They had arrived at the coast as desperately poor
evictees, and their low-income livelihood guaranteed continuing poverty; as
a result the fisherfolk population suffered a considerable degree of social stig-
matization. Living in a few densely populated streets beside the sea, all follow-
ing a single occupation and marrying among themselves, the East Sutherland
fisherfolk retained their Gaelic speech even after most other occupational
groups in the region had made a gradual transition to English. In 1963 there
were about 105 adults in Embo village who spoke Gaelic, but there were no
1 I worked with Embo speakers on site between 1963 and 1978; after a decade-long hiatus
caused by health problems, I resumed work with surviving sources in 1991 by tape-recording
and long-distance telephone. (Phone conversations were recorded with the permission of
the other party.) In several cases during the late 1980s and the 1990s direct in-person work
was also made possible by visits paid me in the U.S.
longer any monolingual Gaelic speakers and there were also no children who
spoke the language. The population of Embo was roughly 250 at the time, and
active bilinguals constituted at least 40–50% of the adult population;2 among
the older people Gaelic was universal. For most of the bilinguals Gaelic had
been their first language, used in the family on a regular basis while they were
growing up and for the most part also in everyday village life; for at least a few
of the oldest remaining speakers Gaelic was their dominant language as well.
All Gaelic speakers were fluent in English. The youngest speakers who still
made use of Gaelic in daily life did so chiefly vertically, with older members
of their kin networks, rather than horizontally among themselves, and English
was the dominant language for these speakers.
East Sutherland Gaelic was a dialect of the extreme northeastern periph-
ery, sharply different in many respects from more mainstream dialects of the
western Highlands. The local dialect, not taught in the schools or used in writ-
ing, was unaffected by any standardization processes. Passive literacy in a stan-
dardized Biblical form of Gaelic was to be met with among a good many men
who had been trained to precent (line out for congregational singing) the met-
rical version of the psalms, but the differences between this archaic written
language and the actual spoken Gaelic of the village were too great to produce
any significant transfer of literacy or to provide any grammatical modeling for
local speakers.
Among the grammatical structures that were very clearly showing change in
progress in Embo Gaelic were locational adverbs and the vocative case, two
prominent elements of Gaelic grammar without a parallel in English. A brief
sketch of each structure follows.
Locational adverbs were originally paired with a set of directional adverbs.
Locational ‘out’ (‘He’s out’) was /(ə) mwĩ(ç)/, directional ‘out’ (‘He went out’)
was /(ə) max/, locational ‘up’ was /hurəd/, directional ‘up’ was /(ə)n ɔ:rd/, and
so forth. But in Embo Gaelic the distinction between locational and direc-
tional ‘in’ had been lost at some earlier point, and the distinction between the
2 Because census figures are entered by parish, no official population figures exist for Embo
village as such. In response to a request for a population estimate for Embo, the General
Register Office for Scotland estimated that the village had a population of 260 at the time of
the national census in 1971.
a. Those of my Embo sources aged 64 and above were regarded in the com-
munity as skilled senior speakers, notable both for the lexical richness
of their Gaelic and for their command of local knowledge (e.g. mastery
3 Expansion of the directional adverbs into additional environments was probably encouraged
by the fact that verb phrases involving any of the adverbs in question (e.g. ‘die out’, ‘tie up’,
‘shut down’) invariably used the directional form.
4 In a few cases there was in addition a change in the final consonant of a noun in the vocative
case, and in still fewer a vocative suffix appeared; these word-final features were very seldom
met with among younger speakers and were sporadic even among older speakers.
5 The fact that initial consonant mutation is itself unmatched as a grammatical device in
English does not appear to be a significant factor in the weakening of vocative marking in
Embo Gaelic. Lenition alone marks the independent past tense for most verbs as well as
the vocative case for most nouns, and no comparable weakening of independent past-tense
marking occurs across the age-and-proficiency continuum. Even the obligatory but non-
informational lenitions that traditionally appear after the numeral /ta:/ ‘two’ and the adverbs
/kle:/ ‘very’ and /rɔ/ ‘too’ are retained to a surprising extent among the youngest fluent speak-
ers and the less than fully fluent speakers (semi-speakers).
One striking difference in the degree to which Embo speakers used Gaelic reg-
ularly can be recognized across these groupings. Apart from the formerly flu-
ent individuals, all OFSs and YFSs had some conversation partners with whom
they used Gaelic as their routine medium of daily communication. This was
not true of any of the SSs, nor of any of the FFSs, partly because of a less com-
plete command of Gaelic (both groups, but especially the SSs), partly because
of their life circumstances (FFSs), and in the case of the SSs also because of the
degree of shift which the community had reached by the time of their birth
and childhood.
Table 1 presents the freely spoken and elicited data from speakers across
the age-and-proficiency continuum for traditional use of locational adverbial
forms and for non-traditional use of directional adverb forms to indicate loca-
tion. Table 2 presents the number (freely spoken and elicited data combined)
and percentage of traditional vs. non-traditional forms used by individual
speakers in environments traditionally calling for locational adverbs.
Table 4.1 Locational-adverb forms used traditionally in environments expressing location vs. directional-
adverb forms used innovatively in environments expressing location. No shading = OFSs, light
shading = YFSs, dark shading = SSs. Fr = freely spoken, el = elicited. The number entered in
brackets represents forms used solely in direct repetition of a non-local speaker and is not counted
in reckoning percentages in Table 2.
E3 [85] m. 2 1
E4 82 m. 9 15 28 1 4 1 1
E6 75 m. 3 1
E7 ?74 f. 1
E8 ?71 f. 2
E9 70 f. 2
E10 70 m. 1 1 1
E13 [67] f. 3 1 1 1 1 1
E14 65 f. 2
E15 65 m. 2
E17 64 f. 18 1 4 8 19 10 1 5
E20 58 f. 2
E22 58 f. 9 1 2 1 3 2 1 2 1
E23 57 m. 3
E24 57 f. 5 1 1
E26* 54 f. 16 2 17 16 4 6
E27 54 m. 10 10 2 3 16 3 2 2 3 1
E28 51 m. 1
E29* 50 f. 1 2 3 1 23 3 43 11 77 11 19 2
E30 49 m. 2 1 1
E32* 47 m. 5 1 3
E33 46 m. 1
E34 45 f. 1 1 4 5 2 2
E37 41 m. 1 2 1 6 10 1
E38 40 f. 6 2 6 4 3 2 1
E39 38 f. 2 2 2 3
E40 36 f. [3] 26 6 30 1 39 5 4
E41 31 f. 1
E42 30 f. 1 4 4 8 12 2
traditional non-traditional
locational forms directional forms
speaker # % # %
E3 [85] m. 3 100
E4 82 m. 57 96.6 2 3.4
E6 75 m. 4 100
E7 ?74 f. 1
E8 ?71 f. 2 100
E9 70 f. 2 100
E10 70 m. 3 100
E13 [67] f. 7 87.5 1 12.5
E14 65 f. 2 100
E15 65 m. 2 100
E17 64 f. 31 47 35 53
E20 58 f. 2 100
E22 58 f. 21 95.5 1 4.5
E23 57 m. 3 100
E24 57 f. 6 85.7 1 14.3
E26* 54 f. 61 100
E27 54 m. 48 92.3 4 7.7
E28 51 m. 1
E29* 50 f. 7 3.6 189 96.4
E30 49 m. 4 100
E32* 47 m. 6 66.7 3 33.3
E33 46 m. 1
E34 45 f. 2 13.3 13 86.2
E37 41 m. 3 14.3 18 85.7
E38 40 f. 8 33.3 16 66.7
E39 38 f. 4 44.4 5 55.6
E40 36 f. 111 100
E41 31 f. 1
E42 30 f. 1 3.2 30 96.8
Table 3 presents the freely spoken and elicited data from speakers across the
age-and-proficiency continuum for traditional vocative-case marking and for
absence of traditional vocative-case marking, giving the total number and per-
centage of such forms for each speaker.
Table 4.3 Number and percentage of instances of traditional vocative forms with initial-
mutation marking vs. number and percentage of instances of unmarked forms used
in direct address. No shading = OFSs, light shading = YFSs, dark shading = SSs.
Fr = free, el = elicited, tot = total. Percentages 90 and above boldfaced.
E4 82 m. 1 1
E6 75 m. 6 6 85.7 1 1 14.3
E7 ?74 f. 1 4 5 100
E9 70 f. 11 11 100
E10 70 m. 4 12 16 100
E13 [67] f. 16 2 18 100
E16 64 m. 1 3 4 100
E17 64 f. 14 14 28 100
E21 58 f. 2 7 9 90 1 1 10
E22 58 f. 13 7 20 95.2 1 1 4.8
E23 57 m. 20 20 100
E24 57 f. 4 1 5 100
E26* 54 f. 40 43 83 92.2 7 7 7.8
E27 54 m. 49 49 81.7 11 11 18.3
E29* 50 f. 7 5 12 100
E30 49 m. 2 2 100
E31 49 m. 3 3 100
E33 46 m. 1 1
E34 45 f. 7 9 16 66.7 5 3 8 33.3
E35 43 f. 2 2 66.7 1 1 33.3
E37 41 m. 1 2 3 23.1 4 6 10 76.9
E38 40 f. 7 7 35 2 11 13 65
E40 36 f. 9 12 21 19.8 57 28 85 80.2
E41 31 f. 3 3 30 7 7 70
E42 30 f. 11 11 100
For both the locational adverbs and the vocative case, it is clear that the tra-
ditional structures are best preserved by the older speakers and less well pre-
served by the youngest speakers (which of course is what marks them as cases
of change in progress), but on the whole vocative-case marking persists better
than does use of the traditional locational adverbs. OFS E17, fewer than half of
whose locational adverbial forms were traditional, produced traditional voca-
tive marking at 100%; FFS E26*, who showed no retention of locational adver-
bial forms at all, produced more than 90% traditionally marked vocatives.
Speakers as young as E34 and E35 retained 66.7% vocative marking, whereas in
the same age-range E34 and E37 used locational adverb forms in only 33.3% of
the environments that traditionally called for them. Ten of 23 speakers (43.5%)
used 100% traditionally marked vocative structures, while only nine of 25
(36%) used 100% traditional adverb forms in locational-adverb environments.
6 This development does not appear to reflect English influence. Negative imperative expres-
sions with ‘don’t go’ + gerund (e.g., ‘don’t go making trouble’), moderately common in
American English, are not in use in any form of English to be heard in eastern Highlands (or
anywhere else in Scotland, so far as I’m aware).
The traditional negated past tense of the irregular verb ‘to be’ differs in
Embo Gaelic from the negated past tense of all other verbs in not showing the
preterite particle /tə/ in the surface structure. In the negated past tense most
verbs have the structure /(x)a/ + /tə/ + Lverb root (negative particle + preterite
particle + lenited verb root); apart from ‘to be’, even the few that have supple-
tive past-tense forms still show the preterite particle. For ‘wasn’t/weren’t’ the
equivalent structure is /(x)a/ + /rɔ/ (negative particle + suppletive past-tense
form of ‘to be’); no preterite particle appears. Although this is the traditional
structure, the prevalence of the preterite particle /tə/ with all other verbs, and
even in certain other grammatical structures involving /rɔ/, has led for some
speakers to the appearance of /tə/ in the negated past tense of ‘to be’ as well.
Tables 4 and 5 present the data for traditional and non-traditional forms
of the negative imperative and of the negated past tense of the irregular verb
‘to be’.7
As before, better preservation of traditional structures by older speakers
than by younger speakers marks these as instances of change in progress. And
again, as in the case of the two unmatched structures, one of the grammatical
structures is retained in traditional form somewhat better than the other: 13 of
28 speakers, or 46.4%, used 100% traditional forms of the negated past tense of
‘to be’, as compared with eight of 21 (38.1%) who used 100% traditional Embo
forms of the negative imperative. Speakers as young as E37 and E38 used 100%
traditional negated past-tense forms of ‘to be’, whereas E31 was the youngest
speaker to use 100% traditional forms of the negative imperative.
Table 6 presents percentages for traditional versus non-traditional forms of
all four structures, unmatched and matched.
7 Negative imperatives using /a/ followed by an unlenitable consonant were discounted for
Table 4 purposes, since they could not be assigned either to /(n)a/ or to /(x)a/.
Table 4.4 Number and percentage of traditional negative imperative constructions with
/(n)a/ + unlenited gerund or /(x)a/ + lenited gerund vs. number and percentage of
non-traditional negative imperative constructions with /a čhe:ǰ/ + lenited gerund or
/a čhe:ǰ/ + lenited root. No shading = OFSs, light shading = YFSs, dark shading = SSs.
Fr = free, el = elicited, tot = total. Percentages 90 and above boldfaced.
E4 82 m. 2 2 100
E6 75 m. 2 3 5 83.3 1 1 16.7
E9 70 f. 1 4 5 100
E10 70 m. 4 8 2 9 23 100
E13 [67] f. 6 6 100
E17 64 f. 2 9 2 5 18 94.7 1 1 5.3
E21 58 f. 1 1
E22 58 f. 6 11 17 100
E23 57 m. 1 1 2 25 6 6 75
E24 57 f. 1 1 2 100
E26* 54 f. 3 2 6 11 22 78.6 1 5 6 21.4
E27 54 m. 9 8 17 100
E29* 50 f. 3 5 6 14 37.8 5 18 23 62.2
E30 49 m. 1 1
E31 49 m. 2 2 100
E33 46 m. 1 1
E34 45 f. 1 1 9 11 78.6 3 3 21.4
E35 43 f. 1 1 50 1 1 50
E37 41 m. 4 1 5 100
E38 40 f. 1 1 2 4 10.3 22 13 35 89.7
E39 38 f. 2 2 50 2 2 50
E40 36 f. 5 38 2 45 100
E41 31 f. 4 2 6 100
E42 30 f. 1 3 4 66.7 1 1 2 33.3
Table 4.5 Number and percentage of conservative (irregular) negative past tense instances
of the verb ‘to be’ without preterite particle vs. number and percentage of innovative
(regular) negative past tense instances with preterite particle. No shading = OFSs,
light shading = YFSs, dark shading = SSs. Fr = freely spoken, el = elicited, tot = total.
Percentages 90 and above boldfaced.
E3 [85] m. 3 3 100
E4 82 m. 49 6 55 75.3 18 18 24.7
E5 77 f. 1
E6 75 m. 2 2 25 2 4 6 75
E7 ?74 f. 4 4 100
E9 70 f. 2 2 100
E10 70 m. 5 10 15 100
E13 [67] f. 13 5 18 94.7 1 1 5.3
E14 65 f. 3 3 100
E15 65 m. 2 2 22.2 7 7 77.8
E17 64 f. 441 23 464 99.1 4 4 .9
E18* 60 f. 4 4 100
E20 58 f. 8 8 100
E21 58 f. 4 4 100
E22 58 f. 12 14 26 100
E23 57 m. 1 1
E24 57 f. 1 9 10 76.9 3 3 23.1
E25 56 m. 1 1
E26* 54 f. 88 36 124 93.9 7 1 8 6.1
E27 54 m. 88 28 116 83.6 18 5 23 16.5
E29* 50 f. 17 3 20 7.3 213 40 253 92.7
E30 50 m. 29 2 31 100
E31 48 m. 1 1
E32* 47 m. 12 12 100
E33 46 m. 1 1 20 4 4 80
E34 45 f. 8 4 12 24 24 14 38 76
E36 41 m. 1 1
E37 41 m. 1 11 12 100
E38 40 f. 22 65 87 100
E39 38 f. 8 14 22 88 4 4 12
E40 36 f. 88 79 167 100
E41 31 f. 3 3 100
E42 30 f. 127 17 144 100
Nancy Dorian - 978-90-04-26193-8
Downloaded from Brill.com02/23/2019 02:45:48PM
via King's College London
negative borrowing in a shift to the dominant language 127
Table 4.6 Percentages of traditional structures (left) and non-traditional structures (right) used by Embo
speakers across the age-and-proficiency continuum for two Gaelic grammatical constructions
without parallels in English grammar (columns 1, 2, 5, 6) and two Gaelic grammatical
constructions with English parallels (columns 3, 4, 7, 8). No shading = OFSs, light shading = YFSs,
dark shading = SSs. Percentages 90 and above boldfaced.
1 2 3 4 5 6 7 8
locatnl voc neg imper neg past locatnl voc neg imper neg past
adv ‘be’ w/o adv ‘be’ w.
part part
speaker % % % % % % % %
1 2 3 4 5 6 7 8
locatnl voc neg imper neg past locatnl voc neg imper neg past
adv ‘be’ w/o adv ‘be’ w.
part part
speaker % % % % % % % %
accomplished Gaelic speaker and a ready conversationalist. Among the SSs the
two who spoke most comfortably in Gaelic were E39 and E40, but the degree
to which they used traditional forms for these structures was sharply differ-
ent: E39 was the most traditional and E40 the least so. The community itself,
furthermore, had no experience of Gaelic standardization and was not in the
least puristic about these structures. With just one exception, neither endorse-
ment of the traditional forms nor disapproval of the non-traditional forms was
ever expressed by speech community members over the 40-year period of this
study. The single exception was a remarkable 1995 conversation with E29* dur-
ing which she commented extensively on the use of personal names in direct
address. But even though she was explicit (by way of examples of what one
would and wouldn’t say) about the need for lenited initial consonants in direct
address, she expressed this wonderingly, as a curious property on the part
of her native language, rather than as a “rule” which some people “violated”
or were “wrong” about. She did not so much as mention having heard other
speakers fail to produce the lenition she was endorsing.
All four of the grammatical structures under consideration here are com-
monplace, occurring at high frequency in ordinary conversational interac-
tion. However two of them, the negative imperative and the vocative, did not
occur frequently in my freely spoken corpus, even though that corpus accu-
8 There are a great many instances of my own name used in direct address in the corpus, but
because it begins with an unlenitable consonant this yielded no marked vocatives.
E3 (1) (1) E20 (1) (1) E26* (2) (2) E38 (2) (2)
100 100 100 100 46.1 86.3 34.2 55.2
E4 (1) (2) E21 (1) (1) E29* (2) (2) E39 (1) (2)
96.6 87.7 90 100 1.8 22.6 44.4 69
E6 (2) (2) E22 (2) (2) E32* (1) (1) E40 (2) (2)
92.9 54.2 95.4 100 66.7 100 9.9 0
E7 (1) (1) E23 (1) (1) E41 (1) (2)
100 100 100 25 30 0
E9 (2) (2) E24 (2) (2) E42 (2) (2)
100 100 92.9 88.5 1.6 33.4
E10 (2) (2) E27 (2) (2)
100 100 87 91.8
E13 (2) (2) E30 (2) (1)
93.8 97.4 100 100
E14 (1) (1) E31 (1) (1)
100 100 100 100
E15 (1) (1) E34 (2) (2)
100 22.2 40 51.3
E17 (2) (2) E35 (1) (1)
73.5 96.9 66.7 50
E37 (2) (2)
18.7 50
The ultimate interest in this comparison lies in how often the average percen-
tage of traditional expression of matched structures is higher than the ave-
rage percentage of traditional expression of unmatched structures, since some
such imbalance is what the notion of negative borrowing would predict. For
five OFSs and three YFSs there is no difference, because they preserved 100%
traditional expression of both kinds of structure, either in one or in both of the
instances representing the category. For two OFSs (E13 and E17), for five YFSs
(E21, E22, E27, E34, and E37), for all three FFSs, and for three of the five SSs (E38,
E39, and E42), the difference was in the predicted direction, with higher ave-
rage percentage of traditional expression for the matched structures. Contrary
to the prediction, however, the reverse was true for three OFSs (E4, E6, and
E15), for three YFSs (E23, E24, and E35), and for two SSs (E40 and E41): they
showed higher average percentages for traditional unmatched-structure
expression. For the two OFSs and one of the YFSs (E24), introduction of the
preterite particle into the negated past tense of ‘to be’ was the non-traditional
feature that caused the discrepancy in favor of unmatched-structure expres-
sion, since they preserved higher rates of all the other traditional structures.
For the other YFS (E35) better retention of traditional vocative marking than
of traditional negative imperative marking was the source of the discrepancy.
For both of the SSs use of some traditional vocative marking also produced
the discrepancy, since none of their other structures showed any traditional
forms at all.
In terms of the overall picture it appears that the formerly fluent speak-
ers are particularly susceptible to negative borrowing, since all three of them
showed considerably greater use of non-traditional forms in the unmatched
structures. This is a reasonable enough outcome, in view of the fact that they
spoke mainly English during most of their adult lives. OFSs appear more resis-
tant to negative borrowing than YFSs, with only two out of 10 OFSs showing
the predicted imbalance in favor of matched structures, whereas five out of
11 YFSs conformed to the prediction. This again is an unsurprising outcome,
given the greater role of Gaelic in the lives of OFSs, during whose childhoods
Gaelic was unrivaled as the dominant language of Embo village. The chief sur-
prise comes with the SSs, for two of whom the residual tenacity of traditional
vocative marking produces results contrary to the prediction entailed by the
negative borrowing concept.
6 General Discussion
The Embo village data offer only rather modest support for negative borrow-
ing as a language-contact phenomenon in obsolescence. This is noteworthy
in view of the fact that three of the changes in progress produce greater regu-
larity and therefore one sort of greater simplicity. Eliminating the distinctive
locational adverb forms would result in a single set of adverbs used both as
independent adverbial forms and as constituents in verb phrases (see note 3).
Introducing exclusive use of /a čhe:ǰ/ as an invariant dummy negative impera-
tive would eliminate a choice between two negative particles with different
effects, one producing no mutation and one requiring lenition. Introducing
the preterite particle into the negated past tense of ‘to be’ would eliminate the
distinction between that verb and all others in one very high-frequency envi-
ronment.9 As for the fourth case, eliminating vocative marking would reduce
by one the number of grammatical categories requiring special segmental
marking in the local Gaelic. Direct address would then, as in spoken English,
be signaled only by suprasegmental features.
The cases of the negative imperative and the vocative pose an interest-
ing challenge to the negative borrowing principle. In the (matched) nega-
tive imperative case, nine of 21 speakers (43%) used conservative forms at a
90–100% level, while in the (unmatched) vocative case, a higher proportion
(13 of 23 speakers, or 56.5%) used conservative structures at that level. At the
other end of the retention scale, six of 21 speakers (28.6%) used conservative
forms at a level below 50% in the (matched) negative imperative case, while
a smaller proportion (five of 23 speakers, or 21.7%) used conservative forms
of the (unmatched) vocative at that low level. That is, not only did a greater
proportion of the speakers perform at a very high conservative level in the
unmatched case, but a greater proportion performed at a very low conservative
level in the matched case. In each case, particles are central to the construc-
tion, and yet it is the unmatched structure – in which furthermore the particle
is not usually present in the surface structure – that survives better at the low
end of the age-and-proficiency continuum. Part of the explanation may lie in
the fact that only a single particle requiring lenition is involved in the conser-
vative vocative structure, while a choice between two particles, one producing
no mutation and the other requiring lenition, is involved in the conservative
negative-imperative structure. Non-structural factors may also be at work in
the unusually strong persistence of the vocative in this speech variety. Women
in East Sutherland use both endearments and their interlocutor’s given name
9 An alternative route to regularization is exemplified in this speaker sample by E38: she regu-
larized all the forms of ‘to be’ that involved /rɔ/ by eliminating the preterite particle /tə/
from every one of them, including from structures that routinely showed /tə/ for nearly all
Embo speakers, e.g. after conjunctions or particles that require the initial mutation known as
nasalization.
far more frequently during conversational exchanges than men do, and all of
the semi-speakers in the sample here are female. The relatively high frequency
of marked vocatives in the speech of older women with whom they frequently
interacted may have affected the retention rate among the youngest women in
the sample.
As Mougeon and Beniak (1991), Silva-Corvalán (1994), and Jones (2005) dis-
cuss, negative borrowing is not the only effect at work in cases of contact and
attrition, since internal system pressures are likely to play a role. The better
preservation of the vocative structure among the youngest female speakers
of Embo Gaelic suggests that gender-related speech style may have an unex-
pected effect.
While generalizations about the likely loss of unmatched features (e.g.
Andersen 1972: 97) have an inherent plausibility, based on the potentially
greater efficiency for the bilingual brain of working with matching structures,
it may be that to be effective a match must be structurally closer than is the
case in the Gaelic/English parallels looked at here. Perhaps, too, negative bor-
rowing is less characteristic of structural-change processes shared by the entire
community at the very end of an indigenous speech form’s existence than of
more individual-centered attrition processes such as first-language attrition
among immigrants and exiles or second-language attrition among successful
learners who later use the learned language infrequently. In whole-community
language-contact settings, perhaps it is long periods of regional co-existence
that are more likely to eliminate unshared grammatical structures, as in the
case of Sprachbund areas. In these cases a good deal of unmatched-feature loss
may already have taken place before any late-obsolescence stage. Some of the
differences between East Sutherland fisherfolk Gaelic and more mainstream
Gaelic dialects suggest that negative borrowing may have operated to this
effect over the relatively extended co-existence period prior to the more dra-
matically obsolescent phase which this study documents, so that by the 1960s
East Sutherland Gaelic was already structurally less different from English
than many other Gaelic dialects are, even though most local Gaelic speakers
were highly proficient at the time and some still Gaelic-dominant.
References
Dorian, Nancy C. 1981. Language death: The life cycle of a Scottish Gaelic dialect.
Philadelphia: University of Pennsylvania Press.
Fenyvesi, Anna. 1995. Language contact and language death in an immigrant language:
The case of Hungarian. University of Pittsburgh Working Papers in Linguistics 3, 1–117.
Jones, Mari C. 2005. Transfer and changing linguistic norms in Jersey Norman French.
Bilingualism: Language and Cognition 8(2), 159–175.
Mougeon, Raymond, and Beniak, Édouard. 1991. Linguistic consequences of language
contact and restriction: The case of French in Ontario, Canada. Oxford: Clarendon
Press.
Sasse, Hans-Jürgen. 1992. Language decay and contact-induced change: Similarities
and differences. In Matthias Brenzinger (ed.) Language death: Factual and theoreti-
cal explorations with special reference to East Africa (pp. 59–80). Berlin/New York:
Mouton de Gruyter.
Silva Corvalán, Carmen. 1994. Language contact and change: Spanish in Los Angeles.
Oxford: Clarendon Press.
Thomason, Sarah G. 2001. Language contact: An introduction. Washington D.C.:
Georgetown University Press.
their lives. But on the whole the assumption that the reduced use of a language
will lead also to a reduced form of that language seems realistic.
Haas’ assumption can best be tested in a terminal language community
where a continuum of proficiency is available, from full fluency to the barest
skills necessary for communication in the dying language. Such a continuum
is available in the coastal East Sutherland area of mainland Scotland where
I have worked for 11 years. In a total pool of Gaelic speakers which numbered
about 140 in 1972, there were at the upper end of the spectrum a few individu-
als who were more comfortable and proficient in Gaelic than in English,1 in
the middle range many who were skilled bilinguals, fluent in both languages,
and at the lower end some who could make themselves understood in imper-
fect Gaelic but were very much more at home in English. These last I have
called “semi-speakers” (Dorian 1973: 417). It is the identification of these semi-
speakers which constitutes a major problem for the fieldworker dealing with a
dying language, since he needs to know how representative and how reliable
his data are. If even the youngest fluent speakers showed notable grammati-
cal change in their Gaelic as compared with the oldest fluent speakers, then
the semi-speakers would presumably show still more radical departures from
the conservative norm, and data recovered from semi-speakers would need to
be handled with caution in the writing of grammars or in the reconstruction
techniques of historical linguistics.
One of the questions that is of interest to the investigator in the field is
whether the community’s own judgments of proficiency have any basis in
fact: is there any significant difference between the usage of the youngest of
those who have a reputation for fluency and that of the eldest of those whose
skills are little thought of? Can the community itself evaluate linguistically
to the extent of identifying the semi-speaker whose language is “reduced”?
Bloomfield’s experience with the Menomini (Bloomfield 1927) shows that such
judgments are made even in illiterate societies, but the question still remains
as to how finely and accurately they may be made. The smallest and most
strongly Gaelic speaking of the East Sutherland villages, Embo (population
ca. 275), provides an excellent test case. The oldest children of the skilled septu-
agenarian speaker B.R. are all considered fluent speakers, down to and includ-
ing the fourth, the 45-year-old son A.R. But the two daughters who are next
in line, J.R. and W.R., are considered less than fluent, even though the elder of
the two, J.R., is only one year younger than A.R.2 Since A.R. habitually speaks
Gaelic by preference whenever circumstances permit (that is, when he has a
1 There were no Gaelic monolinguals, however, and had been none for perhaps 40 or 50 years.
2 All three of these children are unmarried and live in the mother’s household.
3 This is fortuitous. I heard about possible male counterparts, but they were either unwilling or
unable to serve as informants during the limited time available for the study.
4 One of the fluent speakers in his 50’s is also an exile, but his wife is a Gaelic speaker, too, and
Gaelic is the normal language of the home. Their daughter, likewise living in exile, served as
one of the imperfect-speaker informants.
5 B.R. was the only fluent speaker to do this, but four of the semi-speakers other than J.R. also
did so at least once.
While I suspect that all of these phenomena are in fact characteristic of lan-
guages in decline, only one of them here seems to have significance for the
community judgment of proficiency. The first two were displayed not only by
the putative semi-speaker J.R., but also by her mother. If we plot the perfor-
mances of B.R., A.R., and J.R., we find the following:
It is only in the last case, the case of the analytic treatment of the first per-
son conditional, that we find B.R. and A.R. clearly lined up on the one side of
the statistics versus J.R. on the other. As it happens, the use of an analytical
construction for the first person singular conditional also constitutes a case of
analogical leveling, because all of the other persons of the conditional, both
singular and plural, are likewise formed analytically in East Sutherland Gaelic.
If we now look at the difference in performance among the three speakers
on the measure of analogical leveling as such, the pattern which emerges is
striking:
6 This sort of inference is risky, but combined with other evidence (halting manner of delivery;
surprising lexical gaps; and especially the presence of the missing categories in related dia-
lects or languages) may be helpful.
spot. The young Embo semi-speaker I.F., for example, has all but lost one entire
initial mutation, that is, one of the omnipresent morphophonemic changes
common to the languages of the Celtic language family. Were there no other
speakers, no other extant Celtic languages, and no written records, the only
evidence in I.F.’s speech that the morphophonological system of “nasalization”
had ever existed would be the occurrence of one seemingly irregular (but con-
sistent) alternation between the citation form [thε] “house” and its counter-
part [dε] after the definite article. Considering the tendency toward analogical
leveling which we have already noted in the performance of semi-speakers, the
utterly consistent appearance of such an apparent irregularity ought probably
to be given great weight in the analysis of a language where the only infor-
mants are probable semi-speakers.
Certainly the performance of all the semi-speakers in my sample indicates
the accuracy of Haas’ assumption that reduction in the use of a language
will be matched by reduction in its structure. Some of the kinds of reduc-
tion noted for East Sutherland Gaelic may prove to be universally character-
istic of dying languages when more evidence is in; I am thinking particularly
of loss of entire grammatical categories and of reduction of stylistic options.
The latter has already been documented by Hill for Luiseño and Cupeño (Hill
1973). Substitution of analytic for synthetic structures, on the other hand,
can of course only occur in languages with polymorphemic word structure.
Analogical leveling is again potentially universal, since it may be either mor-
phological or syntactic.
Conclusion
7 Hill, for example, reports that there seem to be no semi-speakers among the Luiseño and
Cupeño: “You either speak fairly well or not at all” (personal communication). In the Cupeño
case this may again reflect the speed of the total extinction process; the Luiseño approach to
extinction has been somewhat slower, but is still rapid compared to that of East Sutherland
Gaelic (Hill 1973: 34).
8 Krauss (1963–70), for example, commonly merely notes that the form is suspect (“inappropri-
ate here”, “distorted”, “confused towards end”, “inconsistent”, “confused”, all from p. 44) and
only rarely specifies the feature which provokes his comment (“ti.1-class-mark . . . missing”,
p. 116); very occasionally he provides the judgment of a second informant (“Lena rejects this
morpheme, or more probably, sporadic allomorphic variant . . .”, p. 45).
look for clues from the informants’ manner or from the coherence of the data
to provide the means for evaluating a corpus?
In sum, much more work needs to be done on the incidence of the semi-
speaker phenomenon, on the social and linguistic circumstances which give
rise to it, and on the linguistic features typical of semi-speaker performance.
References
Bloomfield, Leonard. 1927. Literate and illiterate speech, American Speech, Vol. 2:
432–39.
Dorian, Nancy C. 1973. Grammatical change in a dying dialect, Language, Vol. 49:
413–38.
Haas, Mary R. n.d. Tunica (Extract from Handbook of American Indian Languages,
Vol. IV). New York, J. J. Augustin.
———. 1968. The last words of Biloxi, International Journal of American Linguistics
34:77–84.
Hill, Jane H. 1973. Subordinate clause density and language function, You Take the High
Node and I’ll Take the Low Node (Papers from the Comparative Syntax Festival, ed. by
C. Corum, T.C. Smith-Stark, and A. Weiser). Chicago, Chicago Linguistics Society.
Krauss, Michael E. 1963–70. Eyak Texts. Photocopy.
Labov, William. 1970. The study of language in its social context, Studium Generale
23:30–87.
Swadesh, Morris. 1948. Sociologic notes on obsolescent languages, International
Journal of American Linguistics 14:226–35.
* The research reported in this paper was supported by grants from Bryn Mawr College (1974),
the American Philosophical Society (1976), and the National Science Foundation (1978)
(BNS 77-26295). I am indebted to Michael Silverstein for helpful comments on the first ver-
sion of this paper. The conclusions reached in this paper were presented in part at the 24th
Annual Conference of the International Linguistic Association, March 1977.
been discouraged, even suppressed, in Scottish education and public life for
some centuries (Campbell 1950). The economic base which supported the East
Sutherland Gaelic (ESG) linguistic community disappeared with the end of
the local fishing industry in the years between the two World Wars, and the
patterns of residential segregation and endogamy which had kept the Gaelic-
speaking fisherfolk apart began to weaken as a result. English, the language
of the social elite locally as well as nationally, has the support of virtually all
national institutions as they affect local life: law, education, government, the
military and (preponderantly) the media. Standard Scottish Gaelic has mini-
mal institutional support from one religious sect and from the media, the
local Gaelic dialect none at all. The contest between English, as a language of
wider currency with powerful governmental support, and ESG, as a language
of restricted currency with no institutional support of any kind, has been
unequal for generations. So long as the fisherfolk lived apart with an adequate
economic base, they were able to retain their language even while English
schooling and frequent commercial contacts with the English-speaking groups
allowed them to become fluent in English as well. With the loss of the fish-
ing and the separate marriage and friendship sub-population networks that
an independent economic base made possible, shift to English set in rapidly in
this last bastion of east coast Gaelic.
In the vicinity of Hamburg, in Berks County, Pennsylvania, the German dia-
lect which has been in use for over 200 years is likewise fading rapidly from
the scene. Negative prestige attaches to Pennsylvania “Dutch” in the area, as
to ESG in Sutherland, and the acceleration of urbanization, with the concomi-
tant weakening of a rural farming economy and its special social network,
has led to a shift away from the ancestral German dialect. The Hamburg area
“Dutch” are not religious separatists (that is, not Anabaptists of any persuasion,
whether Mennonite/Amish or Hutterite), and the move away from the farms
has exposed them broadly to the values of the larger American society with
its total commitment to English. There is minimal support for German (of the
standard variety, however) from the church, and latterly a few schools have
offered a little Pennsylvania Dutch in addition to regular courses in standard
German;1 off and on some area newspapers print material, usually humorous,
in the dialect. Beyond these slender efforts there is no institutional support for
the dialect.
Neither in eastern Sutherland nor in Berks County are the bilinguals
unaware of the language shift in progress. People comment readily on the
1 This is very recent. Policy for many years in this century was the active discouragement of
Pennsylvania Dutch in and by the schools (see Dorian 1978).
change within living memory in the ratio of speakers of the local language to
speakers of English. They are also acutely aware of the negative social prestige
of the local language and have many anecdotes to relate of social discrimina-
tion against local language speakers; the English of the bilinguals is colored, in
both regions, by the phonology of the home language, and this alone is suffi-
cient to make them conspicuous in a disadvantageous fashion. In each case the
focus of the social disfavor is probably primarily the charge of “backwardness”,
being behind the times socially and intellectually, a kind of bumpkin status;
in Sutherland but not in Berks County this stigma includes the reputation of
excessive inbreeding in the bilingual community. Low income relative to other
groups within the population, in days gone by, adds the notion of material
backwardness to that of social and intellectual backwardness, in both cases.
Although a great many speakers of ESG and Pennsylvania Dutch have moved
out of the relatively isolated settings which spawned these stereotypes, the
stigma tends to follow them.
Given the position of ESG and Pennsylvania Dutch in these two areas, it
seems extraordinary that anyone who had greater skills in English than in the
limited-currency language would nonetheless choose to maintain his role as
a speaker of that limited-currency language – especially when he was a pretty
imperfect speaker at that. Still, this is just what I have found to be the case.
The semi-speaker phenomenon is apparently not universally characteris-
tic of dying languages and dialects. Hill (1973), who worked with two dying
American Indian languages in California, found no semi-speakers: ‘You either
speak fairly well or not at all’ (personal communication). Semi-speakers are
common to ESG and Pennsylvania Dutch, however, and one can recognize
them in a number of accounts of other fading language communities (among
recent American Indian studies alone, for example, Krauss 1963–1970; Miller
1971; Salzmann 1969). Based on extensive interviewing of semi-speakers in east-
ern Sutherland and a lesser amount in Berks County, I have found three fac-
tors which operate to produce the social anomaly of the speaker who chooses
to use a low prestige language which he controls imperfectly. These are: cross
generational linguistic socialization outside the nuclear family; a highly valued
sense of community identity, fostered especially by temporary or permanent
exile; and a personality characterized, especially in the childhood years, by
marked inquisitiveness and gregariousness.2 While all three factors need not
2 There is a fourth factor which can also produce semi-speakers: late birth order in a large
family. Both in Sutherland and in Berks County I have worked with large families in which
the youngest two or three children were semi-speakers; the presence of a considerable group
of older siblings who have been to English-language schools seems to work against full
This woman’s daughter learned her Gaelic chiefly from a cousin of her father’s,
but the commonest figure in the linguistic socialization of the semi-speaker is
the grandmother. One woman, both of whose grandmothers were living dur-
ing her childhood, was sent to reside with the one who had very little English
while some of her siblings were sent to reside with the other. As a result she
learned Gaelic much better than those brothers and sisters, despite the fact
that her parents chose not to use it with their children. Some excerpts from
her account:
Although they spoke the Gaelic to each other, my mother and father,
they didn’t speak it to us. . . . My father and mother used to go to the
fishings, . . . and I stayed with this granny. And I heard nothing else but
Gaelic. . . . She wouldn’t let me off with – eh – anything that wasn’t right.
You know? She corrected me all the time. Because they couldn’t speak the
English, at all. It’s – it’s Gaelic all the time.
a cquisition of the parental language by the youngest members of the family. These semi-
speakers differ, potentially, from the others I shall discuss, in that the conscious or uncon-
scious choosing to acquire the home language may not have been present; some of them
seem to be more accidental semi-speakers than those who form the chief subject of this
paper. However, where a late birth-order child also experiences one of the three factors men-
tioned above, a strong language loyalty may in fact result. For example, two late birth order
semi-speakers in east Sutherland were also temporary exiles from the area, and now, living in
the home village again, they are strong language loyalists.
Another semi-speaker likewise claimed that she spoke better Gaelic than a sib-
ling because of greater contact with her grandmother:
I used to stay with my granny a lot, you see, this is the thing. And I – I sup-
pose my granny and grandfather, they spoke Gaelic all the time. More so,
I suppose – I suppose I heard my granny and grandfather more than my
mother and father, really. . . . My sister’s older, but funny enough, I think
I – I probably knew more than she did. And I think it was because I lived
with my granny so much. I was never out of my granny’s.
Perhaps the most improbable of the ESG semi-speakers is a woman in her late
thirties who has lived in London ever since she was six years old and has still
managed to maintain some hold on the language. Despite weak grammar and
a limited vocabulary, she can speak fairly freely in an unconventional but intel-
ligible fashion. The grandmother in whose household she and other family
members lived (during the war) until she was six is the figure to whom she
attributes the key linguistic influence in her life; her own parents followed
during her childhood the familiar pattern of speaking Gaelic in her presence
but not to her, though once she was old enough she began to insist on some
Gaelic interaction with them, especially with her mother. Gaelic for this exile
has strong emblematic value; it signals her Highland Scottishness, which she
greatly prizes, and the fishing-village heritage she identifies with:
This same exile semi-speaker exemplifies also the third feature typical of the
genesis of the semi-speaker: the exceptionally inquisitive, gregarious personal-
ity. Even her grandmother showed a strong tendency to speak English rather
than Gaelic to her, but eventually she herself demanded Gaelic from her par-
ents and her grandmother:
I don’t think it would have worried Mum and Dad if I didn’t speak it,
‘cause they n – I mean, they never bothered to tell me all the answers,
every time I asked. But they never sort of said to me “Come on, sit down
and we’ll teach you some Gaelic.” No. Even Granny. I used to ask her to
speak Gaelic to me, and she’d still revert to English.
As a child living in east Sutherland, and even later, living in a London house-
hold where every adult resident and visitor was a Gaelic speaker,3 this girl had
an active desire to understand what was said around her:
And anybody that came in, they were all Gaelic speakers. You know.
Perhaps I was frightened of missing out on something if I didn’t know
[Gaelic].
3 There was a large and rather cohesive group of related East Sutherland Gaelic speakers living
in exile in London during her childhood. Adults in this group spent most of their leisure time
in each other’s company.
today.4 The genesis of the semi-speaker is thus far from inevitable. As with
community language shift, where giving up a language is sometimes easier to
understand than maintaining it, it is often easier to see why one individual did
not become a semi-speaker than why another did. The London environment
can scarcely be called favorable to Gaelic, and in neither home did the parents
make any real effort to pass on a knowledge of Gaelic to the children. One
cousin insisted on access to Gaelic; the other, less surprisingly, did not.
The notion that there is a personality type which gives rise to the semi-
speaker is supported by a highly anomalous Pennsylvania Dutch case, where
the speaker is a boy of 15. A dialect speaker of that age in the Hamburg area
is almost unheard of; older speakers have sometimes been incredulous when
I told them of this boy’s ability. Like the London speaker of ESG, he speaks
an unconventional version of the dialect with aberrant grammar and a great
many loanwords from English, but, also like her, he speaks it quite readily. It
is characteristic of these two gregarious, out-going semi-speakers, in fact, that
they speak more freely than many other semi-speakers whose linguistic con-
trol of the dialect in question is noticeably better. The mother of this young
Pennsylvania Dutch semi-speaker gave an account of his acquisition of the dia-
lect which virtually duplicates the London-based ESG semi-speaker’s account
of her own dialect acquisition:
We never – pushed him, you know. “Now say this”, or “Say that”, or “This
is how you say it”. I mean, he was constantly asking us, you know. If there
was something he didn’t understand. . . . He always wanted to know. He
was constantly questioning you. . . . He just picked it up from being curi-
ous. Asking questions. And, of course, hearing it.
Like so many other semi-speakers, this young boy was responsive to older peo-
ple as a child:
4 The two cousins are of identical social-class background and are both of good intelligence
and similar educational background; differences of socioeconomic status or intelligence do
not seem to play a role, therefore. The semi-speaker cousin has a strong interest in language
in general which may or may not be shared by the monoglot cousin; it is impossible to say
whether this interest develops out of her bilingualism or, perhaps, helped to produce it.
Conclusions
Where the circumstances which favor one language over another in a language-
contact situation are rather overwhelming, it seems to me that explaining resis-
tance to shift demands more of our attention than explaining language shift.
It is usually all too clear why a local-currency, low-prestige language gives way
to some other language of wider currency and higher prestige. Consequently,
in investigating language death5 in eastern Sutherland and in Berks County,
I have concentrated on the anomalous cases where an imperfect speaker of
the local-currency dialect has persisted in making some use of that dialect
despite his awareness of its demographically and socially weakening position
and despite his imperfect control of its structure.
In the cases presented here it can be seen that positive exposure to a
language-loyal kinsperson, especially an older person, outside the nuclear
family can counterbalance even the conscious decision of the parents not to
transmit the local-currency language. Exile, temporary or even permanent,
can foster a sense of community identity that favors maintenance of the
threatened language. And, finally, there would seem to be a shift-resistant
personality, characterized by curiosity and an outgoing nature; people of this
personality type apparently need very little external encouragement beyond
mere exposure to the language in the home, despite the well-known fact that
exposure alone often produces the purely passive bilingual with no productive
skills at all.
These observations are explanatory rather than predictive, however. Where
any two of the conditions prevail, the likelihood of a semi-speaker resulting
is increased, but it is certainly not inevitable. For the individual as for the
5 I regard both these cases of dying dialects as legitimate cases of ‘language’ death, since in
each instance a local speech form, isolated from any other dialect of the same language, is
giving way not to a more standard or prestigious dialect of the same language but to a totally
different language. Where two quite distinct languages were spoken in eastern Sutherland
and in Berks County, only one language will survive; hence the term language death, although
what is dying is a regional dialect whose extinction will not mean the end of either Gaelic or
German.
References
Campbell, John Lorne. 1950. Gaelic in Scottish Education and Life. Edinburgh, W. and
A. K. Johnston for the Saltire Society.
Dorian, Nancy C. 1973. Grammatical change in a dying dialect, Language 49: 413–438.
———. 1977. The problem of the semi-speaker in language death, International Journal
of the Sociology of Language 12: 23–32.
———. 1978. The dying dialect and the role of the schools: East Sutherland Gaelic
and Pennsylvania Dutch in International Dimensions of Bilingual Education, ed. by
James E. Alatis, 646–656. Georgetown University Round Table on Languages and
Linguistics.
Fishman, Joshua A. 1971. The sociology of language, Advances in the Sociology of
Language I, 217–404. The Hague, Mouton.
Hill, Jane H. 1973. Subordinate clause density and language function, You Take the High
Node and I’ll Take the Low Node, ed. by C. Corum, T. C. Smith-Stark and
A. Weiser. Papers from the Comparative Syntax Festival, Chicago, Chicago Linguistics
Society.
Krauss, Michael. 1963–1970. Eyak Texts. Photocopy.
Lewis, Glyn. 1978. Migration and the decline of the Welsh language, in Advances in the
Study of Societal Multilingualism, ed. by Joshua A. Fishman. The Hague, Mouton.
Mackey, William F. 1973. Three Concepts for Geolinguistics. Publication B-42,
International Center for Research on Bilingualism, Quebec.
Mertz, Elizabeth. 1978. The Gaelic-speakers of Nova Scotia: Language maintenance
and language death in comparative perspective, mimeo.
Miller, Wick R. 1971. The death of language or serendipity among the Shoshoni,
Anthropological Linguistics 13: 114–120.
Salzmann, Zdenĕk. 1969. Salvage phonology of Gros Ventre (Atsina), International
Journal of American Linguistics 35: 307–314.
1 The Gaelic of the fishing villages of Easter Ross showed the greatest resemblance to East
Sutherland Gaelic, in all likelihood (see Watson 1974 and Dorian 1978: 143–4). The Gaelic of
the Easter Ross fishing villages is all but extinct, however (Watson, personal communication).
present2 and are distributed over three villages. The Gaelic speakers in any one
village are complexly interrelated to one another, and there are also kin ties
across villages; this is the result of about a century of forced endogamy among
the fisherfolk. Kinship networks are also often friendship networks, so that rela-
tives representing a variety of ages interact a good deal. Families also tradition-
ally ran rather large, so that siblings sometimes span more than a decade in age
ranges. As a result of these two facts, interaction networks which include both
fully-fluent, Gaelic-dominant bilinguals and low-proficiency s emi-speakers or
near-passive bilinguals, and even young English monolinguals, are fairly com-
mon. In one household, for example, four unmarried children, all adult, lived
in the home of their Gaelic-dominant mother, while two somewhat older (and
married) children lived nearby in the same village (village A). The three old-
est siblings were fully-fluent bilinguals, the next two were high-proficiency
semi-speakers, and the youngest of the siblings was a near-passive bilingual.
Relatives who visited regularly in this household, often spending entire eve-
nings there, included both fully-fluent bilinguals and English monolinguals. In
another village (village B), various relatives gathered irregularly in the home
of their eldest kinswoman, a fully-fluent woman who is now a nonagenarian.
Most of them were likewise fully fluent, but the nonagenarian’s high profi-
ciency semi-speaker daughter was often present, and also a low proficiency
semi-speaker kinswoman who was a next-door neighbour. Most of my remarks
about the claims of low-proficiency semi-speakers and near-passive bilinguals
to membership in the East Sutherland Gaelic speech community are based
on long-term participant observation of these two networks and another one
composed of East Sutherland exiles from village A residing in and around
London, plus less intimate knowledge of other similar networks. The networks
I have observed and participated in – these three as well as others – have been
altered by the deaths of one or more members over the 17-year period that I
have known them; therefore I use the past tense to describe them even though
they continue to exist in reduced form.
The most striking feature of the cross-generational interaction networks
among fisherfolk descendants was the ability of the low-proficiency members
to participate in Gaelic interactions. Despite their very limited productive
skills, they were able to understand everything said, no matter how rapidly or
uproariously. They never missed the point of a joke or failed to grasp a signifi-
cant tidbit of gossip. They occasionally supplied a translation of something dif-
ficult to hear or something poorly enunciated for the linguist-guest who spoke
2 The number was about 200 when I began fieldwork in East Sutherland in 1963–64.
a far more grammatical and likewise more fluent East Sutherland Gaelic than
they did.
The second notable feature of the participation of low-proficiency network
members was its sociolinguistic ‘fit’. What they actually said might be very
little, and some of their utterances were always grammatically deviant. But
since their verbal output was semantically well integrated with what preceded
in the conversation, and since it conformed to all the sociolinguistic norms of
the dialect, the deviance could usually be overlooked. Often the semi-speaker
or near-passive bilingual did not even have to finish the sentence; some fully-
fluent member of the group could step in, if there was a marked hesitation,
and supply the anticipated conclusion. Low-proficiency members of these net-
works, unlike the linguist-guest, were never unintentionally rude. They knew
when it was appropriate to speak and when not; when a question would show
interest and when it would constitute an interruption; when an offer of food
or drink was mere verbal routine and was meant to be refused, and when it
was meant in earnest and should be accepted; how much verbal response was
appropriate to express sympathy in response to a narrative of ill health or ill
luck; and so forth.
Two approaches to the speech community which seem more adequate,
in the sense that they do not define out of membership those who have low
productive capacity but high receptive capacity and who conform to the soci-
olinguistic norms, are provided by Hymes (1974) and Corder (1973). Hymes pro-
poses that the social group, rather than the language, be taken as the starting
point, and that we then consider ‘the entire organization of linguistic means
within it’ (1974: 47). This would enable us to start with the participants in
Gaelic verbal interactions, in East Sutherland, including the low-proficiency
semi-speakers and near-passive bilinguals, and define the speech commu-
nity so as to include them. Their inclusion would be appropriate not simply
because they are participants (so, after all, was the linguist-guest), but because
they are highly successful participants whose receptive skills and knowledge
of the sociolinguistic norms allow them to use their limited productive skills in
ways which are unremarkable (that is, provoke no comment).3
3 I have noted that semi-speakers’ East Sutherland Gaelic is grammatically deviant in ways that
are labelled “mistakes” by fluent speakers, which is true enough. But when left to their own
devices, so that they can speak when they wish to, briefly and in the structures they are most
comfortable with, semi-speakers are often able to reduce the deviance to the point where it
can be overlooked, especially in the flow of a general conversation. Semi-speakers are some-
times also “rescued” by a fluent speaker from the necessity of finishing more adventurous
sentences which they may have begun and in the middle of which they then hesitate.
Corder takes the self-perceived group as the basis for the speech community:
“A speech community is made up of people who regard themselves as speaking
the same language; it need have no other defining attributes” (1973: 53; italics
in original). This approach has the advantage of according well with the well-
integrated position in the Gaelic interaction networks of some individuals with
extremely poor active skills. The very low-proficiency semi-speaker who lived
next door to the (now) nonagenarian fluent speaker in village B expressed the
complete ease she felt when her visits to her neighbour coincided with those
of another relative, this one a verbally gifted, notably articulate fluent speaker:
Semi-speaker: J., she’s fluent – Gaelic speaker. But any – any time she’s in,
if she does [speak Gaelic], you know, I don’t – I just take it in my stride,
as – just as if it’s English, you know? It doesn’t worry me in any way.
Or I don’t get mixed up, and I know what they’re – I can join in the con-
versation, because I know everything they’re saying, you know. I haven’t
to stop and think or anything.
Investigator: Yes. Uh-huh. And if you joined in, would you join in in
English or Gaelic?
Semi-speaker: I would join in in Gaelic, you know. As best I could, y’
know.4
When this woman’s active skills were first tested, in the same year (1974) the
above interview was taped, the testing was done in her fluent next-door neigh-
bour’s home where she spent so much time. It proved a distressing experience
for all participants: neither of the women, though neighbours for years, had
realized how little active control of the dialect the younger woman had; nor
for that matter had I, or I would not have exposed her to the embarrassment
of “public” testing. She proved to be one of the very weakest speakers in my
sample, yet none of us had noticed her failings as an active speaker, thanks
to her skillful use of what proficiency she had and to her outstanding recep-
tive skills and sociolinguistic knowledge. Although it’s clear that she acknowl-
edges some weakness in her speaking abilities (“as best I could”), it’s also clear
that she feels included in the interaction (“I know everything they’re saying”).5
4 This interview was conducted in English and is quoted here verbatim. It proved extremely
difficult, almost impossible even, to interview this semi-speaker in Gaelic, since the pres-
ence of the tape-recorder produced a nervousness which compounded her difficulties with
Gaelic.
5 A young man in Berks County, Pennsylvania, a low-proficiency semi-speaker of Pennsylvania
Dutch, expressed exactly the same sense of inclusion in interactions with fluent Pennsylvania
Dutch speakers, for the same reasons.
At least until the testing took place, her neighbour certainly considered her an
adequate member of the Gaelic speech community. That was precisely why
the testing proved so distressing to all of us – it showed plainly that she was
actually less than adequate in productive East Sutherland Gaelic skills. But she
regarded herself, and was in turn regarded by fluent speakers, as a speaker of
East Sutherland Gaelic.
What is interesting about this case (and others in which low proficiency
semi-speakers and near-passive bilinguals participate successfully in Gaelic
interactions) is that it highlights the minimum requirements for membership
in a speech community. Fluency is not required, nor grammatical and/or pho-
nological control of the speech variety common to the participants. The for-
eign learner can achieve those things and still be only a participant in a speech
community and not a member if he or she does not also fully master receptive
skills and sociolinguistic norms. As Hymes has so often insisted, communi-
cative competence depends not only on knowing how to say something, but
also on knowing how to say it appropriately (1964a, 1964b, 1967, 1971, 1974, for
example). In fact it seems that knowing how to say relatively few things appro-
priately is more important than knowing how to say very many things without
sure knowledge of their appropriateness.
Low-proficiency semi-speakers, not to mention near-passive bilinguals,
meet none of Fillmore’s criteria for fluency (in the sense of speaking one’s
language well; Fillmore 1979: 93). In my experience their only productive
skill which is even close to normal is control of what Fillmore calls formu-
laic expressions (1979: 91–2,94). There are a great many formulaic expressions
which can be trotted out on suitable occasions; knowledge of their forms and
their suitability enables the user to participate actively in the verbal interac-
tion and helps to keep the interaction going forward smoothly, and thus earns
the user a measure of social approval.6 Observation of semi-speaker success
with these items offers support for Fillmore’s belief that “a very large portion of
a person’s ability to get along in a language consists in the mastery of formulaic
utterances” (1979: 92).
I noted at the outset of the discussion of low-proficiency semi-speakers and
near-passive bilinguals that any definition of the speech community which
implied productive control of the language in question would eliminate these
6 In the absence of strong skills in the use (as opposed to the form) of formulaic expressions,
my own strategy has been to master a good many East Sutherland Gaelic proverbs. It is easier
on the whole [to know] when they are appropriate, and they also earn the user strong social
approval for the same set of reasons noted in semi-speaker use of formulaic expressions, as
well as for control of highly valued traditional material.
7 It was not uncommon for me to be in all three villages in the course of one day, and I was
simply unable to monitor my speech carefully enough to guarantee only the correct forms for
whatever village I was in at a given time.
8 Control of the phonology is one of the stronger points in my own East Sutherland Gaelic so
that foreign accent [at least in the production of single words at a time] is not an explanation
for this outcome.
The chief reason for the absence of social evaluation of linguistic varia-
tion must be the fact that the East Sutherland fisherfolk constituted until very
recently an undifferentiated social group: all followed the same occupation,
all were poor, all were members of a stigmatized ethnic group, none who
remained in East Sutherland had more than the legal minimum of education.
Intra-group status distinctions rested on skills or moral character, not on differ-
ences in occupation, education, or wealth. All present-day bilinguals grew up
in active fisherfolk households, and all share the same status in the local social
hierarchy as a result.
How deep low-proficiency semi-speakers’ and near-passive bilinguals’
knowledge of regional variation is, I cannot be sure. Everyone in the Gaelic-
speaking group, regardless of level of proficiency, can produce on request a
short list of regional variants which are local stereotypes, much discussed and
frequently imitated for the purpose of poking fun. But among fluent speakers,
despite the fact that all produce the same small list of stereotypes when asked
about regional variation, there is awareness of many more variants than they
typically offer. If asked about words not among the stereotypes, they can often
come up with the “alien” forms used in other villages, and when they listen to
tape recordings made in another village they spot the regional variants read-
ily, without any prompting. They are eager to discuss them, in fact, and will
concentrate on them to the exclusion of content, oftentimes. I have neglected
to press the low-proficiency semi-speakers and near-passive bilinguals to the
limits of their skill in these matters, so that I am unable to say whether they can
match the fluent speakers’ knowledge.
Another issue on which I cannot at present shed any light is the position of
true passive bilinguals in the speech community. I know that there are such
people – individuals who understand what is said, but cannot produce Gaelic
speech – since I once heard a young woman whose inability to pronounce East
Sutherland Gaelic words was both evident and self-admitted translate a Gaelic
conversation for the benefit of a foreign visitor. Unfortunately I have not myself
worked with any true passive bilinguals, and so I have little notion of the actual
extent of their passive abilities. Their existence is acknowledged by the bilin-
guals, who complain that one can’t count on keeping a secret through use of
Gaelic if these people are about. Although they are sometimes “participants”
in Gaelic interactions by dint of injecting English comments and responses
into a Gaelic interchange which they have understood, I have never heard any-
one deliberately address a Gaelic remark to them (except as a direct challenge
to them to reproduce it – a test of their abilities), nor have I ever heard anyone
refer to them as Gaelic speakers. In both these respects they differ from semi-
speakers of any proficiency, and even from the near-passive bilinguals, who are
References
One of the regrettable but interesting things about language death is its long
history. It’s anything but a new phenomenon, and we have a lot of extinct lan-
guages littering the shores of linguistic history to prove it. On the other hand,
our own time seems a little curious in one special respect, namely in respect
to the number of languages which have persisted with pretty fair strength for
what seems like a long period, only to weaken in what seems like a rather short
time and suddenly wind up in a downslide toward extinction.
In this country and Canada, for example, some long-established popula-
tions with very distinctive customs and languages which have been secure for
centuries are suddenly in trouble. The geographical region doesn’t seem to
matter – it’s the same story regardless of location. Cajun French in Louisiana
is in the same trouble as French Canadian in Maine. Pennsylvania Dutch (that
is, German) among the secular (non-Anabaptist) Pennsylvania Dutchmen is
threatened in the same fashion as Scottish Gaelic in Cape Breton. None of
these is a particularly johnny-come-lately immigrant language – the oldest
of them have been in place for several centuries, and their speaker populations
have been relatively loyal and stable, sometimes also reinforced by continu-
ing immigration (this is the case with Canadian reinforcement of the French-
speaking population in Maine and Highland Scottish reinforcement of the
Gaelic-speaking population in Nova Scotia, whereas the Cajun and Penn Dutch
populations seem to have recruited more by absorbing incomers or non-native
locals than by major inflows of new immigrants).
In general the twentieth century seems to be notable for the large num-
ber of languages which are either obviously dying out or showing marked
signs of contraction such as simplifying structure, functional restriction, and
loss of speakers at the margins of the community. Whether this century is actu-
ally any more characterized by these phenomena, or whether we’re only better
informed about the number of cases and their wide geographical distribution,
is unclear.
Some people are inclined to argue that this is a particularly pernicious time
for languages which are isolated, or enclaved, or represented by rather thin
populations, or heavily outbalanced by languages of wider currency. People of
this persuasion usually point to ease of modern travel, the “global village” phe-
nomenon, the power of the modern nation-state to affect the lives of even its
most outlying citizens, the savage thoroughness of the more modern instances
of genocide or attempted genocide, the spread of literacy, the penetration of
radio and television, and so forth.
I think there is no denying any of these factors. They are all very real and
very potent. Anyone who has worked with even a single threatened language
can attest to the force of negative policies (or even only negative attitudes)
spreading out from a central government and discouraging or perhaps penal-
izing speakers of languages or dialects other than the officially state-promoted
language. Similarly the ouster of traditional activities which fostered minority
languages – social gatherings like the ceilidh in Scotland and Ireland, pedagog-
ically-oriented verbal routines such as Aesopian tales, fairy tales and rhyming
genres (all directed toward children) in Albanian-speaking Greek communi-
ties (Tsitsipis 1983: 27), the most formal styles of public speaking in the Cupeño
and Luiseño communities in California (usurped by English; Hill 1973: 45) –
by passive or active verbal events which involve only or mainly the state-
promoted language has a pronounced, unmistakably deleterious effect on the
strength of the minority languages in most cases.
This is the usual outcome, more or less the predictable outcome, and it sur-
prises no one. It’s not the inevitable outcome, however, since people seem to
be capable of quite remarkable segmentation of their lives, including linguis-
tic segmentation. It’s hardly encouraging for a language to be excluded from
the schools, ignored in broadcasting, discouraged in public life, and unpro-
vided for in any officially sponsored activities whatever. But in some societ-
ies it seems to be possible for people to accept a very restricted role for their
native speech form, such that they assume it will be used only in the hearth-
and-home sphere; they may even welcome the specialization of their mother
tongue as an in-group marker. Where there is a deep gulf between the minor-
ity-language group and the dominant-language group, as with certain Native
American tribes, the home language may be jealously guarded from mem-
bers of the majority language group, treated along with things like religious
ceremonials as a privileged form of in-group knowledge, not to be casually
exposed to outsiders or shared with them. There are entire societies in which
the home language has good standing but has been traditionally restricted in
use without any threat to its ultimate viability (German Switzerland, where
Schwyzertütsch is seldom written and almost never used in circumstances of
any formality is a case in point), and of course quite a lot of societies exist in
which the language of highest prestige is not the local language – most often
where religion is involved, as in many Islamic but non-Arab societies.
Since there are recognized instances of all these exceptions to any general
tendency to succumb to centralizing dominant-language pressures, the ques-
tion may be why there aren’t more such exceptions rather than why there are
any. In connection with the relatively long-standing ethnic communities now
experiencing survival difficulties in the U.S. and Canada, it seems to be the tem-
per of the times which works most against compromises which would allow
continuance. Despite the “melting pot” myth, special provision for certain
mother-tongue rights of long-established non-English populations was made
in several cases into at least the early twentieth century: French in Louisiana
(Kloss 1977: 112–113), German in Pennsylvania (ibid.: 146–147), Spanish in New
Mexico (ibid.: 130–1:31), for example. Assimilative pressures have nonetheless
been strong, of course, and the great nineteenth-century waves of European
immigration undoubtedly created tensions for longer-established populations
as concerns were increasingly voiced over the effect of home-school bilingual-
ism on intelligence and on loyalty to the national state.
Although functional segregation in language use is a perfectly feasible way
of managing and maintaining two or more languages, unless the wider com-
munity is one in which this is the norm (as in German Switzerland and in
Somalia, for example; see Pride 1971 for the latter case), there seems generally
to be little support for this course and little understanding of its frequency of
occurrence in a good many parts of the world. In most of western Europe and
the areas colonized by western European nations, the prevailing attitudes have
most definitely not been favorable to full-fledged linguistic dualism of any sus-
tained kind.
On the basis of my own work with two minority languages, one in Great
Britain and one in the eastern U.S., and also on the basis of reports from other
researchers working in similar settings, I would propose a rather widespread
phenomenon which I have dubbed “tip” in describing the British case (Dorian
1981: 51). This phenomenon can be conceived metaphorically as a gradual accre-
tion of negative feeling toward the subordinate group and its language, often
accompanied by legal as well as social pressure, until a critical moment arrives
and the subordinate group appears abruptly to abandon its original mother
tongue and switch over to exclusive use of the dominant language. Because of
the seeming suddenness of the switch-over, it’s rather like watching a structure
slowly eaten invisibly away at the bottom topple over almost without warning.
Yet when the tip has occurred and one begins to examine the period which led
up to it, the tip is seldom if ever so sudden as it initially appeared.
The most striking level at which tip occurs is, to my own perception, that of
the family. I would like to introduce two cases, one among the Gaelic-speaking
fisherfolk of East Sutherland in the extreme northeast of the Highland Scottish
mainland and the other among the secular Pennsylvania Dutch of the Hamburg
area in Berks County, Pennsylvania. In each of these family cases there were a
good many children: seven in the Gaelic-speaking family, and twelve in the
Penn Dutch speaking family. The parents in each family were skilled bilin-
guals but spoke Gaelic and Penn Dutch (respectively) by habit and preference
with each other and within the home generally at the outset of their family
life. In each of these families the elder children – the first four of the seven in
the Gaelic speaking family, the first nine of the twelve in the Dutch-speaking
family – were raised as, and became, fully fluent speakers of the parents’ origi-
nal mother tongue. In the Gaelic-speaking family no conscious change in the
parents’ linguistic behavior toward the three youngest children seems to have
taken place, whereas in the Dutch-speaking family there was an acknowledged
though unexplained change of that type. In each family the three youngest
children emerged as imperfect speakers (or, in the case of the youngest child
in the Gaelic-speaking family, as a near passive bilingual with very little ability
to generate utterances in the parental mother tongue).
Several aspects of the two cases are especially interesting. One is that the
parents’ intentions probably mattered relatively little, since the results were the
same in a case where the parents deliberately changed their behavior and in a
case where they didn’t. The behavior of the peer group outside the family and
also the sheer number of older siblings who had attended English-language-
only schools and were using a good deal of English among themselves in or
around the home most likely had more impact on the language-acquisition
patterns of the youngest children than the parents’ own linguistic behavior or
transmission plans, since in both homes the parents continued to use the origi-
nal mother tongue with each other and with the older children (and with all
the children in the Gaelic household). This means that the youngest children
received at least a good deal of exposure to that language.
Another interesting facet of the two cases is the clarity of the fully-fluent
as opposed to the less than-fully-fluent demarcation line among the children,
and the unimportance of the size of the age-gap where the demarcation line
falls. No one in either family is in any doubt about which child is the last of the
fully fluent and which is the first of the imperfect speakers. The three youngest
children in each family are just as aware of their less-than-fully-fluent status as
the older children are, although in the Gaelic speaking family the three young-
est are not particularly sensitive about it whereas in the Penn Dutch family
the three youngest mind very much that they are not as competent in Dutch
as their older siblings. In the Gaelic family the last of the fully-fluent children
is two years younger than the next oldest fully fluent sibling and only one year
older than the first of his imperfect-speaker siblings. In the Dutch family the
last of the fully-fluent children is three years younger than the next oldest
fully-fluent sibling and two years older than the first of the imperfect-speaker
siblings.
My data from tests among the Penn Dutch-speaking siblings are not yet fully
analyzed. But I can present here two sets of results from testing of the Gaelic-
speaking family, offering clear indication of how sharply the fluency line can
be drawn between siblings only a year apart in age. In Table 1, Speaker 1 is the
mother of the family. Speaker 2 her fourth child, and Speaker 3 her fifth child.
Neither child was married; both lived in the mother’s household. There is only
a year’s difference in age between the two siblings. The test was for analogical
levelling in four structures, the frequent use of analogically levelled forms being
a notable marker of the imperfect Gaelic of the less-than-fully-fluent. Identical
sentences were presented to each of the three speakers, in individual elicita-
tion sessions, for translation from English into Gaelic. (It should be noted that
translation is a relatively natural, high-frequency occurrence in a community
where kin networks include both bilinguals and monolinguals, since remarks
or conversations in one language will often be recounted in translation at a
later time to a kinsperson with whom the language of the original interchange
is not the normal language of social interaction). All three of these speakers
knew me well, were comfortable with me, and had done this kind of work with
me before; I had been around the district over a period of a good many years
and the test sentences were couched in a form of English which was reason-
ably normal for the local English dialect.
Speakers 2 and 3 may be only a year apart in age, but Speaker 2 is much
closer to his mother, 29 years older than he, than to his sister one year younger,
in his linguistic usage on this measure.
In Table 2, Speaker 2 is compared with Speaker 3 again and also with
Speaker 4, the latter being the next younger sibling, another sister four years
younger than Speaker 3 (and so five years younger than Speaker 2). The struc-
tures tested (by the same type of elicited translation procedure) were control
of three tenses and control of three embedded structures (for discussion of
the theoretical difference in the difficulty of the structures concerned, see
Dorian 1982).
Although Speakers 3 and 4 show some marked differences in their control
of the particular structures tested (Speaker 3 being distinctly better than her
sister at using the conjunction ‘that’ and distinctly worse at forming relative
clauses and at constructing the conditional), the really striking difference is
between their performances taken together as compared with their brother’s.
He in fact misproduced only one form in the entire set of sentences, whereas
his sisters misproduced 15 each. Overall, then, they performed much like each
Speaker 1 1a 0 0
Speaker 2 2 0 0
Speaker 3 8 3 5
Speaker 1 0 43 1 2
Speaker 2 0 43 2 4.5
Speaker 3 5 43 21 49
other, despite the four years between them; the brother, though only a year
older than Speaker 3, performed quite differently – namely like the fully-fluent
speaker he is.
For the Berks County Pennsylvania Dutch I can’t be certain that intrafam-
ily tip of the dramatic abruptness I found in the 12-sibling group is a frequent
occurrence, since I worked extensively with two kin networks only. But among
the East Sutherland fisherfolk I know of several similar cases where groups of
siblings were sharply and abruptly divisible according to full fluency versus
imperfect-speaker control of the local Gaelic, even though I undertook close
testing of only this one highly available and highly cooperative family. The
community at large was well aware of the phenomenon, in fact, and readily
identified cases in their own kin networks or others.
When it comes to tip on the community-wide level, the cessation of home-
language transmission can seem equally sudden and surprisingly datable. In
the smallest of the East Sutherland fishing villages, for example, I found that
people were able to identify the last primary-school class whose members
Table 2. Comparison of control of three tenses and three embedded structures by three
siblings in a single Gaelic-speaking household
Speaker 2 1 73 1
Speaker 3 15 75 20
Speaker 4 15 75 20.5
a The number of instances of a given structure sometimes differs across speakers because a
particular speaker offered two variants for a particular structure, each of which was recorded,
evaluated, and counted in arriving at the tabulation.
b The figures here differ very slightly from those recorded for Speaker 3 in Dorian 1982: 39,
where she appears as WR; results of testing of the future were retabulated and recounted
subsequently, with one additional instance recognized for WR.
regularly used Gaelic on the playground whenever they were let out to play
during the schoolday. The class only one year younger, everyone agreed, might
occasionally use Gaelic on the playground, but did so seldom; and they did not
typically become, or remain, fully-fluent speakers, whereas their immediate
predecessors did. No one could give a particular reason why this change in lan-
guage behavior should have come exactly when it did, but they agreed on its
timing. It was as if a consensus had tacitly been reached among the children –
and that was that. Not merely coincidentally, the brother identified as Speaker 2
in the tables above was a member of the last primary-school class to use Gaelic
regularly on the playground, and the sister identified as Speaker 3 was a mem-
ber of the immediately following class, which did not regularly use Gaelic on
the playground. This again suggests that even had the parents in that family
made a concerted effort to keep their last three children in the fluent-speaker
fold, the climate among the youngsters themselves would have made it an
extremely uphill battle. Very strict and very determined parents are certainly
known to succeed in producing fluent bilingual children, and then to succeed
in maintaining that bilingual fluency in their children, within communities
unfavorable to the phenomenon; but in my own experience most such cases
involve either middle-class (often intellectual) parents, or, alternatively, an
only child. (One exception which comes to mind did involve a Scottish Gaelic
family. They lived in a very isolated district on the west coast of Scotland: the
parents were not middle class, nor, so far as I can recall, was there only one
child, but the father was considerably older than the norm for a parent in that
community and was a formidable and demanding figure in the household life.)
Reports of community-wide tip turn up with some frequency in the growing
literature of language shift (see, for example, Gal 1979, Hinojosa 1980, Mertz
1980). Because parents in communities where transmission failure seems sud-
den often simply decline to raise their children as bilinguals, usually citing
concern for the children’s success in school or ability to get ahead in the world
as reasons (Denison 1971: 166–167; Dorian 1981: 104; Huffines 1980: 52; Pulte 1973:
426; Timm 1980: 30), some scholars have raised the question of whether the
passing of such languages ought rather to be considered “language suicide”
than language death (e.g. Denison 1977, Greene 1972). But this is to ignore the
long history, usually stretching centuries into the past, of relentless pressure on
the non-dominant language.
The Zapotec case presented by Hinojosa (1980) is particularly interesting,
because Zapotec in fact showed a relatively unusual degree of resistance to
the spread of Spanish among Mexican Indian populations. The Zapotec had
been doubly resistant to dominant language pressure, what is more, since in
the pre-Conquest period they successfully fought off the Aztecs and retained
their independence and identity to a unique degree (Hinojosa 1980: 28). The
town of Juchitán, not served by the railroad and the loser in the rivalry for
capital-city status in the Isthmus of Tehuantepec, remained strongly monolin-
gual in Zapotec for a surprisingly long time. But in the early 1970s the discovery
of oil in a nearby coastal area led to the creation of a new port, Salina Cruz,
relatively near to Juchitán. For the first time there was strong economic incen-
tive for Juchitán natives to acquire Spanish, since good jobs became available
to those with control of Spanish (op. cit.: 28, 30). It would seem that the early
so dramatically after persisting with anomalous strength for so long. The fail-
ure of linguistic will under these circumstances is a measure of the potency of
long-brewing negative pressures and the fragility of isolation as a buffer, since
a serious breaching of that isolative buffer can produce such rapid decline in
a previously resistant population. It’s possible that less isolated communities,
with longer experience of compromise (for example, such a compromise as
the linguistic domain-separation discussed above) have an advantage in sur-
vival potential precisely because they have had a prolonged period in which to
learn to cope with pressures for linguistic assimilation. Metaphorically speak-
ing, the more isolated linguistic groups may resemble North American Indian
tribes or South Pacific island populations exposed to measles for the first time
and carried off in disastrous numbers by the unfamiliar contagion. The mea-
sles virus was long in existence, but slow to reach them: when it did, they suc-
cumbed with terrible swiftness. In something of the same fashion the “virus”
of hostility to non-mainstream languages may gather strength for a very long
time and when it finally breaks through to an isolated community, carry the
minority language off in an equally swift and deadly wave of social contagion,
producing the phenomenon of linguistic tip.
References
fuller set of Tariana evidentials, for which he is faulted, apparently by the com-
munity in this case, too (ibid.).
Dika, also in his seventies, speaks both Tariana and Tucano to his children,
but his Tariana is “known to be hard to understand and frequently ungram-
matical” (2001: 423). He uses younger people’s forms “often” and inserts a
Tucano morpheme “now and again”; as a result “he is constantly made fun of
and called . . . ‘useless’ ” (ibid.).
The Tariana of the next generation shows certain recognizable deviations
from that of their fathers, with structural intrusions diffused from East Tucano
languages (e.g. possession marked with juxtaposition of terms instead of pos-
session marked with prefixes) and with certain phonological or structural
changes (e.g. reduction of –wa to –a, use of –hipe instead of –hipita as the
classifier for ‘land’). Aikhenvald tabulates the deviations in question (2001: 417–
418) and recognizes on the basis of their use a “younger speakers’ ” or “younger
people’s” Tariana (2001: 423–424).
Candi’s eldest son Maye is “the best Tariana language expert” and “the most
respected representative” of this first-descending generation, with a very clear
pronunciation and “morphologically complex” speech (2001: 423). He “only
occasionally” uses younger speakers’ forms, and he has “good lexical and cul-
tural knowledge” (ibid.). Maye is actually just one year younger than his uncle
Leo, but Leo as a member of Candi’s own sibling set (and of Candi’s generation
in that sense) “has more authority” than Maye (ibid.) Candi’s eldest daughter Oli
is a fluent speaker “with a comprehensive knowledge of the traditional kinship
system, for which she is respected”, but she speaks “a typical younger speaker’s
Tariana” and the occasional Tucano morpheme creeps into her Tariana (ibid.).
Maye, Leo, and Candi correct her speech and “behind her back she is called
Yasenisado ‘Tucano woman’ ” (ibid.). Gara, another of Candi’s sons, is “a good
and witty storyteller”, but his Tariana is “full of Tucano calques and ungram-
maticalities” (ibid.). Aikhenvald reports that “his speech is condescendingly
referred to as Gara yarupe ‘Gara’s thing’ ” (2001: 424). Candi’s youngest son Ze,
a trained schoolteacher, is “very fluent in younger people’s Tariana” but aspires
to more traditional speech: Aikhenvald describes him as “desperately trying
to make his speech as archaic as possible” and as “the greatest partisan of the
correct Tariana spoken ‘the way our fathers speak it’ ” (2001: 424). His effort to
become a language authority “is appreciated by the community” even though
his practice falls short (ibid.).
Yuse Paiphe’s children are all fluent in younger speakers’ Tariana, but only
his son Ñu “is considered ‘reasonable’ ” as a speaker, with a clearer pronuncia-
tion than the others (unreduced vowels and aspirated consonants). Ñu’s older
brothers Saba and Kiri have reputations as “ ‘deficient speakers’ ”, their stories
full of Tucano calques, their speech lacking the traditional person marker
and without the full set of Tariana evidentials: “their lack of language compe-
tence is constantly commented upon” (2001: 424). Yuse Paiphe’s daughters, like
Candi’s daughter Oli, are “frowned upon as ‘Tucano women’ ” (ibid.), evidently
for the same sort of Tucano influence in their Tariana.
Dika’s sons Emi and Raimu both use younger speakers’ forms and insert
Tucano morphemes “occasionally”. Emi, the elder, is “acclaimed as one of the
best storytellers in the village”, but Raimu “does not feel confident enough to
tell a full story in Tariana”. Emi and Raimu, like Saba and Kiri, are considered
deficient speakers: Aikhenvald reports that “their inability to ‘speak correctly’
is constantly lamented by other members of the community” (2001: 424–25).
These and other profiles offered by Aikhenvald are consistent with her
depiction of the Tariana concept of “speaking well”, and overall it is clear that
she intends to convey the community’s judgments of the speakers in question.
Her portraits leave no doubt that age in the sense of generational membership
has a bearing on speaker status, and her identification of particular linguistic
features that distinguish a more traditional form of Tariana from a younger
people’s Tariana demonstrate that age has a bearing on language structure
as well, even though not all older-generation speakers represent traditional
Tariana equally well (Yuse Paiphe and Dika falling particularly short of the
ideal for their generation).
and sounds bad, and another too much like a shaman’s preaching or archaic
(‘the way the old, old people talked’)” (ibid.). But Bloomfield notes that he was
surprised to find that despite his own relatively slight acquaintance with the
language he was able to share in the judgments that native speakers made,
just as a foreigner could come to share such judgments about English speak-
ers’ usage: “[I]f he listened to us long enough and . . . fortune favored him”, he
would pick up the social values (“normal good form”, “deliberate and elevated”,
“unidiomatic, vulgar, pedantic”) that English speakers attach to competing but
semantically equivalent structures (ibid.).
The individual whose Menomini speech Bloomfield rates most highly is a
woman in her sixties, Red-Cloud-Woman, who “speaks a beautiful and highly
idiomatic Menomini”; fluent also in Ojibwa and Potawatomi, she may have spo-
ken a little Winnebago as well but “knows only a few words of English” (1964:
395). Her husband, Storms-At-It, speaks Potawatomi as well as Menomini but
knows no English at all. In Menomini he “often uses unapproved, – let us say,
ungrammatical, – forms which are current among bad speakers”, yet he also on
“slight provocation” shifts “into elevated speech, in which he uses what I shall
describe as spelling-pronunciations, together with long ritualistic compound
words and occasional archaisms” (ibid.).
A man in his fifties, Stands-Close, son of a man known as “an oracle of old
traditions”, speaks a “well up to standard” Menomini “though less supple and
perfect than Red-Cloud-Woman’s”; but it is “interlarded with words and con-
structions that are felt to be archaic” (ibid.).
Little-Doctor, in his sixties, had huge vocabularies both in English and in
Menomini. In English he had “a passion for piling up synonyms” and his vast
Menomini vocabulary made him an explicator of rare words to fellow-speak-
ers. Bloomfield reports that “in both languages his love of words sometimes
upset his syntax, and in both languages he was given to over-emphatic diction”
of the spelling-pronunciation type (ibid.).
At the very low end of the proficiency scale Bloomfield places White-
Thunder, a man of about 40, who speaks “less English than Menomini”, a seri-
ous condemnation in view of the fact that Bloomfield describes his Menomini
as “atrocious”: “his vocabulary is small; his inflections are often barbarous; he
constructs sentences of a few threadbare models” (ibid.).
Because Bloomfield states that he is able to make the same judgments that
Menomini community members make, it is not always clear whether the
specific characterizations he offers are his own or the community’s unless,
as in Stands-Close’s case, he uses a passive construction like “are felt to be
archaic”. Unlike Aikhenvald he does not provide directly quoted renderings of
In ongoing work with the Gaelic dialect spoken by fisherfolk and their descen-
dants in the village of Embo, on the east coast of the northern mainland
county of Sutherland in Highland Scotland, I have had occasion to refine my
original age- and proficiency-based speaker categories, developed on the basis
of fieldwork in the 1960s and ’70s, in the light of resumed fieldwork in the 1990s
and the first decade of the 21st century, and to attach less blanket importance
to the role of age as a correlate of proficiency (Dorian 2010). This change was
prompted above all by two conspicuous instances of discrepancy between my
assessment of speaker skills, as a linguist focusing especially on grammatical
conservatism, and the community’s own assessment of speaker skills, reflect-
ing a different conception of “speaking well”.
Material collected in Embo in the ’60s and ’70s showed a surprisingly large
amount of change in progress in Embo speakers’ Gaelic, with older speakers by
and large using more traditional grammatical forms (and to a lesser extent also
phonology) than younger speakers, within the ranks of the fully fluent, and
with a final group of imperfect speakers maintaining a limited use of Gaelic
with noticeably deviant grammar and phonology. The community itself rec-
ognized these imperfect speakers, although they had no special term for them
and even though the group members differed from one another in grammati-
cal intactness and level of fluency. I termed these imperfect speakers semi-
speakers in order to distinguish them from the fully fluent (Dorian 1973: 417).1
The community did not recognize an older/more traditional vs. younger/less
traditional division among fluent speakers, but my particular speaker sample
happened to divide rather neatly between a group of speakers among whom
a linguistically well respected same-age pair were the youngest, on one side
of the divide, and a group of fluent speakers among whom the oldest were six
years younger, on the other. In view of the age divide I termed the older part
1 The term semi-speaker does not necessarily translate well to other speech communities. As
noted, the East Sutherland Gaelic semi-speaker group includes a variety of levels of fluency
and proficiency, so that I have sometimes distinguished among high-proficiency, middle-
proficiency, and low-proficiency semi-speakers in characterizing degrees of grammatical
intactness. The characteristics shared by all members of the group go beyond the presence
of considerable grammatical deviation to the conspicuous use of certain strongly disfavored
elements in their Gaelic and the absence of any conversation partners with whom Gaelic is
their default language. Evaluation of speaker skills and the semi-speaker concept/category
are discussed more fully in Dorian 2010.
of the sample older fluent speakers (OFSs) and the remainder younger fluent
speakers (YFSs). In describing the differences between the Gaelic of these two
fully fluent groups I then tabulated the data that reflected grammatical change
in progress in terms of group-based characteristics.
Taken as groups, the OFSs and YFSs were clearly distinguishable by their
different degrees of grammatical conservatism (see Dorian 1981: 147, Table 7).
As in the Tariana case, that is, it was possible to identify a more conservative
older speakers’ fluent Embo Gaelic and a somewhat less conservative younger
speakers’ fluent Embo Gaelic. But in one paper arising from that early fieldwork
I noted the anomalous finding that in extensive translation-task testing the
Gaelic of one young man considered fully fluent by the community resembled
that of his OFS mother, 29 years older than he, mainly in one particular respect,
namely the preservation of irregularities in four grammatical structures, while
in a good many other respects his grammar more nearly resembled that of a
semi-speaker sister only one year younger than he (Dorian 1977).2
Continuing fieldwork in the 1990s (conducted by tape recorder and tele-
phone, since health difficulties precluded a return to the field site)3 confirmed
my earlier view that “Seònaid”, the youngest individual among the OFS group,
enjoyed a high reputation as an Embo Gaelic speaker and that the young man
“Rory” was considered a fully fluent speaker, while his one-year-younger sis-
ter “Elsie” was not, in spite of the similarities in the siblings’ grammar apart
from the different degree to which they preserved the four irregular structures.
But ongoing fieldwork also confirmed that Seònaid was less conservative in
regard to certain grammatical structures than most of her OFS peers, and that
Rory was more deviant in certain grammatical structures than most other YFSs
and even than some semi-speakers. That is, again as in the Tariana case, the
2 Translation-task testing was unusually easy and effective with fluent East Sutherland Gaelic
speakers. In the fisherfolk communities the routine practice in producing reported speech
was to use whichever language a conversation partner favored. With language shift well
underway, all remaining Gaelic speakers had younger kinfolk, and sometimes neighbors,
who were English monolinguals and for whose benefit remarks originally made in Gaelic
were regularly “quoted” in English; the same was true in reverse of remarks originally made
in English, which were “quoted” to fellow Gaelic speakers in Gaelic (Dorian 1997). In these
small communities remarks of high interest were constantly passed along in the course of
daily conversation, and the frequency of translation in both directions seemed to have made
all speakers proficient translators. This was less true of semi-speakers, but the Embo semi-
speakers in particular (as compared with semi-speakers among the fisherfolk speakers of
Brora village) proved surprisingly adept at translation and adventurous about undertaking it.
3 Telephone conversations were recorded with the permission of the other speakers.
4 Rory died suddenly in his sixties early in 1993, whereas Seònaid died in her nineties early
in 1999. The fact that she outlived him by six years during the latter phase of my fieldwork
accounts for the fact that her token counts are considerably higher than his.
5 The small number of surnames in Embo village, together with traditional naming practices
that resulted in repetition of certain given names across adjacent generations of any fam-
ily, gave rise to multiple instances of identical official names within the village. The official
names were of little practical use, consequently, and individuals were identified instead by
by-names. Superficially these resemble nicknames, but because a great many of them had
humorously mocking elements they were used only in reference and not in address. By-name
elements were frequently shared by some or all of an individual’s children and often carried
over into a third generation as well. They therefore served to identify lines of descent and
relationship, just as family names would in a community with a wider selection of surnames.
5 Conclusion
by Kroskrity, are useful correctives in this respect. The value of the briefer
sketches is greatest, however, if, as in Aikhenvald’s Tariana case, community
language ideology is fully considered and local views of speaking well are
clearly identified. This sort of consideration has proved essential in the Embo
case to explaining the sharp difference in the way age-adjacent brother and
sister Rory and Elsie are evaluated as speakers by their own speech commu-
nity, and it may prove to be important more generally in revealing differences
between the criteria linguists apply and the criteria community members
apply in appraising individuals’ ability to speak well. Linguists may therefore
need to be more cautious about attaching importance to the role of age as a
correlate of proficiency.
References
Aikhenvald, Alexandra Y. (2001). Language awareness and correct speech among the
Tariana of Northwest Amazonia. Anthropological Linguistics 43:411–430.
——— (2003). Teaching Tariana, an endangered language from Northwest Amazonia.
International Journal of the Sociology of Language 42:125–39.
Bavin, Edith L. (1989). Some lexical and morphological changes in Warlpiri. In
Investigating Obsolescence, Nancy C. Dorian (ed.), 267–286. Cambridge: Cambridge
University Press.
Bloomfield, Leonard ([1927] 1964). Literate and illiterate speech. In Language in culture
and society: A reader in linguistics and anthropology, Dell Hymes (ed.), 391–396. New
York: Harper and Row.
Dorian, Nancy C. (1973). Grammatical change in a dying dialect. Language
49:413–438.
——— (1977). The problem of the semi-speaker in language death. International
Journal of the Sociology of Language 12:23–32.
——— (1981). Language death: The life cycle of a Scottish Gaelic dialect. Philadelphia:
University of Pennsylvania Press.
——— (1997). Telling the monolinguals from the bilinguals: Unrealistic code choices
in direct quotation within Scottish Gaelic narratives. International Journal of
Bilingualism 1:41–54.
——— (2010). Investigating variation: The effects of social organization and social set-
ting. New York: Oxford University Press.
Jones, Mari C. (1998). Language obsolescence and revitalization: Linguistic change in two
sociolinguistically contrasting Welsh communities. Oxford: Oxford University Press.
Kroskrity, Paul V. (1993). Language, history, and identity: Ethnolinguistic studies of the
Arizona Tewa. Tucson and London: The University of Arizona Press.
Schmidt, Annette (1985). Young people’s Dyirbal: An example of language death from
Australia. Cambridge: Cambridge University Press.
Silverstein, Michael. ([1981] 2001). The limits of awareness. In Linguistic anthropology:
A reader, Alessandro Duranti, ed., 382–401. Blackwell: Oxford and Malden, MA.
Voegelin, C. F., and Voegelin, F. M. (1977). Is Tübatulabal de-acquisition relevant to
theories of language acquisition? International Journal of American Linguistics
43:333–338.
Another point is that in the Vaupés, language is by far the most impor-
tant marker distinguishing language-aggregates and their members. It
is primarily the Bará language which all Bará Indians share and which
separates them as a category from Indians affiliated to other languages.
In most other multilingual situations which have been reported on, lan-
guage is but one of several such markers, others such as physical charac-
teristics, dress, differences in technology, eating patterns, etc., being of
equal or greater importance, at least in the eyes of the natives. As stated
above, formal language affiliation in the Vaupés is determined by mem-
bership in a named patrilineal descent group, which also confers the
right to manufacture certain ceremonial artifacts (that is limited to adult
men) and to use various chants and names associated with the language
in its role of father-language. No other differences exist which coincide
with language-aggregate membership, regardless of whether one is look-
ing for markers used by the Indians themselves to classify one another
or looking for more subtle differences the Indians may not be aware of or
choose not to acknowledge.
1 The research reported in this paper was supported by grants from Bryn Mawr College (1974),
the American Philosophical Society (1976), and the National Science Foundation (1978, grant
BNS 77-26295). I am grateful to Dell Hymes and Michael Silverstein for comments on an ear-
lier draft of this paper.
The Fulnio of Brazil have given up their lands several times during the
last three centuries, moving in order to preserve their language and
annual religious ceremony (to which proper use of the language is essen-
tial) as basis of their identity. In contrast, the Guayqueries of islands off
Venezuela seem to have given up their aboriginal language and native
religion so early that no trace remains. They have preserved their iden-
tity on the basis of a special socio-economic structure, as demonstrated
by the fact that a small group who abandoned the special structure are
today indistinguishable from other Venezuelans. Language, together with
religion, has served as separatist and unifying function in the one case,
but not in the other.
It would seem that the fisherfolk of coastal East Sutherland, in the far north
of the Highland Scottish mainland, are simply one more such case of a people
preserving their ethnicity largely through the use of a distinctive language. Up
to at least World War I, the fisherfolk were distinctive in a rich variety of ways.
Their occupation was itself distinctive, and they (especially the men) had a dis-
tinctive style of dress which accompanied it. Because of the way they earned
a living, their diet was different from that of the rest of the population; they
not only ate fish with greater frequency, but they also ate more varieties of fish
than other people, and they had a number of specially-prepared fish dishes
peculiar to themselves. Residentially the fisherfolk were almost wholly segre-
gated. Two of the villages concerned had a separate district where the fishers
lived (“Lower Brora” as opposed to “the upper village” in Brora; “the West End”
as opposed to “the East End” in Golspie); the third, Embo, was a satellite settle-
ment, composed wholly of fisherfolk, to the nonfishing village of Dornoch. All
three of these fisherfolk residential areas were known, unofficially and impo-
litely, as “Fishertown”. The fisherfolk were also entirely endogamous. Before
World War I, marriages between fisherfolk and other groups in the population
were almost unheard-of.
Today nothing remains of these patterns but some residential clustering
(very weak now in Brora and Golspie, quite strong still in Embo). But the for-
mer fisherfolk – who no longer fish at all – are still highly distinctive in terms of
their language: a local variety of Scottish Gaelic is their mother tongue, though
all are also fluent in English. Even younger speakers whose control of Gaelic
is imperfect (“semi-speakers”: Dorian 1973, 1977) explicitly claim Gaelic as
their mother tongue despite the fact that they are more proficient in English.
They define “mother tongue” temporally, as the language used in their homes
in infancy and childhood, and express no doubt as to the claim of Gaelic to
mother-tongue status.
The ethnic identity “fisher” was a severely stigmatized one in East Sutherland
at the time when the current bilinguals were growing up. It is clear that they
understood the connection between speaking Gaelic and being identified as a
“fisher”, since the automaticity of that connection was a factor in the abandon-
ment of Gaelic by some individuals:
I think myself, as the children from Lower Brora got older, they . . . were
ashamed to speak the Gaelic, in case they would be classed as a – a fisher.
(Brora bilingual, 1974)
At the present time the connection is wholly warranted: any fluent speaker of
any variety of East Sutherland Gaelic is “of the fisherfolk”. There are no excep-
tions. Language and a residual ethnicity are perfectly correlated. And apart
from the remaining residential patterns, which are very weak in two of the
villages, there are no other markers of fisherfolk ethnicity.2
The perfect correlation between language and ethnicity works only syn-
chronically, however. As soon as one goes back a bit in time, the picture
becomes complicated by the fact that other segments of the population also
spoke Gaelic in the past. The crofters, sub-subsistence agriculturalists who
lived in the country districts around the villages we are concerned with, were
still fairly strongly Gaelic-speaking into the period between the two World
Wars; the last Gaelic-speaking crofters died only in the 196os. Earlier still,
2 In a few homes, some old-style fisherfolk dishes are prepared, but even this small difference
in cuisine is not general, and nowadays people of fisherfolk descent eat no more fish than
does the rest of the population.
before World War I, some Gaelic speakers remained even in the villages them-
selves. Yet despite the presence of other Gaelic-speaking groups, informants
continue to talk as though Gaelic were important to fisherfolk ethnicity in
the early years of this century. The information one gleans can be confusing
and contradictory. In one interview about the roots of the prejudice against
“fishers”, a septuagenarian Golspie native, monolingual in English, identified
the fisherfolk’s Gaelic as a possible source of hostility, and then in the next
breath denied it, when I pressed the issue, on the very reasonable grounds that
other people, towards whom there was no prejudice, spoke Gaelic, too:
English monolingual: I don’t know just what it was. Was it people despised
them for their Gaelic, and then some of them weren’t very clean, y’ know.
Some of them. I don’t know just what it was. . . .
Investigator [a few seconds later]: You say they might have been despised
for their Gaelic.
English monolingual: Oh, no, I don’t think for the Gaelic. I don’t think so,
because a lot of people that I knew in the East End had Gaelic, y’ see? My
own mother had it, and a lot more but her had it. No, it wasn’t that.
(Golspie monolingual, 1976)
After bringing up fisherfolk Gaelic on her own, without any prompting from
me, she then categorically denied her own suggestion when confronted with it.
The confusion experienced by this informant clears away, however, if one
accepts that she was completely correct in identifying a linguistic component
to fisherfolk ethnicity, and also accepts that that linguistic component did not
consist of the Gaelic language as such. That is, language can be an important
part of ethnicity without taking the form of the presence or absence of some
particular language or dialect. In East Sutherland this becomes clear in a dia-
chronic view of the situation. As far back as 1897, the fisherfolk of the area were
being singled out for a distinctive linguistic behavior. The writer quoted below
was a Gaelic scholar who was comparing the relative purity, in the sense of lack
of adulteration by English loanwords, of the eastern, northern (Reay Country)
and western (Assynt) dialects of Sutherland:
The fishing communities were out of line linguistically, not in the sense that
they spoke a language which the rest of the population did not speak, but
in the sense that they spoke it differently (in this case, better). Here we have
what can be recognized as a first form of linguistic lag.
Correlated with the relative excellence of the fisherfolk in Gaelic was a rel-
ative weakness in English. The expansion of the English language into East
Sutherland was evident by 1841, the date of the second Statistical account of
Scotland. The Statistical accounts were sketches of each parish in the country,
in which the parish minister dealt with a variety of subjects such as topography,
climate, population, economics, and education. The first Statistical account of
Scotland appeared in 1793, and there was no mention whatever of language by
the ministers of the three parishes where Brora, Golspie, and Embo are located.
The second Statistical account, published 48 years later, contained comment
by all three of the ministers from these same parishes on the rise of English;
two of the three even predicted the death of Gaelic in the area, so vigorous was
the expansion of English. The reasons for this need not concern us here (see
Dorian 1978: 6–10 for the historical and social causes). We can take the com-
ment of the Golspie parish minister as typical:
Forty years ago, the Gaelic was the language generally spoken in the par-
ish. But . . . that language is now fast on the decline; and among the young
there is now hardly an individual who does not understand and speak
English (Statistical account of Sutherlandshire 1841: 35).
The Golspie minister might, like Gunn, have excepted the fishing communi-
ties, since Gaelic monolingualism was still to be found among the fisherfolk
into the twentieth century. An Embo septuagenarian said in 1976: “My mother
couldn’t speak English. Couldn’t speak any English. Very few words.” Pre-school
monolingualism also lingered late, most especially in Embo. Here are two
Embo bilinguals in their mid-sixties discussing the subject in 1978:
The last instances of pre-school monolingualism in Embo date from the 1940s,
a full century after the Golspie minister asserted that bilingualism was the
norm. Here again the fisherfolk were out of line linguistically, continuing to
speak only Gaelic when all other segments of the population knew English, or
both English and Gaelic. This represents a second form of linguistic lag. It is
undoubtedly as early as the “Gaelic purity” lag; both may be assumed to have
begun in the second half of the nineteenth century.
Naturally enough, the second form of lag led to deficiencies in English rela-
tive to the rest of the population when the fisherfolk did become bilingual:
And, their English was very much poorer than ours. . . . [By-name of a
fisher woman], she used to come into the shop, and she used – ‘How
much are your tuppenny sponges?’3 Y’ see? . . . They would hear these
[English] words, y’ see, and – and they didn’t really know. (Golspie mono-
lingual, 1976)
Many present-day bilinguals tell amusing stories about the difficulties they had
with English as youngsters and laugh at the “howlers” that they came out with
through insufficient experience with English. One could even acquire a tempo-
rary by-name thanks to an English blunder. One Golspie bilingual, now a sep-
tuagenarian whose English is fluent, was for years called “That’s-a-dog-of-me”
by a local shopkeeper because of the Gaelic-patterned English she mustered
as a young girl to claim the dog that the shopkeeper was trying to chase out of
his shop.
Again we see that the fisherfolk were linguistically distinct, this time for
their flawed English when all the rest of the population was thoroughly pro-
ficient in English. This represents a third form of linguistic lag. Finally, at the
present day, a fourth form of linguistic lag appears in an anachronistic Gaelic-
English bilingualism which characterizes the fisherfolk when all other seg-
ments of the population are monolingual in English. The pattern seems clear.
For a hundred years or more, the East Sutherland fisherfolk have had a lin-
guistic component to their ethnic identity, but it has not consisted of the use
of any particular language or dialect as such until very recently. Rather, their
3 Taking “tuppeny” as a kind of sponge, but not knowing that it was itself an answer to the
question, “How much?”
Gaelic of the late nineteenth century East Sutherland fisherfolk was superior
to the adulterated Gaelic of the surrounding population. Judging by the invidi-
ous comparisons I have heard among Gaelic speakers in Sutherland and in
other parts of Highland Scotland, it is safe to say that the non-fishers of East
Sutherland would themselves have agreed with Gunn; the number of English
loanwords in one’s Gaelic has long been a sensitive matter among Gaels. On
the other hand, the “broken” English of the imperfectly bilingual fisherfolk dur-
ing the early twentieth century was certainly negatively valued and subject to
much criticism. The positive and the negative lag were equally distinctive for
fisherfolk identity.
Giles, Bourhis, and Taylor (1977: 325) make what can be taken to be a “ classic”
statement about language and ethnicity:
To this formulation I would add, in the light of the East Sutherland fisherfolk
case, that an ethnic group may also manifest its distinctiveness from others by
means of a linguistic behavior which may be popularly perceived in terms of a
language, but can be seen in diachronic perspective to consist of a time differ-
ential: a persistent lag in linguistic habits as compared with the linguistic hab-
its of neighboring groups. The precise nature of the out-of-step behavior may
vary over time, as it has done in Sutherland, but the constant is the presence
of some difference in linguistic behavior, and not a particular different lan-
guage or dialect as such. The ethnic marker is rather the lag than the language.
The traditional notion of the form the “language” component takes, in the rela-
tionship between language and ethnicity, needs to be broadened to allow for
such cases.
References
Chapman, A. J., J. R. Smith, & H. C. Foot. 1977. Language, humor, and intergroup rela-
tions. In H. Giles (ed.), Language, ethnicity and intergroup relations. N.Y.: Academic
Press. 137–69.
De Vos, George. 1975. Ethnic pluralism: Conflict and accommodation. In G. De Vos &
L. Romanucci-Ross (eds.), Ethnic identity: Cultural continuities and change. Palo
Alto, California: Mayfield Publishing Co.
Dorian, Nancy C. 1973. Grammatical change in a dying dialect. Language 49, 413–38.
———. 1977. The problem of the semi-speaker in language death. International Journal
of the Sociology of Language 12, 23–32.
———. 1978. East Sutherland Gaelic. Dublin: Dublin Institute for Advanced Studies.
Fishman, Joshua A. 1977. Language and ethnicity. In H. Giles (ed.), Language, ethnicity
and intergroup relations. N.Y.: Academic Press. 15–57.
Giles, H., R. Y. Bourhis, & D. M. Taylor. 1977. Towards a theory of language in ethnic
group relations. In H. Giles (ed.), Language, ethnicity and intergroup relations. N.Y.:
Academic Press. 307–48.
Gunn, A. & J. MacKay, (eds.). 1897. Sutherland and the Reay Country. Glasgow: John
MacKay.
Hohenthal, W. D., & T. McCorkle. 1955. The problem of aboriginal persistence.
Southwestern Journal of Anthropology 11, 288–300.
Hymes, D. 1966. Two types of linguistic relativity (with examples from Amerindian eth-
nography). In W. Bright (ed.), Sociolinguistics. The Hague: Mouton. 114–67.
Jackson, J. 1974. Language identity of the Colombian Vaupes Indians. In R. Bauman &
J. Sherzer (eds.), Explorations in the ethnography of speaking. Cambridge: Cambridge
University Press. 50–64.
Petersen, William. 1975. On the subnations of western Europe. In N. Glazer and
D. P. Moynihan (eds.), Ethnicity: Theory and experience. Cambridge, Mass.: Harvard
University Press. 177–208.
The statistical account of Sutherlandshire, by the ministers of the respective parishes.
(1841). n.p.
There are a good many possible outcomes of extensive language contact, and
by no means all of them include language loss, whether on the community
level or on the individual level. By definition, though, we are interested on this
occasion in the language contact situations which do result in same form of
loss, and my remarks will be confined to such cases. But because there is some
tendency in North American cultural and intellectual life to assume that exten-
sive language contact produces one form or another of loss, I feel obliged to
stress that this needn’t be so, even though I will be dwelling on the cases where
it is so.
One common result of extensive language contact on the community-wide
level is language shift, the gradual displacement of one language by another in
the lives of the community members. This occurs most typically where there
is a sharp difference in prestige and in levels of official support for the two
(or more) languages concerned. Where such differences in prestige and official
support exist, there are usually also marked differences in the utility of the
two (or more) languages for the speakers, (To simplify terminological matters,
I will speak henceforth of two languages and of bilingualism, but my remarks
should be understood to apply also to situations in which more than two lan-
guages are spoken and to multilingualism.) The two “classic” settings in which
this phenomenon has been relatively well studied are the indigenous minority
language and the transplanted immigrant language.
On the level of the individual, the result of extensive language contact may
also be a complete shift, over a lifetime or over a briefer transition period; but
at least as common, and perhaps more so, is a partial shift or even a partial
merger, so that at least one of the two languages does not retain its full com-
plement of functions or perhaps even of forms. Sometimes neither language
retains its full complement of functions or of forms.
To take up first the case of shift on a community-wide level, there are several
questions in connection with that phenomenon which are of potential inter-
est to us here. It seems usual for members of any given speech community to
have very little difficulty ranking two coexisting languages in terms of pres-
tige. There are societal attitudes toward specific languages and to their native
speakers. How are these attitudes formed? Are they usually local in origin and
1 One can also have a positive language orientation in a positive situation, but the concept of
language loyalty is rarely introduced in such cases since the assumption is that loyalty to the
language in question will be high. In negative situations there can be no such assumption,
and where loyalty is nonetheless encountered, it is consequently worthy of comment.
of the United States (and, of course, Great Britain before it) and of the Soviet
Union. Occasionally, a politically dominated area possesses a cultural tradition
of such strength that its language survives conquest and perhaps even prevails
over the language of the conquering power; Greek maintained a strong position
under the domination of the Romans, and the Norsemen adopted a Romance
tongue in France. Various compromises are also possible, such as the Arabic-
and Persian-encrusted “Turkish” of the Ottoman Empire, or the Normanized
lexicon of resurgent English in the thirteenth and fourteenth centuries.
Two factors which have undoubtedly played a role in the shift characteris-
tic of historic conquest situations have been identified in contemporary cases
of impending language extinction as well: pragmatism, and a cultural stance
which either does or does not favor language maintenance. The two may or
may not be related. That is, pragmatism is sometimes a part of a cultural stance
which favors language shift, but in other cases it may be a separate issue and
may favor bilingualism rather than language shift.
Examples abound of communities where the local language was delib-
erately not transmitted to the children because it had no practical value in
the national or even local setting. That is, no jobs and no social advantages
accrued through mastery of the language in question. A native speaker of
East Sutherland Gaelic, a moribund variety of Scottish Gaelic, told me that
she hadn’t taught the language to her children because “Gaelic’s no use to you
through the world.” Denison reports for Sauris, an originally German-speaking
area in northeastern Alpine Italy, a pattern of parental use of German among
themselves but Italian to their children:
The reason given by informants for this use of Italian is in almost all cases
the desire to ease the path of their children at school: a few have men-
tioned the general usefulness of Italian as compared with the other lan-
guages (the local German dialect and Friulian) . . . (Denison 1971: 166–67).
brought about radically different economic and social conditions. Before 1974,
“socio-economic conditions were such that there was no advantage identified
with the learning of Spanish for those who did not plan to migrate”; but the
greater utility of Spanish for jobs within Oaxaca, since the development of a
port near Juchitan, has already produced a weakening of language loyalty to
Isthmus Zapotec in Juchitan in the few years since 1974 (Hinojosa 1980). Susan
Gal’s masterful study of the language-shift processes in Oberwart, Austria,
shows that the current shift to German on the part of the originally Hungarian-
speaking segment of the population had its counterpart earlier, when the dis-
trict was part of Hungary, in a tendency among some German speakers to shift
toward Hungarian (Gal 1979: 42). In short, people know very well on which side
their linguistic bread is buttered.
For the transplanted immigrant, there is often a major reversal in the utility
of the mother tongue (unless, of course, he emigrates to a colony or former
colony where his mother tongue is still spoken). In the new environment, his
mother tongue will be useful only insofar as he interacts with other immi-
grants from the same home country. Otherwise, a quite different language will
become much more useful. Karttunen considers the homogenous settlements
peopled by Scandinavian immigrant groups in the northern forest and plains
areas of the United States and Canada to be particularly favorable for language
maintenance: rural immigrants there continued to speak the languages of
their home countries longer than did immigrants in urban centers. But even in
these ethnically homogeneous settlements, language shift was soon underway:
“For a while, such settlements had rather limited contact with English, and it
was possible to remain monolingual in North America. Once contact began,
however, English rather quickly took over.” (1977: 174).
In such cases as these, it’s clear that pragmatism is a major factor. Language
loyalty persists as long as the economic and social circumstanees are condu-
cive to it, but if some other language proves to have greater value, a shift to that
other language begins.
Pragmatism is not the only value operating, however. There seem also to
be cultural dispositions which do or do not favor language maintenance, or
cultural-historical dispositions. Hymes drew attention more than a decade
ago to a study by Hohenthal and McCorkle (1955) which contrasted two South
American Indian groups in terms of retention of language (and identity). One
group, the Fulnio of Brazil,
have given up their lands several times during the last three centuries,
moving in order to preserve their language and annual religious cer-
emony (to which proper use of the language is essential) as the basis of
their identity. (Hymes 1966: 126)
The other group, the Guayqueries, who live on islands off the coast of
Venezuela, apparently abandoned their language and their pre-Christian reli-
gion very early, since no trace of the original language or religion survives. The
Guayqueries maintain their identity by way of a particular socio-economic
structure; in the case of one small group of the Guayqueries who gave up that
social structure, their distinctiveness was lost, and they became indistinguish-
able from other Venezuelans. Hymes comments: “Language, together with reli-
gion, has served a separatist and unifying function in the one case, but not in
the other” (ibid.). McLendon, in studying a Californian Indian language, found
that linguistically exogamous marriages coincided with a thoroughly prag-
matic approach to language learning and unlearning. Eastern Pomo Indians
acquired whatever languages might be spoken in their homes very readily but
forgot them just as readily if the use for them passed (via the death of a grand-
parent, for example). Characteristic of the Eastern Porno, also, is “a matter-of-
fact valuation of language and lack of romanticism about or idealization of
only one language” (McLendon 1978: 143). Clearly this is not a cultural stance
which fosters a language-centered ethnic symbolism, and McLendon suggests
that English entered the Eastern Pomo speech community on the same prag-
matic basis as other useful languages in the past, but outweighed all the rest,
ultimately, in its long-run utility – including Eastern Pomo itself, which is now
approaching extinction (op. cit.: 146).
The strongest statement on the role of cultural dispositions is probably to
be found in a recent paper of Eric Hamp’s. He points out that “cultures . . . have
as a part of their complex of traits a specific view or policy concerning integ-
rity of the culture and the role of diversity” with “important consequences for
language contact, and the more so to a dramatic degree in the case of small
communities” (1978: 160). His most striking case in point is the very dissimi-
lar fates of Albanian· language enclaves in Greece and in Italy. In Italy, where
a tradition of cultivating and valuing localism (including local dialects) pre-
vails, the Albanian enclaves are preserving their language; while in Greece,
with an ethnocentric reverence for all things Greek as the dominant feature,
the Albanian areas – once quite extensive – are losing their distinctive lan-
guage. Hamp maintains that it is entirely in keeping with, even predictable
from, these different cultural traditions, that Italian immigrants in the United
States usually give up Italian after the first generation in favor of the language
of the new locality, while “immigrant Greeks in the United States will cling to
their language to the third generation” (op. cit.: 161–62). As a veteran of two
and a half years of “Greek school”, I can testify to the determination of Greek
Americans to maintain their language; the very numerous Italian Americans
in my home town, on the other hand, provided no language-school classes for
their children at all.
So very many languages are dying, during our time, in so very many locali-
ties, especially in the face of a vast expansion of a relatively few languages of
enormous political and cultural potency like English, Russian, Spanish, and
Mandarin Chinese, that one might well wonder whether even the most resis-
tant linguacentric cultural stance would be enough to preserve a language
with limited geographical, cultural, or demographic scope today. The dying
languages are typically within the boundaries of a country in which a “major”
language is spoken. That is, English is not displacing indigenous languages in
India or Africa (see Fishman 1977: 114), but it is doing so, rapidly, in Anglophone
Canada, the United States, and Australia; Russian is completely displacing a
good many languages within the Soviet Union (Lewis 1972), but not in the rest
of Eastern Europe. This phenomenon highlights again the political aspect of
language maintenance. And, indeed, where threatened languages have made a
comeback, it has usually been in connection with a sharp rise in nationalistic
sentiment. Thus, there were some notable success stories in Europe as national-
ism gained strength: Finnish successfully replaced the prestige language which
had threatened to displace it in its own homeland (that is, Swedish), and Czech
ceased to lose ground to German with the dissolution of the Austro-Hungarian
Empire and the establishment of an independent Czechoslovakia (Ellis and
mac a’ Ghobhainn 1971). German itself was once actually endangered by a
fashion for French which spread beyond the aristocracy and beyond even the
middle class, threatening to become truly “popular”. Historians of the German
language often quote Voltaire, who claimed that during his rather extensive
experience of eighteenth-century Germany (he was a guest at the court of
Frederick the Great), he needed German only for travelling: “Je me trouve içi en
France. On ne parle que notre langue. L’allemand est pour les soldats et pour
les chevaux: il n’est nécessaire que pour la route” (quoted in Waterman 1966:
138). But the German language survived to emerge later in the eighteenth and
nineteenth centuries as a stable standard (and eventually national) language.
Hebrew, too, is the great exception to the requirement that a reviving language
must have a certain numerical strength (of as yet undetermined size) in the
population in order to succeed: for, although it seems clear that large numbers
of speakers will not save a language (cf. Irish, which gave way to English during
the eighteenth and nineteenth centuries, despite the fact that it boasted great
numerical superiority at the start of that period over English, the language of
a relatively small elite; see Macnamara 1971), it also seems that a comeback on
the part of a threatened language rarely occurs without a substantial demo-
graphic base, even if that base is largely among the peasantry.
If we now look at all these factors – the utility and prestige of a language,
cultural values which do or do not support identification and maintenance
Moving now from the very general to the very particular, I would like to draw
on the in-depth study which I’ve been making of dying East Sutherland Gaelic
(ESG), on the extreme northeast coast of mainland Scotland, for information
about attrition in the language extinction process. Among the 13 imperfect
speakers4 of ESG whom I’ve tested since 1974, three different acquisitional his-
tories are identifiable. Most of the imperfect speakers have always been exactly
that: they lacked sufficient exposure or motivation to become fully proficient
speakers of ESG and have never at any time spoken a grammatically normal
form of the language. One imperfect speaker has a very different history, how-
ever: she was fully fluent (and almost certainly Gaelic-dominant, to judge by
other members of her family whom I know) up to the age of 18, when she left
her home village for the Lowlands. There she ultimately married a monolin-
gual English speaker and raised her family. At the time that I tested her (1974),
she had been away from the home village for 40 years with very little oppor-
tunity to use Gaelic, apart from vacation trips home roughly once a year for
about two weeks at a time. Her passive knowledge of Gaelic was virtually per-
fect (as in fact is that of all of the imperfect speakers, including those who have
never been fully profieient),5 and her phonology was essentially normal for
her age group, but her spoken Gaelic was halting and her grammar fairly devi-
ant. Two other imperfect speakers, a brother and sister pair, have yet another
acquisitional history, in that they were fully f1uent child speakers of ESG up to
five or six years of age, when they entered school. It is abundantly clear from
family anecdotes that they were Gaelic-dominant at school entrance, and the
most convincing evidence of that fact today is that their phonology is unusu-
ally good compared to that of other imperfect speakers. In this last respect,
they resemble the exiled speaker mentioned above.
For comparison with the imperfect speakers, eight fully proficient speakers
were also tested: four “older fluent speakers” relatively close to the c onservative
norm for the dialect[,] . . . established on the basis of the Gaelic used by the
oldest and best f1uent speakers available to me in the early 1960s when my
4 I use the term “imperfect speaker” throughout this paper rather than the term “semi-speaker”,
which I have frequently used in other papers (e.g., Dorian 1977, 1978b) because the sample of
speakers reported on here includes one individual with a different acquisitional history and a
different level of proficiency from the semi-speakers on whom I have previously reported. . . .
5 The perfect passive bilingualism of the imperfect speakers is apparent from: their flawless
interactions with fluent speakers; their ability to provide instant English translations of any-
thing said in ESG; their very evident comprehension of jokes, teasing, and rapid-fire banter;
and so forth. They also state that they understand everything said in their presence in the
local Gaelic.
study of ESG began (see Dorian 1978a); and four “younger f1uent speakers”
whose Gaelic departed from the conservative norm in quite a few respects
(see Dorian 1973, for example), but in ways and degrees not generally appar-
ent to the community. The tests consisted of sentences in English presented
for translation into Gaelic, a task made less artificial and less difficult by the
following factors: translation into and out of Gaelic is a frequent activity in
the community due to mixed networks of monolingual and bilingual kin and
friends; the tests were conducted in the speakers’ own homes, in a friendly
atmosphere and with much encouragement; I had worked in the community
for eleven years before these tests were undertaken and was well known to
almost all of the speakers, and very well known to a good many.
Very much by herself, in terms of test results, was the exiled speaker. She
was neither fish nor fowl, linguistically speaking. She was markedly deficient
in morphophonology, morphology, and syntax by comparison with the two
groups of fully proficient speakers, yet she was notably better than the other
imperfect speakers in some (though not all) respects. Some examples: (1) she
chose the conservative synthetic form of the first-person singular conditional
verb 100% of the time, as did seven out of the eight fully proficient speakers,
but none of the other imperfect speakers did (only one of whom ever used it);
(2) she and two other imperfect speakers were the only speakers (out of 16
people tested on this point) to use fewer conservative placements of a pronoun
object than nonconservative placements; (3) like all other imperfect speakers,
but only one fully proficient speaker, she lacked the imperative plural mor-
pheme entirely; (4) in the negative imperative, she used none of the most con-
servative possible forms, but also none of the least conservative forms favored
by most of the imperfect speakers, producing only intermediate forms (except
for one ambiguous form which may either have been an out-and-out mistake
of a kind made otherwise only by imperfect speakers or have been a slightly
deviant rendering of a conservative form); (5) all of her tenses were intact, a
record matched by only three other imperfect speakers; (6) she controlled all of
the subordinating conjunctions except the most infrequent ones (e.g.; ‘unless’,
‘although . . . not’), which was true of no other imperfect speaker: (7) in a test
for retention of irregular noun plurals, she gave exactly as many analogical plu-
ral formations as conservatively irregular ones, a much weaker performance
than any fully proficient speaker, and also weaker than four out of eight other
imperfect speakers, but stronger than the remaining four tested on this point.
Since this speaker’s acquisitional history is unique in my sample of subjects,
no firm conclusions can be drawn about what remains and what disappears
in the speech of the ESG-speaking individual who is totally fluent in late ado-
lescence but has very little subsequent use for the language. All the same, it is
clear from her test results that she has suffered marked attrition compared to
her peer group who remained in the home community. Her younger sister and
her same-age brother-in-law were among the younger fluent speakers tested.
They outperformed her on every measure, since even where her morphology
and morphophonology were intact, as in tense formation, she made mistakes
which they did not make (such as the choice of the wrong root-form on which
to build the tense-form itself).
Although I have not had the opportunity to broaden my sample of exiles
who have suffered attrition,6 there is a wealth of anecdotal material available
in East Sutherland on the subject. The reason for this abundance of anecdotal
material is itself of interest: it seems that there are enormous individual dif-
ferences in the retention of the language among exiles, both in Britain and
overseas, and the differences are so great that the community is at a loss to
account for them except by attributing them to differences in loyalty to the
home community and to its language. Since the sole remaining group which
speaks Gaelic in East Sutherland today is a stigmatized ethnic group, willing-
ness to use Gaelic is valued as a sign of group solidarity, and unwillingness
to use Gaelic is resented as a sign of rejection of one’s origins and, hence, of
social climbing. For this reason, a great deal of attention is paid to the state
of any formerly fluent exile’s Gaelic if he or she comes home for a visit. There
are, as I’ve noted, vast individual differences in retention. Families which
have siblings who have came home after thirty or forty years in Australia with
their Gaelic essentially intact boast of this for years afterwards, and families
whose siblings are unable to converse in Gaelic after an equal length of time
abroad make excuses miserably (“There was no one else from our part of the
Highlands in his area”, and “He was married to an English speaker, you know.”).
Extremely harsh condemnation is the lot of individuals who claim no longer to
be able to speak Gaelic, or whose Gaelic in fact measures up poorly, after only
6 In actual fact there were two other exiles in my test sample, ESG speakers who had been liv-
ing in London for a great many years. They were married to each other, however, and used
Gaelic habitually in the home. I have often stayed with them in London and have had ample
opportunity to observe that not only they but almost all adult visitors to their home are
habitual Gaelic speakers. If one avoids looking out the window, there is nothing to remind
one that one is not in East Sutherland. These two London residents have some relatively
non-conservative usages, and some skewing of lexicon and idiom, but they are really fully
proficient speakers of ESG and hence not useful subjects for the study of attrition despite
their removal over a long period of time from the home community.
a relatively short time away – say, two to five years in an English-speaking area.
The ESG speech community members emphatically do not believe that the
performance differences they have observed over the years are due to actual
differences in retention. They attribute all of the performance differences to
attitude differences, that is, to differential language loyalty.
As an outsider, I disagree. It seems evident to me, after witnessing from the
inside of several different households the return of an exile-sibling, that there
are real individual differences in the capacity to retain a once-dominant lan-
guage in conditions in which there is no opportunity to speak it. I don’t doubt
that same individuals work at it in some fashion; as with child learners, there
are apparently some exiles who literally “practice”, whether by carrying on
internal monologues, by talking to themselves when working alone, by praying
regularly in the original home language, or by cultivating a repertoire of songs
in it. (I heard tell of one man in Australia who had kept his Gaelic alive at least
partly through songs.) But some people apparently don’t have to work at it very
much. The language just stays with them, through no special effort on their
part. Other people, like the exile whom I tested, retain full receptive knowl-
edge of the language, but lose a good deal of their active control. Whether
they could prevent this by special effort is entirely unknown. It may be as it is
among adults with adopting the accent of the place you’ve gone to live: some
people can’t help it, and others couldn’t do it if they tried.
The issue of whether retention can be achieved by any individual through
the exercise of will, or with the spur of highly positive attitudes, is one which
interests me intensely because of the vehement feelings which attach to the
Gaelic performance of returned exiles in East Sutherland. It should not be dif-
ficult to approach the matter objectively. Many American towns have some
exiles from the smaller language groups of Europe – for example, Gaels, or
Frisians, or Friulians – who have little opportunity to speak their original home
language because they are too few and too scattered. Questionnaires could be
devised to evaluate the strength of their attachment to the mother tongue and
the degree of integration into any accessible exile community, as well as self-
evaluation of the use they make of the mother tongue and their proficiency
in it (both for ordinary purposes and for inner speech). A single investigator
could administer such questionnaires and also some measures of vocabulary
retention and productive abilities to a sample of just 50 to 100 exiles of identi-
cal mother tongue in various localities, and we would very quickly have some
idea of the range of individual variation in these matters. Ideally, one would
like also to administer a foreign language aptitude test, in order to determine
whether language retention capacity is linked to language learning abilities.
spoke together [in Gaelic]. You know, the three of us. Because – y’know,
we just liked speaking.
I’ve asked her the same question more than once, several years apart, and her
answer was the same each time: that spending time away from the community,
in company with other exiles, led her to value her mother tongue more and to
try to hold on to it. Furthermore, since her return home to marry, she has lived
for many years immediately next door to an older Gaelic-speaking kinswoman
and her husband, with other Gaelic-speaking relatives a little farther down the
street. She interacts regularly with such of these relatives as are still alive and
with still others a little farther away in the village, and fully acknowledges, both
publicly and privately, her membership in the stigmatized Gaelic-speaking
group. That is, she does not make the attempt to “pass”, socially, by suppressing
aspects of her identity (such as knowledge of Gaelic) which identify her as a
member of the group in question.
The issue of emotional distance seems worthy of special attention because
it has been found to operate in second language learning (cf. Schumann 1978;
Fillmore 1979). But where retention of the mother tongue is concerned (and
almost all of the imperfect speakers consider Gaelic their mother tongue, inter-
estingly enough, regardless of the level of proficiency they display), I find that
not only the stronger imperfect speakers, but also the weaker imperfect speak-
ers, are relatively loyal to Gaelic: favorable to the language in their attitudes,
positive in their assessment of its beauty and richness, desirous of its being
taught in the schools, and so forth. That is, a positive attitude toward Gaelic
does not characterize only people who have succeeded in retaining it quite
well, but rather it characterizes all imperfect speakers, even the very weak ones.
It is probably a minimum condition for some retention of a language in what
are clearly the final decades of that language’s existence; but it is not in any way
predictive of the degree of success in maintaining control of the language. This
contrasts sharply with studies of second language acquisition, where attitude
often is predictive of success, both inside and outside the classroom. (In addi-
tion to Schumann and Fillmore, cited above, see Gardner and Lambert 1972,
and the extensive bibliography which it contains, and Schumann 1975). One
possible explanation for the difference between second language learning and
mother tongue retention in this respect is the total receptive bilingualism of
the imperfect mother tongue speakers, to which I’ve already referred above.
They interact with fully fluent speakers superbly by drawing on their perfect
comprehension and their knowledge of the sociolinguistic norms of the com-
munity (i.e., when to speak, when to be silent, whether to phrase something
as a statement or a quotation, etc.), and by using their sometimes very limited
productive skills judiciously. Since the second language learner normally has
everything to learn, receptive as well as productive skills and sociolinguistic
norms, the forward thrust provided by emotional involvement (integrative
motivation; see Lambert 1967: 102) may be crucial to successful mastery. And,
of course, the second language learner does not in any case have the option
of leaning on passive and sociolinguistic skills to lessen the need for active
control.
One of the more suggestive findings of my Gaelic language-death studies for
linguistic attrition is that the imperfect speakers in my sample show certain
kinds of reductive phenomena in common in their Gaelic, regardless of acqui-
sitional history. For example: a greater or lesser use of analogically regularized
allomorphs in place of irregular allomorphs; complete loss of morphemes
that are already showing weakness in the fully f1uent population’s Gaelic; loss
of “inventory”, as in both the examples already mentioned and also in loss of
vocabulary from both open and closed classes. On the other hand, my one-
person sample of the “formerly f1uent adult” category showed the following
two differences from all of the other imperfect speakers: she did not show any
certain evidence of loss of syntactic options except by loss of inventory (she
retained forms appropriate to her age group for the two ESG passive structures,
for example); and she did not show a tendency to make synthetic structures
analytic, as all the other imperfect speakers quite strongly did. Thus, there
seem to be both some important similarities and some important differences
between the once-fluent adult who ceases to use a language and the speaker
who has never actually achieved full proficiency. Analogical regularization is
one of the former; resistance to analysis of synthetic forms is one of the latter.
Obviously we need much richer evidence before we speculate much on why
this should be so – most of all, we need to know whether these tendencies hold
up over additional test cases.
One other phenomenon which deserves mention is the amount of admix-
ture or interference that shows up in the imperfect speaker’s weaker language.
Like the second language learner, the imperfect speaker of a dying language
controls some other language much better than the home language he has
either partially forgotten or never fully mastered (or both). It has surprised me,
in working with ESG, how little structural interference there is from English
in the speech of imperfect speakers, all of whom are, of course, fully f1uent
in English. There is a great deal of lexical interference, and some of the weak-
est imperfect speakers introduce unnatural syntactic patterns from English
(perhaps a good deal of this promoted by the artificiality of the translation
tests). But I’m struck by the fact that, although both ESG and English form most
plurals by suffixation, and imperfect speakers greatly extend, through analogy,
References
Costello, John. 1978. “Syntactic change and second language acquisition: The case for
Pennsylvania German.” Linguistics, 213:29–50.
Denison, N. 1971. “Some observations on language variety and plurilingualism.” In:
Edwin Ardener, ed., Social anthropology and language. London: Tavistock
Publications.
Dorian, Nancy C. 1973. “Grammatical change in a dying dialect.” Language, 49:413–38.
———. 1977. “The problem of the semi-speaker in language death.” International
Journal of the Sociology of Language, 12:23–32.
———. 1978a. East Sutherland Gaelic. Dublin: Dublin Institute for Advanced Studies.
———. 1978b. “The fate of morphological complexity in language death: Evidence
from East Sutherland Gaelic”. Language, 54:590–609.
Ellis, Peter Berresford, and Seumas mac a’ Ghobhainn. 1971. The problem of language
revival. Inverness: Club Leabhar Limited.
Fillmore, Lily Wong. 1979. “Individual differences in second language acquisition.”
In: Charles L. Fillmore, Daniel Kempler, and William S.-Y. Wang, eds. Individual
differences in language ability and language behavior. New York: Academic Press.
Fishman, Joshua A. 1977. “The spread of English as a new perspective for the study of
‘Language maintenance and language shift’.” In: Joshua A. Fishman, Robert L.
Cooper, and Andrew W. Conrad, [eds]. The spread of English. Rowley, Mass.:
Newbury House. pp. 108–33.
Gal, Susan. 1979. Language shift. New York: Academic Press.
Gardner, R. C., and W. E. Lambert. 1972. Attitudes and motivation in second-language
learning. Rowley, Mass.: Newbury House.
Hamp, Eric. 1978. “Problems of multilingualism in small linguistic communities.” In:
James E. Alatis, ed. International dimensions of bilingual education. Georgetown:
Georgetown University Press. pp. 155–64.
Hinojosa, Maria de la Paz. 1980. “The collapse of the Zapotec vowel system.” Penn
Review of Linguistics, 4:28–39.
Hohenthal, W. D., and T. McCorkle. 1955. “The problem of aboriginal persistence.”
Southwestern Journal of Anthropology, 11:288–300.
Hymes, Dell. 1966. “Two types of linguistic relativity (with examples from Amerindian
ethnography).” In: W. Bright, ed. Sociolinguistics. The Hague: Mouton.
Karttunen, Frances. 1977. “Finnish in America: A case study in monogenerational lan-
guage change.” In: Ben G. Blount and Mary Sanches, eds. Sociocultural dimensions of
language change. New York: Academic Press.
Lambert, Wallace E. 1967. “Social psychology of bilingualism.” Journal of Social Issues,
23:91–109.
Lewis, E. Glyn. 1972. Multilingualism in the Soviet Union. The Hague: Mouton.
Macnamara, John. 1971. “Successes and failures in the movement for the restoration of
Irish.” In: Joan Rubin and Bjorn H. Jernudd, eds. Can language be planned? Hawaii:
University of Hawaii Press.
McLendon, Sally. 1978. “How languages die: A social history of unstable bilingualism
among the Eastern Pomo.” In: Margaret Langdon, Shirley Silver, and Kathryn Klar,
eds. American Indian and Indo-European studies. The Hague: Mouton. pp. 137–50.
Rankin, Robert L. 1978. “The unmarking of Quapaw phonology: A study of language
death.” Kansas Working Papers in Linguistics, 3:45–52.
Richards, Jack C. 1973. “Error analysis and second language strategies.” In: John W. Oller,
Jr., and Jack C. Richards, eds. Focus on the learner. Rowley, Mass.: Newbury House.
pp. 114–35.
Schumann, John. 1975. “Affective factors and the problem of age in second language
acquisition.” Language Learning, 25:209–35.
In the fifteen years during which I was constantly visiting the East Sutherland
district of the northeast Scottish mainland or living there, actively studying
the dying Scottish Gaelic dialect of the region and clearly deeply engaged by
its distinctive history and unique flavor, I was asked many times whether my
activities or indeed anyone else’s could make any difference to the ultimate
fate of the dialect. But I was never asked that question by a native of the region.
Both academics and laypeople in other parts of the world thought it theo-
retically possible that the ebb tide of linguistic retreat might turn or be turned.
And theoretically of course such a thing is perfectly possible and has in fact
happened in the case of a good many languages. The heroic, near miraculous
example of Hebrew, regaining vernacular status after centuries of more or less
fossilized existence, is always to the fore, and the number of other languages
once threatened but now perfectly secure is sufficient to make the question
reasonable enough. For a compilation of success stories, see EIlis and mac a’
Ghobhainn (1971); for examples of enclaved peoples who have survived against
all odds, sometimes with their languages as part of the persisting identity, see
Castile and Kushner (1981).
Nonetheless, at the risk of casting a dark shadow over a subject already
rendered gloomy by the sheer number of languages known to have died or
acknowledged to be in the gravest danger of doing so very shortly, I wish to deal
here with some of the more knotty problems which pose severe obstacles to
well-intended and even well-funded efforts to promote the survival (whether
in terms of true maintenance or of revival from a barely existing population
base) of threatened languages. I will deal first with the East Sutherland Gaelic
case and then add examples from the Irish experience. My reason for empha-
sizing the negative rather than the positive in this paper is not a wish to deny
the value of language-maintenance programs in general (see the closing sec-
tions of this paper in evidence), but rather an uneasy sense that maintenance
programs are too easily and comfortably invoked as a solution to the decline of
any speech form. The reality, as usual, is more complex and difficult.
1 I am indebted to Mr. Séamus Ó Ciosáin of the Roinn na Seirbhíse Poiblí, Dublin, for his read-
ing of the prepublication form of this paper and for several corrections and suggestions.
Marked political and/or economic change is the scenario most often sug-
gested as favorable for a corresponding change in linguistic fortunes. Examples
are not lacking. A case not only in point but also in progress is Faroese, where
a new degree of relative freedom from political and economic subservience
to Denmark has made possible a resurgence of Faroese linguistic activity, in
schooling, journalistic and creative writing, scholarship, and popular speech
(Wylie and Margolin 1981). The outlook for Faroese is vastly more positive than
it was only fifty years ago, and Faroese ethnic identity has gained notably in
conjunction with the progress made in establishing two forms of the language
as valid “national” written and spoken norms (ibid.).
The question then arises: could not “devolution”, partial or total disengage-
ment from a general British polity in which Scottish and particularly Highland
interests are given disgracefully little attention, create a similarly positive cli-
mate for Gaelic? If the revenues from North Sea oil and whiskey export were to
be kept “at home”, as Scottish Nationalists have often enough urged, could huge
infusions of financial support for Gaelic make a significant difference?
Where Scottish Gaelic generally is concerned, such an outcome is not
impossible, however unlikely. For East Sutherland Gaelic, the answer must be
negative. The reasons are all too numerous. First, East Sutherland Gaelic is an
isolated dialect, a speech island cut off from other dialects of the same lan-
guage. Second, as one might expect in such circumstances, it is an unusual
dialect, quite unlike demographically better represented, less widely separated
dialects in the western part of the country. Third, it is an unwritten dialect,
lacking any tradition for rendering it visually. Fourth, such fluent speakers as
survived even in the 1960s, when I began my work there, were on the whole
poorly educated and elderly, ill suited to take on leadership roles in promoting
or codifying the dialect. Fifth, by reason of the deviance of the dialect and the
lack of a written tradition, well-meaning outsiders would find it difficult to
learn or (if speakers of other dialects) adapt to the local norms and unconge-
nial to promote a dialect so far from the more usual or better known varieties.
Any promotion of Gaelic in East Sutherland would have to mean, and would
have had to mean even in the early 1960s, when the speaker population was still
200 or more, promotion of a more nearly standard form of Gaelic, somewhat in
keeping with dominant western dialects such as those of Skye or Lewis or the
Outer Hebridean islands south of Lewis. In sophisticated populations with a
tradition of literacy and therefore usually some experience of standardization
or of compromise in the general direction of standardization, such an intro-
duction of outside norms might be welcomed, or at least tolerated, as pref-
erable to loss of the language altogether in the local area. In unsophisticated
populations lacking such tradition or experiences, a response of that kind
is unlikely. This is the more true because unwritten, nonstandard minority
. . . (I)f you’re not satisfied with that [the local version of a sentence given
out for translation], go and take your own Gaelic. [From an adult woman
who abandoned an adult Gaelic class because of dialect conflict.]
The effect of these efforts to promote Gaelic by teaching it can be seen overall
to have been alienating rather than reinforcing, and this is a common enough
result where a threatened language displays pronounced dialect differences
and no established standardization tradition exists to temper the alienating
effect of encountering authority figures who attempt to inculcate nonlocal
norms. Languages struggling to survive have often had to make heroic efforts to
bridge dialect differences in creating written forms which could be promoted
without prejudice to one or another region and could therefore enlist the sym-
pathy of the speaker population as a whole. For the ingenious and unusual
solution devised for Faroese, see Wylie and Margolin (1981: 82–94). See Timm
and Kuter (1982: 12–14) for the ongoing difficulties in devising an acceptable
pan-Breton orthography.
For members of societies with relatively long histories of literacy, even
though dialect variation may be pronounced (as in Germany or Italy, for exam-
ple), it can be hard to appreciate the aversion among self-conscious and under-
confident dialect speakers to “strange” forms of “the same” language. Because
of inadequate or absent literacy, experience in relating local forms to other
forms via an intervening standard form to which each can be referred is also
lacking, and the effort involved in working out the equivalences necessary to
easy understanding is correspondingly greater. Thus even a west-coast speaker
of Scottish Gaelic complained to me once that she wished a certain woman in
her neighborhood would just speak English to her instead of insisting on using
a nonlocal (but also western) Gaelic dialect, as it was such a bother trying to
“translate” the other dialect forms into local equivalents at high-enough speed
for easy communication.
In beleaguered speech communities where there is competition from and
heavy pressure in favor of some language of wider currency, it often seems also
to be the case that tolerance of dialect differentiation in the threatened lan-
guage is low. Perhaps the awareness that dominant-language speakers typically
already have a negative attitude to the minority language makes the minority-
language speakers hypersensitive to what seem to be aberrations within the
minority-language varieties. Scottish Gaelic speakers generally are severely
critical of their fellows, in my experience, for such failings as heavy use of
English loanwords or calques of English constructions; they also react rather
vehemently, again in my experience, to dialects of Gaelic which show collapse
of grammatical categories maintained in the most conservative dialects, like-
wise to loss of lexical richness and to loss of phonemic contrasts – this without
regard to (or interest in) the possible long-standing history of such changes for
the dialects in question.
A classic example of interdialect intolerance, comical really because of the
historical absurdity of the attitudes when one is aware of the actual etymolo-
gies involved, is the contempt sometimes expressed by Scottish Gaelic speak-
ers whose dialects use the word coinean to mean ‘rabbit’ when they encounter
speakers whose dialects use the word rabaid instead for the same meaning.
Coinean users heap scorn on the rabaid users, taunting them for merely put-
ting a Gaelic pronunciation on so obviously English a word.
The irony, and the absurdity, in this case, is that coinean is equally an English
loanword, but borrowed at a somewhat earlier period when the English dialect
form coney was in common use in Scotland with the meaning ‘rabbit’. Sic tran-
sit historia verbi.
Still the question remains whether large-scale changes in the political cli-
mate (with attendant changes in attitude toward indigenous minority lan-
guages), especially if accompanied by favorable economic developments,
could work enough magic to bring a struggling minority language back from
the brink. For a language as a whole, the answer is surely yes, since there are
well-documented cases to prove the point. For local forms of a given language,
the answer may still be no, if decline has progressed beyond a certain point.
One might imagine a modern-day Carnegie emerging among the many
hundreds of Gaelic-speaking East Sutherlanders who went abroad in search
of opportunity. No matter how large the infusion of money and goodwill pro-
vided by such a benefactor, the wherewithal for revival would not be avail-
able. Not only would personnel who could speak, or learn to speak, the local
dialect fluently be virtually absent at the outset, but the time period required
to develop an orthography, produce texts, create curricula, and train teach-
ers would almost certainly coincide with the time period during which the
last native speakers fluent enough to act as adequate source people for these
undertakings died out entirely. Whatever might emerge from such an effort, it
would not be East Sutherland Gaelic as we now know it. One could just pos-
sibly reinstate Gaelic in East Sutherland, but the outcome would resemble the
introduction of Oxford English into an Alabama or a Vermont where the local
American English had disappeared.
It is not of course by any means certain that any sort of Gaelic could be
reinstated in East Sutherland, even given a riotously wealthy benefactor dedi-
cated to the revival of his ancestral tongue. It is here that the experiences of the
Irish revival effort are relevant, and they certainly do not provide grounds for
unbounded optimism. Four accounts bearing on the Irish experience, dating
from 1971 to 1985 and offering therefore the perspective of fifty to sixty years
since the establishment of an Irish state with the avowed objective of reviving
Irish as a national language, may be cited in evidence. ‘The Irish language as
the national language is the first official language. The English language is rec-
ognized as a second official language’, according to the 1937 constitution, still
in effect (quoted in Ó Ciosáin 1983: 12).
Most poignant of these accounts is John Macnamara’s little anecdote of
being chidden as a boy of about eight for not speaking Irish with his older
sister when they went into a shop to buy sweets. Though speechless before
the shopkeeper’s reproach, the young boy asked his sister afterward why they
had been reproved, and on learning that the sweetshop lady expected them
to speak Irish because they were being taught it at school, he could only ask
in surprise, “Is Irish for talking?” His judgment on the episode, looking back
on it in adulthood, is that this was “the inevitable effect when society at large
The attempt [to save the Gaeltacht] was based on a false assumption
which was made not merely by the Irish government, but also by the
language movement and the Irish people generally. This assumption
was to the effect that the state bureaucracy, the semi-state companies,
and particularly Gaeltarra Eireann [a state company for the economic
development of the Gaeltacht, set up in 1958], could stop the Gaeltacht
shrinking. Acting on this assumption, the government gave that task to
these agencies. But the assumption that these agencies could perform
that task was quite mistaken. . . . If there is a territory in which a particu-
lar language is usually spoken, and it is contracting continually through
language change on the fringes, who can stop this contraction? Clearly,
only the people of that territory – by deciding to do so and by taking
appropriate measures. So another way of explaining why the state failed
to save the Gaeltacht is by saying that the government failed to perceive
this fact, and failed therefore to take action accordingly. It made no
serious attempt to persuade the people of the Gaeltacht to decide to
end the erosion – it never even asked a representative assembly of them
whether they would try to end it – nor did it establish a representative
regional institution which would have enabled them to “take appropriate
measures” (Fennell 1981: 36–37).
toward the lrish language generally), and especially the ratio of native Irish
speakers to native English speakers both within the work force and within the
district (where the spouses and children of outsiders were also a factor), indus-
trialization projects might or might not work to favor retention of Irish in a
given district. Two of the projects investigated had been successful in terms
of the intended goal; the third had actually had an adverse effect on use of
Irish. The third project had been particularly likely to fail in that respect, in the
judgment of the authors, because it was too highly technological in nature and
therefore required not only that more than the average number of outsiders be
brought in to run it, but also that too many English technical terms be pressed
into service in order to discuss plant operations (O’ Cinneide et al. 1985).
Looking at the results of Irish educational efforts overall as of 1975, David
Greene, basing his figures on a large-scale research survey (Committee on Irish
Language Attitudes Research 1975), notes that an outcome of roughly 220,000
at least fairly fluent non-native speakers is a moderately respectable result for a
government-sponsored language-promotion effort (and he might have added
that this is especially so when the dominant rival language is one of world-
language status, namely English, and the only near neighbour states are offi-
cially English-speaking). But he points to two serious weaknesses which almost
nullify the seeming numerical success: the new speakers are largely from the
well-educated middle c1ass, leaving the working c1ass unaffected, and they are
not residentially clustered in any way, so that no actual neighborhoods have
been created where transmission of Irish to a new generation on more than
an individual family basis is likely to take place (Greene 1981: 6). The Action
Plan for Irish 1983–1986 clearly recognizes this latter problem and proposes –
with what success remains to be seen – that “community schemes” be put
into operation throughout Eire, plus two new city projects in north and south
Dublin, in order to bring Irish speakers into regular contact with each other
and to promote larger-scale, more systematic, and more group-oriented use of
the language (Bord na Gaeilge n.d.: 15).
The emphasis in this chapter thus far on the difficulty and even in some
circumstances the impossibility of promoting language maintenance on an
official or individual-benevolent basis is not, despite appearances, intended to
suggest that such efforts are altogether without value. There are a number of
possible reasons for undertaking efforts of this kind, sometimes in the face of
almost certain failure, or in the certainty that one will be in effect introducing
Oxford English into a region which once spoke Alabaman American English.
The first is that one of the commonest reasons for failure – negative atti-
tudes internalized by the speakers or potential speakers themselves – is in
itself a serious reason for attempting to promote the language. A middle-aged
also become adept, in the favorable climate which prevails, at reworking their
findings for popular radio and TV broadcasts – and these more widely “con-
sumable” versions are often also published in inexpensive, readily available
editions. Schoolchildren have copious materials to acquaint them with their
ancestral culture; an official Curriculum Development Unit puts out richly
illustrated readable books for that purpose. Not all Irish, adults or children, are
especially interested in their Celtic heritage. The point is, however, that if they
should be, there are no obstac1es whatever to learning about it. Irish school-
children, however distant from the Gaeltacht their homes, are most unlikely to
be denied the opportunity to study Irish if they wish it; over most of Highland
Scotland, schoolchildren and their parents are still told either that there are no
teachers available to teach Gaelic or that there is no room in the curriculum
for the subject.
It is not that members of the Irish public must engage with their ethnic
past, but that there is ample opportunity to do that if they are so inclined. In
Scotland, except of late in the areas where bilingual education projects have
begun (the Outer Hebrides, the Isle of Skye), the dedicated intellectual or the
tenacious amateur historian can contrive to acquaint himself with his ethnic
past, but only by dint of major personal effort.
The legacy of language-maintenance efforts in Ireland is far from uniformly
positive. Active aversion to Irish, nurtured mostly in the years of premature
maintenance efforts when the policy of “compulsory Irish” was in force despite
woeful lack of personnel with the requisite language skills or the requisite
methodological training, is still quite widely met with. On the other hand, cul-
tural disinheritance is at least potentially a thing of the past, writers in Irish
have a modest reading public, and fluent speakers have job preference in some
branches of governmental service where a knowledge of Irish is highly useful.
It is by comparison with Highland Scotland, where none of these conditions
prevail (except active aversion to Gaelic, in this instance because of genera-
tions of suppression and negative stereotyping), that the Irish situation gains
a certain luster.
Maintenance efforts on behalf of East Sutherland Gaelic cannot work.
Maintenance efforts on behalf of Irish have barely worked, speaking strict1y in
terms of the ongoing transmission of the language. Yet if asked whether main-
tenance efforts nonetheless serve some useful purpose, have some value, one
can still find reasons for answering in the affirmative.
References
Bord na Gaeilge (n.d.) Action plan for Irish 1983–1986. Republic of Ireland: Bord na
Gaeilge.
Castile, G. P., and Kushner, G. (eds.). 1981. Persistent peoples: cultural enclaves in
perspective. Tucson: University of Arizona Press.
Committee on Irish Language Attitude Research. 1975. Report. Dublin: Oifig Dhíolta
Foilseachán Rialtais.
De Vos, G. 1975. Ethnic pluralism: conflict and accommodation. In Ethnic identity:
Cultural continuities and change, G. De Vos and L. Romanucci-Ross (eds.), 5–41. Palo
Alto, Calif.: Mayfield.
Dorian, N.C. 1978. The dying dialect and the role of the schools: East Sutherland
Gaelic and Pennsylvania Dutch. In International dimensions of bilingual education,
J. E. Alatis (ed.), 646–656. Georgetown University Round Table on Languages and
Linguistics 1978. Washington, DC: Georgetown University Press.
———. 1981. Language death: The life cycle of a Scottish Gaelic dialect. Philadelphia:
University of Pennsylvania Press.
Ellis, P.B., and mac a’ Ghobhainn, D. 1971. The problem of language revival. Inverness:
Club Leabhar.
Fennell, D. 1981. Can a shrinking linguistic minority be saved? In Minority languages
today, E. Haugen, J. D. McClure, and D. Thompson (eds.), 32–39. Edinburgh:
Edinburgh University Press.
Greene, D. 1981. The Atlantic group: Neo-Celtic and Faroese. In Minority languages
today, E. Haugen, J. D. McClure, and D. Thompson (eds.), 1–9. Edinburgh: Edinburgh
University Press.
Macnamara, J. 1971. Successes and failures in the movement for the restoration of Irish.
In Can language be planned? J. Rubin and B. H. Jernudd (eds.), 65–94. Honolulu:
East-West Center, University of Hawaii.
Ó Cinneide, M. S., Keane, M., and Cawley, M. 1985. Industrialisation and linguistic
change amongst Gaelic-speaking communities in the west of Ireland. In Language
Planning and Language Problems 9 (1), 3–16.
Ó Ciosáin, S. 1983. Bilingualism in public administration: the case of Ireland. Revista de
Llengua i Dret 1, 11–19.
Spolsky, B. 1978. American Indian bilingual education. In Case studies in bilingual
education, B. Spolsky and R. L. Cooper (eds), 332–361. Rowley, Mass.: Newbury
House.
Timm, L. A., and Kuter, L. 1982. Language problems and language promotion in
Britanny. Unpublished manuscript.
Wylie, J., and Margolin, D. 1981. The ring of dancers: Images of Faroese culture.
Philadelphia: University of Pennsylvania Press.
1 Introduction
If you were to ask a random group of people what language is for, most would
probably reply that language is for communication: it’s what we use to express
our thoughts and feelings. This is a good and sensible reply, but not a com-
plete one.1 If we used language only for communication, we would not main-
tain more than one language, or try to acquire more than one, unless we had
(or foresaw) some practical need to communicate across a language barrier. But
many people who already speak a language of wider communication (LWC)
are willing to maintain or acquire some other less widely spoken language for
reasons that are not related to practical need. Typically those other reasons are
related to identity. Not that language is the only available signal of identity, of
course: any number of other behaviors or traits can serve the same purpose,
and in any case not all groups that are considered ethnically distinctive have a
language of their own. Still, language is well recognized for its special culture-
carrying capacity, so that learning another language, or keeping up one from
your home, gives you a gateway to the culture embodied in that language.
It’s a general truth that when a language gains in practical value, it also gains in
speakers: people see a need for it and are willing to go to the trouble of acquir-
ing it. In our time a relative handful of the six or seven thousand languages
spoken around the world have gained exceptionally wide distribution, making
them exceptionally useful. These few very useful languages have been adding
more and more new speakers whose parents and grandparents were mother-
tongue speakers of some other language. Where smaller languages have been
swept into the orbits of expanding languages, language shift has followed in
what seems like an irresistible process, creating what many consider a crisis of
language endangerment.
But in actuality this process is interruptible and even potentially reversible.
Political, demographic, and cultural conditions can change, and when they do,
changes in relations between ethnic populations and their languages may fol-
low. The breakup of the former Soviet Union created new conditions for lan-
guages such as Kazakh and Uzbek in Central Asia, for example, and in Solomon
Islands, in the South Pacific, the advantage enjoyed by Bugotu, favored by mis-
sionaries, was lost when the influence of the missionaries was replaced by
other cultural forces (Terrill 2002, 207–208).
a nationwide restoration of Irish. But one scholar has pointed out that
when the Gaelic League was founded in 1893, 99 percent of the population
of Ireland spoke English and 85 percent of those did not speak Irish; he sug-
gests that successfully reawakening a national consciousness after centuries of
relentlessly hostile colonial rule was more surprising than failing to persuade
85 percent of the population to change their language (Ó Cuív 1969, 128–129).
Another scholar describes Irish-language policy since 1922 as a struggle to
find a fair and suitable balance between two objectives: maintaining Irish in
the discontinuous western districts where it was still a first language in 1922
(the Gaeltachtaí), and reviving Irish elsewhere in the country by increasing
the number of speakers through education (Ó Riagáin 2008, 56). No dispas-
sionate assessment can assert that these policy aims have since been fulfilled.
Changing employment and social network patterns in recent decades have
weakened Irish as a community language in the Gaeltachtaí (Ó Riagáin 2008,
57); at the same time, national economic development has created pressures
leading to a relaxation of language requirements that promote Irish by linking
job opportunities to proficiency in Irish (Ó Riagáin 1988, 45–47). Worrisome
to such knowledgeable Irish scholars as Ó Riagáin and Gearóid Ó Tuathaigh is
the fact that the largely middle-class networks of active Irish users that have
emerged outside the Gaeltachtaí via the education system have not produced
any self-sustaining urban communities, nor has the Irish state promoted the
development of such communities (Ó Riagáin 2008, 59–60; Ó Tuathaigh 2008,
41). Ó Tuathaigh notes the connection many urban Irish-speaking networks
have with one or another of the Gaeltachtaí, whether as place of origin or place
of fluency acquisition, and wonders whether those networks could survive if
the shift to English evident among Gaeltacht schoolchildren were to result in
the disappearance of Gaeltacht base communities for Irish (Ó Tuathaigh 2008,
40–41).
On the other hand, the glass can appear half-full when the Irish situation
is compared with many others. Irish has weakened, especially as a first lan-
guage, but it has not disappeared, even as a first language, despite the fragile
demographic position it was already in well over a hundred years ago. If state
support for Irish has not produced a nation of fully fluent bilinguals (and it has
not), it has produced an education system with options by way of which new
high-fluency Irish speakers can be generated (and continue to be generated,
albeit not routinely). Certain desirable occupational avenues are still reserved
for those with an advanced knowledge of Irish. Irish-language broadcast
media, now especially television, continue to develop, providing jobs that are
particularly attractive to young people. All-Irish schools, in which the medium
of instruction is Irish throughout the curriculum, continue to increase in
ancestral language are often relatives of those same elders, young people who
lost their chance to acquire the language because of a break in the transmis-
sion process; aware of how much escaped them, they are now actively trying to
overcome the loss (Hinton 2001).
In East Sutherland, apart from the disappearance of the livelihood that had
kept fisherfolk separate and the powerful social stigma that that low-income
occupation had produced, there was yet another reason to foresee that the local
form of Gaelic had a limited life expectancy. It differed a good deal from most
other forms of Scottish Gaelic and was somewhat reduced by comparison with
more mainstream dialects in terms of its sound system and its grammar. The
Gaelic language has a codified written form that has long been used for writ-
ing more mainstream dialects, but local East Sutherland pronunciations and
grammatical features did not correspond well to those of the written language.
Because English had been introduced earlier there than in most other parts of
the Highlands, the local Gaelic also showed certain unusual features reflecting
long contact with English. It was easy to predict that if Gaelic were to enjoy
an unforeseen rise in support, locally or nationally, it would not be the local
Gaelic that was promoted but a form closer to the written language. And this is
in fact exactly what has happened. Adult Gaelic classes are now fairly regularly
on offer in eastern Sutherland, and Gaelic-medium primary education exists
both in an East Sutherland village a little to the south of the former fishing vil-
lages and in a neighboring county a relatively short drive away. But the instruc-
tors all speak, and teach, a non-local Gaelic that conforms better to the written
language; only once has an instructor made an effort to include local material,
on an occasion when he had several adult students of fisherfolk descent.
Even so, it would not be true to say that local East Sutherland speakers have
derived no benefit from the heightened visibility of Gaelic in Scotland and its
new instructional availability in Sutherland and other parts of the country.
During the greater part of their lifetimes Gaelic – especially the local Gaelic,
but to some extent Gaelic in any form – was regarded as a sign of cultural back-
wardness, a survival of an outdated way of life that had little or no value for
the future. It was rare for local speakers to encounter people who were seri-
ous about learning Gaelic or showed a strong interest in its history and tradi-
tion; the occasional dialectologist, linguist, or folklorist might visit the region,
some of them taking an interest in the local form of the language, but normally
there was no prestige to be gained through a native knowledge of Gaelic and
frequently there were distinctly negative overtones attached to that status. As
activism on behalf of Gaelic took hold in the last two decades of the twentieth
century and the first decade of the twenty-first, this began to change. There
were more Gaelic learners on the scene as Gaelic classes multiplied, and a
few local Gaelic speakers were able to enjoy, for the first time in their lives, a
certain special standing as native speakers of a suddenly more desirable lan-
guage; they found themselves consulted by learners about pronunciations or
constructions and occasionally approached by hopeful learners for practice in
conversation. This revalorization of a language that was once disdained can be
one of the worthwhile by-products of language revitalization and restoration
efforts, finding expression not just in an improved self-regard among speakers
and the ethnic population but in a general upsurge of interest in the language,
its history, and its culture.
Despite its psychological benefits, however, Gaelic-language activism has
not “saved” East Sutherland Gaelic, which no longer has any fully fluent speak-
ers. Both for the skeptical and for the hopeful, the long-term effectiveness of
support efforts for Gaelic remains an open question. On the one hand, recent
developments favorable to Gaelic have been notable, including the opening of
Gaelic-medium units in a rising number of schools and even a few free-standing
all-Gaelic schools, with 2,766 pupils receiving some or all of their education
through Gaelic in 2008 (Gaeliconline 2008). A Gaelic Language Act passed by
the Scottish Parliament in 2005 created a policy body, Bòrd na Gàidhlig (the
Gaelic Board), committed to the protection and promotion of the Gaelic lan-
guage. A digital television channel dedicated to Gaelic came into being late in
2008. Government funding for these and other Gaelic initiatives is currently
estimated at many millions of pounds.
On the other hand, Gaelic is spoken by a tiny minority of the country’s pop-
ulation (1.2 percent as of the 2001 census), many of whom live in peripheral
rural communities at a great distance from the political and economic centers
of Scottish life. Speaker numbers are expected to drop again at the next cen-
sus, since the deaths of elderly speakers still far outstrip gains in new speak-
ers via home transmission and Gaelic-medium schools. Gaelic, in competition
with other languages (originally Pictish and British, subsequently Scots and
English) throughout its entire existence, does not serve as a symbol of Scottish
identity in the way that Irish serves as a symbol of Irish national identity. In a
time when cost-effectiveness and the bottom line are always a factor in offi-
cial policies and funding decisions, the relatively favorable current position of
Gaelic is very precarious.
and it may be that those who will benefit most from these efforts are those
who have lost the most, members of ethnic groups whose languages are some-
times spoken of as “sleeping,” that is, viewed as a central part of a cultural leg-
acy but unrepresented by active fluent speakers at the time when restoration
movements begin. In spite of the colossal effort that goes into recovering the
structure, the lexicon, and the sounds of such languages, every word or sen-
tence that is discovered or reconstructed and made available in some form to
the ethnic community is a precious heritage to those who wish to reclaim their
languages.
Improving the position of still spoken but sharply receding languages is not
a simple matter, and in particular it is not a matter of language policy alone.
Even the most supportive-seeming language policies can be ineffective if they
are formulated in isolation from other far-reaching social and economic poli-
cies that might have contrary effects. In the Irish case, where this has become
particularly apparent, Ó Riagáin points to the counterproductive effect on lan-
guage revitalization of certain policies undertaken in the economic sphere,
in regional planning, and in education. These other policies were rarely con-
sidered in terms of their impact on language patterns, he notes, yet they may
in the end have more effect on those patterns than explicit language policies
themselves (1997, 170–171). For example, if school districts are created that
send Irish-speaking young people from adjacent areas to separate secondary
schools in each of which English is the dominant language, this will have a neg-
ative effect on maintenance of Irish when compared with a policy that would
have funneled those Irish-speaking young people into a single secondary
school where they constituted a majority (Ó hIfearnáin 2007, 514–515). A more
narrowly language-related policy – awarding prizes for school achievement in
Irish, say – is of limited impact by comparison. What is more, some students of
the Irish revitalization movement argue that elements of the official language-
support policy itself are counterproductive: in adopting an “official” Irish that
is no one’s actual spoken language for the purposes of country-wide educa-
tion in Irish, schooling in Irish actually weakens the use of authentic local dia-
lects without in return winning the allegiance of young Gaeltacht residents
(Ó hIfearnáin 2008).
There is a startling lack of consensus about even some of the most basic
questions in connection with language endangerment and language-support
efforts, as seems clear from one recent review of the prospects for indigenous
language survival: the author surveys the field and concludes that opinions dif-
fer on the threshold of endangerment (in terms of some critical mass of speak-
ers necessary for language survival, for example), on just what level of linguistic
and cultural proficiency is involved in being a speaker (and so in determining
how many speakers a language does or does not have), on the methods and
prospects for stemming language loss, on the reasons that some groups suc-
ceed better than others, on whether outside expertise can or should contrib-
ute significantly to indigenous revitalization efforts, and so forth (Walsh 2005).
There is by now some very solid, nuts-and-bolts advice available to would-be
revitalizers (for example, Brandt and Ayoungman 1989 and Hinton et al. 2001),
but among indigenous, small-language populations there has also long been
a disturbing tendency to concentrate on schooling and to neglect parent-to-
child transmission in the home, even among the ranks of indigenous-language
teachers themselves (Hinton 2009).
Fishman pointed long ago to the limitations of schooling where transmis-
sion and language continuity are concerned (1991, 368–373) and to the need for
intimate, small-scale network processes “too gratifying and rewarding to sur-
render” if a small language is to be maintained and transmitted (1989, 399). Such
network processes are difficult to bring into existence. There is some evidence
that coherent residential or religious spheres of linguistic influence (residen-
tial neighborhoods or religious congregations, the latter sometimes also resi-
dentially grouped), where use of a particular, non-mainstream language can be
cultivated, are particularly effective approaches (Maguire 1991, Burridge 2002,
Al-Khatib and Al-Ali 2005, Hansen 2010), but even the Amish and Mennonites,
whose most conservative communities are famous for language retention, have
difficulty securing ownership of enough land to support residentially coherent
communities. There is also some evidence that situating language revitaliza-
tion efforts within an overall context of cultural revitalization is a relatively
effective approach: this strategy is part of the Master-Apprentice Program, for
example. And always, political and social contexts can change, sometimes very
quickly, making strategies and technologies obsolete almost as soon as they are
devised. But if it would be unwise to declare many particular language move-
ments successes, it would also be unwise to declare them all unequivocal fail-
ures. Even when it appears to be stark, the arithmetic of language maintenance
and revitalization is always to some degree ambiguous.
References
Al-Khatib, Mahmoud, and Mohammed N. Al-Ali. 2005. Language and cultural main-
tenance among the Gypsies of Jordan. Journal of Multilingual and Multicultural
Development 26: 187–215.
Brandt, Elizabeth A., and Vivian Ayoungman. 1989. Language renewal and language
maintenance: A practical guide. Canadian Journal of Native Education 16: 42–77.
Burridge, Kate. 2002. Steel tyres or rubber tyres – maintenance or loss: Pennsylvania
German in the “horse and buggy” communities of Ontario. In Language endanger
ment and language maintenance, David Bradley and Maya Bradley (eds.), 203–229.
London: Routledge Curzon.
Dorian, Nancy C. 1981. Language death: The life cycle of a Scottish Gaelic dialect.
Philadelphia: University of Pennsylvania Press.
——— 1987. The value of language-maintenance efforts which are unlikely to succeed.
International Journal of the Sociology of Language 68: 57–67.
Fishman, Joshua A. 1989. Language and ethnicity in minority sociolinguistic perspective.
Clevedon: Multilingual Matters.
———. 1991. Reversing language shift. Clevedon: Multilingual Matters.
Gaeliconline. Feb. 26, 2009. Retrieved from http://www.gaeliconline.co.uk.
Hansen, Magnus Pharao. 2010. Nahuatl among Jehovah’s Witnesses of Hueyapan,
Morelos: A case of spontaneous language revitalization. International Journal of The
Sociology of Language 203: 125–37.
Hinton, Leanne. 2001. The master-apprentice language learning program. In The green
book of language revitalization in practice, Leanne Hinton and Ken Hale (eds.), 217–
226. San Diego: Academic Press.
———. 2009. Language revitalization in the home. Paper presented at the First Inter
national Conference on Language Documentation and Conservation, University
of Hawaii, Manoa, March 2009. Retrieved from http://scholarspace.manoa.hawaii
.edu/bitstReam/10125/5165/9/5165script.pdf.
———, Matt Vera, and Nancy Steele. 2001. How to keep your language alive. Berkeley:
Heyday Books.
Kuter, Lois. 1989. Breton vs. French: Language and the opposition of political, eco-
nomic, social, and cultural values. In Investigating obsolescence: Studies in lan
guage contraction and death, Nancy C. Dorian (ed.), 75–89. Cambridge: Cambridge
University Press.
Maguire, Gabrielle. 1991. Our own language: An Irish initiative. Clevedon: Multilingual
Matters.
Matras, Yaron. 2005. Language contact, language endangerment, and the role of the
“salvation linguist.” In Language Documentation and Description, Vol. 5, Peter K.
Austin (ed.), 225–251. London: Hans Rausing Endangered Languages Project.
Ó Cuív, Brian. 1969. The changing form of the Irish language. In A view of the Irish lan
guage, Brian Ó Cuív (ed.), 22–34. Dublin: Stationery Office.
Ó hIfearnáin, Tadhg. 2007. Raising children to be bilingual in the Gaeltacht: Language
preference and practice. International Journal of Bilingual Education and
Bilingualism 10(4): 510–519.
———. 2008. Endangering language vitality through institutional development.
In Sustaining linguistic diversity, Kendall A. King, Natalie Schilling-Estes, Lyn Fogle,
Jia Jackie Lou, and Barbara Soukup (eds.), 113–128. Washington DC: Georgetown
University Press.
Ó Riagáin, Pádraig. 1988. Bilingualism in Ireland 1973–1983: An overview of national
sociolinguistic surveys. International Journal of the Sociology of Language 70: 29–51.
———. 1997. Language policy and social reproduction: Ireland 1893–1993. Oxford:
Oxford University Press.
———. 2008. Irish language policy 1922–2007: Balancing maintenance and revival.
In A new view of the Irish language, Caoilfhionn Nic Pháidín and Seán Ó Cearnaigh
(eds.), 55–65. Dublin: Cois Life.
Ó Tuathaigh, Gearóid. 2008. The state and the Irish language: An historical perspective.
In A new view of the Irish language, Caoilfhionn Nic Pháidín and Seán Ó Cearnaigh
(eds.), 26–41. Dublin: Cois Life.
Romaine, Suzanne. 2008. Irish in the global context. In A new view of the Irish language,
Caoilfhionn Nic Pháidín and Seán Ó Cearnaigh (eds.), 11–25. Dublin: Cois Life.
Terrill, Angela. 2002. Why make books for people who don’t read? A perspective on
documentation of an endangered language from Solomon Islands. International
Journal of the Sociology of Language 155/56: 205–219.
Tollefson, James W. 1997. Language policy in independent Slovenia. International
Journal of the Sociology of Language 124: 29–49.
Valenzuela, Pilar. 2010. Ethnic-racial reclassification and language revitalization
among the Shiwilu from Peruvian Amazonia. International Journal of the Sociology
of Language 202: 117–30.
Walsh, Michael. 2005. Will indigenous languages survive? Annual Review of Anthro
pology 34: 293–315.
Yamane, Linda. 2001. New life for a lost language. In The green book of language revital
ization, Leanne Hinton and Ken Hale (eds.), 429–432. San Diego: Academic Press.
The grammar books teach a lot which long ago ceased to be of any rel-
evance to standard Arabic as it is practiced today . . . Since the rules of
Arabic grammar are based on prescriptive rules instead of actual usage,
they will remain hopeless and unattainable goals for the vast majority of
Arab learners. (Ibrahim 1989: 42)
far fewer middle-aged speakers, and perhaps none at all among young people.1
For another thing, introducing a language which can still be modeled for
potential new speakers by remaining fluent speakers is considerably easier
than introducing to them, in any convincing way, a language which exists in
recorded texts or in books, but is not in ordinary use by any living person.
Puristic attitudes should, in theory, be more of a potential problem in revi-
talization than in revival, since bringing about alterations in what people are
already saying could be expected to produce more resistance than prescrib-
ing certain ways of speaking a language they have yet to learn. Yet in actuality
puristic attitudes are likely to cause problems in both sets of circumstances, as
a few illustrations will indicate.
The Tiwi language, spoken on Melville and Bathurst Islands off the north
coast of Australia, shows an all too typical profile for an indigenous language
overtaken by the rapid expansion of a wide-currency language, in this case
English (Lee 1987, 1988). With exposure to intense pressure from English, quite
radical changes have taken place in the structure of the language over a short
time period; thus an older, already largely bilingual generation – which knows
(and among its own members still uses) a conservative traditional form of the
language – co-exists with a younger and wholly bilingual generation, which
uses a much modified form of the same language. The traditional language
is polysynthetic, with a particularly complex verb structure. The elderly still
control this form of the language; but younger speakers’ Tiwi shows changes
in phonology, lexicon, noun classification, syntax, and, above all, in verbal
constructions.
2 The abbreviations are: P ‘past’, CON ‘concomitative’; the parenthesized element in the second
TT line is optional. Note that in each case the NT verb is a borrowing from (Pidgin) English:
wokapat derived from walkabout and lukim from look.
The older people within the Tiwi community of Bathurst wanted the tra-
ditional language used, passing to the school the job of teaching the children
this conservative form of the language which was no longer being transmitted
naturally via the family. This is in itself dubious policy. Fishman 1991 devotes a
full chapter to “limitations on school effectiveness in connection with mother
tongue transmission,” pointing out that “without considerable and repeated
societal reinforcement schools cannot successfully teach either first or sec-
ond languages” (371). The Tiwi situation was more difficult than average, since
attempts to use Traditional Tiwi for school purposes meant not just teaching
literacy, but also providing oral Tiwi instruction: the children were being
introduced to a form of Tiwi quite different from the range of Tiwi styles they
were most familiar with, and sharply different from the kind of Tiwi they actu-
ally spoke. The difficulties were such that the bilingual program shifted over
time away from purely Traditional Tiwi toward the various styles of what Lee
(1987: 80) calls Modern Tiwi: “a modified/simplified traditional Tiwi.” Lee
herself, in her teaching efforts on Melville Island for the Summer Institute of
Linguistics, tried to lessen the gulf between what young people actually spoke
and what they would encounter in their first efforts at reading Tiwi by prepar-
ing comics (a genre she thought might seem acceptable for use of the modified
Tiwi of the young) with the text in “a formal style of N[ew] T[iwi]” (91); she also
encouraged teacher trainees to write stories initially in language they would
ordinarily speak, and subsequently to put that material into a more “proper”
style if they wished, as most did (1988: 92). She encountered difficulty, however,
when a wave of puristic conservatism greeted the texts she had prepared.
Lee worries, not without grounds, about the utility of all the support work
for Tiwi. She fears that the language will not derive realistic benefit from text
preparation, school programs, the compiling of a dictionary, teacher-training
programs, or anything else that might be undertaken, if the children who are
the targets of all these efforts do not find the Tiwi they are exposed to enough
like the Tiwi they speak to interest and encourage them to use the language:
In the Tiwi situation, there is an echo of the problem that plagues Arabic: puris-
tic norms militate against the teaching of the students’ actual language forms,
and instead promote unrealistic norms from an earlier version of the language.
Tiwi, like other hard-pressed minority languages in intense contact situations
(Dorian 1981: 151–53, Schmidt 1985: 213–14), has undergone major change in a
sharply foreshortened time frame. In the span of three or four co-existing gen-
erations, it has reached something resembling the distance from a traditional
model which Arabic has reached over many centuries. But in the case of Tiwi,
it is neither the teachers at the Roman Catholic school nor the linguists of the
Summer Institute of Linguistics who chose to promote the most conservative
form of the language, but “influential members” of the community itself.
The problems related to purism which arise in the case of Irish are differ-
ent, but no less difficult. Puristic attitudes operate in two different directions
in the Irish case. One set of conservative attitudes exists in the tiny remaining
heartlands of the Irish language, consisting mainly of several non-contiguous
extreme-western pockets of Donegal, Mayo, and Kerry; the other exists in
the positions taken by the standardizers who were responsible for arriving at
a normalized Irish suitable for country-wide use in textbooks, official docu-
ments etc. The two seem irreconcilable.
Irish, in contrast to Tiwi, has an exceptionally long literary history, with a
written tradition dating to the seventh century. All the same, when the revital-
ization movement took hold in the late 19th century, the last great period of the
written language (Early Modern Irish, also known as Classical Irish) lay more
than 200 years in the past. In the interim stretched a period of drastic decline
in the number of speakers, of restriction in geographical distribution, of con-
traction of spheres of use, and of repression or neglect of the language by the
governing powers. What had been a brilliant literary language survived in the
monuments that had been produced by its practitioners; but so far as the spo-
ken language was concerned, what remained was rustic in character, surviving
in daily use almost exclusively among a peasantry. There were marked dialect
distinctions from region to region.
When political independence lent practical means to the hopes of those
who wished Irish Gaelic to become once again the distinctive language of the
Irish people, and to come into regular use among them, modernizing and stan-
dardizing tasks loomed large. It was not feasible to take one of the living Irish
dialects as the clear-cut basis for a modern standard language, since none of
the three main dialects had any obvious superiority in prestige or numbers
(Ó Baoill 1988: 111); without such a realistic basis for selection, the speakers
of each dialect were certain to object to any one dialect among the three
being singled out for official favor. Compromise was necessary, but the result
was inevitably artificiality. Revitalization required a single, normalized form
of Irish which could transcend dialect differences. To speakers of living Irish
dialects, however, the result is Gaeilge B’l’ Ath’ ‘Dublin Irish’, a stilted, unnatu-
ral form of Irish (Hindley 1990: 60). The puristic conservatism of native Irish
dialect speakers takes authenticity as its chief virtue, and “Dublin Irish” fails
the test.
If authenticity is the form which conservative attitudes take in the rural
Gaeltachtaí – the scattered residual areas where one of the living dialects is
still spoken natively – historicity might be said to be the form which conserva-
tive attitudes take in the “official” Irish ultimately produced by the long labors
of the standardizers. The standardized form of Irish steers clear of extreme
regionalism, but makes less effort to steer clear of the grammatical complexi-
ties of conservative forms of the language.
Literature written in the standard form creates very few problems for the
average reader. In trying to use the system, however, even the most com-
petent users have to often consult the dictionaries or handbooks. This is
due mainly to the complicated morphological system of Irish, and the
standard now evolving has not succeeded to any great degree in reduc-
ing the complex system of grammatical rules involving the use of inflec-
tions and the mutations of initial consonants of Irish to express different
shades of meaning. Many of the rules and forms advocated by the stan-
dard have been simplified in the speech of native speakers . . . (Ó Baoill
1988: 117).
some of the more complex features of the traditional grammar; but in each
locality they preserved the distinctive speech of their own region with its own
forms, phraseology, and idioms. While each form of Irish has certain clear
advantages – supra-regionalism and uniformity in the case of the standardized
Irish, realistic local vividness in the case of the regional dialects – each has
faults that limit its overall usefulness.
By far the fullest and most penetrating account of the threat which pur-
ism can pose to a small language community is Jane Hill and Kenneth Hill’s
study (1986) of Mexicano (Nahuatl, i.e. modern Aztec) in the Malinche region
of Mexico. Mexicano and Spanish have co-existed for centuries, with contact
phenomena appearing in both languages as spoken regionally; but Spanish
influences in Mexicano are the more evident. Although the Hills take the view
that “an ecological perspective can see linguistic syncretism as having a posi-
tive, preservationist effect on a language when its speakers must adаpt rapidly
to changing circumstances” (59), some local attitudes are less tolerant. As part
of a heightened attention paid to the ethnic boundaries of the local popula-
tion and their towns, native purists reject the syncretic form of Mexicano, with
its high Hispanic content, and insist on an artificial variety which is not only
unrealistic, but in some instances even inauthentic (140). The result is a self-
consciousness and exclusivity with potentially harmful consequences.
Purism in the Malinche towns may work against the survival of the
Mexicano language. Since Mexicano is considered to be of very little eco-
nomic utility, many people question the instrumental value of the lan-
guage. Purism, which deprecates all modern usage, inspires speakers to
question the moral and aesthetic value of Mexicano as well . . . Since no
formal education about Mexicano is available in the Malinche towns, it
is unlikely that young speakers can be educated to a purist standard, and
when young speakers feel that their Mexicano is inadequate, they may
choose to use only Spanish. (Hill & Hill, 140–41).
The Hills note (122–23) that Mexicano purism focuses most zealously on lexi-
con as a particularly salient locus for contamination. External lexical influence
is usually conspicuous in contact situations; in most revitalization efforts, the
problematics of conservatism is very clearly to be seen in connection with
attempts to nativize the lexicon while also updating it. There are normally
two obvious options: to borrow or to coin. The difficulty is that the remain-
ing native speakers often reject both. By way of illustration I offer anecdotal
evidence from my personal experience with the reception of a coinage, on the
one hand, and a pair of borrowings, on the other, by Scottish Gaelic speakers
(cf. Dorian 1978).
efforts. Yet today’s widely used coinage or extension was often yesterday’s
laughingstock; transition from the latter to the former can seem random and
mysterious, and the observer may see no reason to account for the successes,
as opposed to the failures.3
It’s reasonably easy to appreciate the difficulties posed by conservative atti-
tudes in instances of revitalization. By comparison, revival settings look invit-
ingly free of potential resistances, since speakers have no entrenched habits to
overcome. In revival settings, however, the hazards of rival proposals, giving
rise to rival factions, pose just as great a threat; purism of one sort or another
is quite likely to be at the heart of the rivalry. If the language to be revived is
well preserved, even though not conversationally spoken, there may be dispa-
rate traditions for rendering it phonologically, as was the case with Ashkenazic
and Sephardic pronunciations of Hebrew. If the language to be revived is not
well attested, or is attested in more than one earlier form, there may be dispa-
rate reconstructions of the language itself, forming the basis of rival teaching
materials for modern-day learners. This is the unhappy situation of Cornish,
the Celtic language of Cornwall, which survived as a spoken language up to
the late eighteenth century but is only moderately well attested as a written
language.
Efforts to recover Cornish, and to create the texts and reference works which
would make it accessible to those who wanted to become acquainted with it,
began in the very early twentieth century. Antiquarian interest in the language
gradually gave way to a more active involvement, with some individuals and
groups espousing revivalist sentiments. Two different approaches to locating
“true” Cornish have led in very recent years to rival revivalist factions, each
promoting its own version of the language. One faction favors Dr. Kenneth
George’s version of Cornish, based on meticulous computer analysis of the
grammar, lexicon, and orthography of the available Cornish texts, which span
the period roughly from the twelfth to the seventeenth centuries (Ellis 1974,
O’Callahan 1989: 27). The other favors Richard Gendall’s version of Cornish,
3 The struggle to eliminate foreign loanwords from the German language produced vast
numbers of coinages and extensions from the 17th century on. Browsing through a few
Verdeutschungswörterbücher produces some smiles at the apparent absurdities that zealous
Germanizers urged on their compatriots; but these are quickly balanced by surprise at the
realization that many words, utterly respectable and ordinary now, were coinages no less
novel and curious in their time than the coinages which failed to gain acceptance and so
produce smiles. In the 19th century, Arthur Schopenhauer found the coinage Stickstoff, rec-
ommended by the writer Campe as a replacement for Nitrogen, so ugly that he suggested
Azot instead (Tschirch 1969: 260). But Campe prevailed; modern German dictionaries offer
Stickstoff for ‘nitrogen’, but no Azot.
4 In the village of Embo, East Sutherland (Scotland), where Gaelic is dying out, it surprised me
originally to find that the two mothers whose children are today the youngest Gaelic speak-
ers were less conservative in their Gaelic than many of their older-fluent-speaker peers. It’s
possible, however, that the willingness of these two women to adapt their Gaelic somewhat
toward young-fluent-speaker norms may have been a factor in producing a home environ-
ment which encouraged their children to speak Gaelic, when most young people of compa-
rable ages were not actively acquiring the language.
to count, and so disdain it. This is a value judgment and should be recognized
as such.
When the convergence in question lies safely in the past, the disagreement
is innocent enough. English emerged from its period of social subordination to
French, in the wake of the Norman Conquest, quite different in form from the
English which had existed before the Conquest. Arguments can be made (and
have been) for its continuation as a distinctly Germanic language – or for its
latterday emergence as a mixed language, or even a creole. When convergence
features are evident in a present-day speech form, the debate can take on more
than academic interest. Maguire notes that the Shaw’s Road children are capa-
ble of some degree of grammatical monitoring and avoidance of English loan-
words in more formal situations (1987: 87, 1991: 228). But overall she finds that
they “adapt their [Irish language] system to suit their own needs,” and her sum-
marizing comment (1987: 88) leans to the conservative side: “Although com-
municative competence and functional adequacy are mastered, a language
which is very much on the defensive must aim higher in order to ensure its
own separateness from the dominant language.” This is at heart a puristically
inclined evaluation. In its absolute form, it is belied by the evidence of the
sectarian speakers of Pennsylvania German; but it is certainly true that a sense
of separate identity is a valuable sustaining feature in ethnic language revival
and revitalization efforts.
Ó Baoill, considering the outlook for preservation of traditional Irish pho-
nological contrasts among speakers of whom many or most will be learners,
in the context of the Irish Republic, considers compromise a likely necessity:
“If Irish is to become a viable means of communication among the general
population, I fear that much leveling will take place, and it is certain that many
of the contrasts now existing in Irish will be lost. If the revival of Irish were to
succeed, then it might all be worthwhile” (1988: 125). Ó Baoill’s is a slightly tooth-
gritting embrace of revival, since he suspects that it can only come at a cost of
phonological leveling in the original language. His predictions for Irish may
have been embodied for Greek Albanian in the young Attican semi-speakers
of Arvanitika whom Hamp encountered.
Maguire and Ó Baoill both hope for the preservation of Irish and the prolif-
eration of speakers of Irish. Maguire, like many language loyalists before her, is
asking how dilute a language can become while still remaining the linguistic
entity it was – distinct from all others, including (and especially) the neighbor-
ing language of wider currency. Ó Baoill, in a more pragmatic tradition, is ask-
ing how traditional a threatened language can afford to remain if its traditional
forms pose obstacles to learnability and hence dissemination. In the best of all
possible worlds, one would not need to choose, of course, and Fishman’s char-
acterization of enlightened planning in a nationalist framework would prevail:
“The enlightenment of nationalist purism in language planning . . . proceeds
along many well-trodden paths: the differentiation between ethnic core and
nonethnic periphery, between technical and nontechnical, the differentiation
between preferred and nonpreferred sources of borrowing, and, finally, the
appeal to common usage among the masses” (1989: 309). But in very small lan-
guage communities which have no nation to their name, and little immediate
prospect of acquiring anything deserving the term “masses,” a choice may be
unavoidable. The rapidity of change and the expansion of contacts with other
peoples add to the pressures. Drapeau (1992: 3) points this out in connection
with Betsiamites Montagnais, an Algonquian language with a moderately solid
speaker base but extensive exposure to French, despite its geographical isola-
tion in Northern Quebec.
On the evidence of the difficulties posed by puristic stances for even very large
modernizing languages, like Hindi and Arabic, and with the suggestive find-
ings of Hamp and Huffines in cases at the other end of the spectrum as an
encouragement, it may prove the wiser course to accept considerable compro-
mise rather than make a determined stand for intactness, where threatened
languages are at issue. If a language survives, after all, it has a future. If it can
never again be exactly what it once was, it may yet be something more than it
now is. Gifted speakers and writers may eventually appear who will coax new
richness of expression from it, and tease it into forms that will be uniquely
its own, even if not those of its past. Ælfric might well have been horrified at
what Chaucer called English, had he lived to see it, since English emerged in
a markedly altered state, both lexically and grammatically, from two centu-
ries of domination by the Norman French and their language. But if Chaucer
wrote in a sharply modified and even gallicized form of English, by comparison
with that of Ælfric, that did not prevent Chaucer from writing masterful and
enduring literary works. Purity need not be a requirement for persistence, and
compromise need not be the death knell, for small languages any more than
for larger ones.
References
Anonymous. 1990. Le cornique: Renaissance d’une langue celtique. Ar Men 29: 3–19.
Coulmas, Florian. 1989. Language adaptation. In Florian Coulmas (ed.), Language
adaptation, 1–25. Cambridge & New York: Cambridge University Press.
Dorian, Nancy C. 1978. East Sutherland Gaelic. Dublin: Institute for Advanced Studies.
———. 1981. Language death. Philadelphia: University of Pennsylvania Press.
Drapeau, Lynn. 1992. Language birth: An alternative to language death. Paper pre-
sented at the XVth International Congress of Linguists, Québec.
Ellis, P. Berresford. 1974. The Cornish language and its literature. London: Routledge &
Kegan Paul.
Fishman, Joshua A. 1989. Language and ethnicity in minority sociolinguistic perspective.
Clevedon, England: Multilingual Matters.
———. 1991. Reversing language shift. Clevedon, England: Miltilingual Matters.
Hamp, Eric P. 1989. On signs of health and death. In Nancy C. Dorian (ed.), Investigating
obsolescence, 197–210. Cambridge & New York: Cambridge University Press.
Haugen, Einar. 1977. Norm and deviation in bilingual communities. In Peter A. Hornby
(ed.), Bilingualism: Psychological, social and educational implications, 91–102. New
York: Academic.
Heyd, Uriel. 1954. Language reform in modern Turkey. Jerusalem: Israel Oriental Society.
Hill, Jane H., & Hill, Kenneth C. 1986. Speaking Mexicano: Dynamics of syncretic lan-
guage in Central Mexico. Tucson: University of Arizona Press.
Hindley, Reg. 1990. The death of the Irish language. London: Routledge.
Huffines, Marion Lois. 1989. Case usage among the Pennsylvania German sectarians
and non-sectarians. In Nancy C. Dorian (ed.), Investigating obsolescence, 211–26.
Cambridge & New York: Cambridge University Press.
Ibrahim, Muhammad H. 1989. Communicating in Arabic: Problems and prospects.
In Florian Coulmas (ed.), Language adaptation, 39–59. Cambridge & New York:
Cambridge University Press.
Kuter, Lois. 1989. A note from the editor. Bro Nevez 32: 40.
Lee, Jennifer. 1987. Tiwi today: A study of language change in a contact situation.
Canberra: Australian National University.
———. 1988. Tiwi: A language struggling to survive. Australian Aborigines and
Islanders Branch, Summer Institute of Linguistics, Work Papers 13 (series B), 75–96.
Maguire, Gabrielle. 1987. Language revival in an urban neo-Gaeltacht. In Gearóid
MacEoin, Anders Ahlqvist, and Donncha Ó hAodha (eds.), Third International
Conference on Minority Languages: Celtic papers, 72–88. Clevedon, England:
Multilingual Matters.
———. 1991. Our own language: An Irish initiative. Clevedon, England: Multilingual
Matters.
* I’m indebted to Christina Bratt Paulston for helpful criticisms of the first draft of this chapter
and for suggestions for its improvement.
opinions of the various regional speech forms, which were labeled coarse and
stupid and were considered to keep the people ignorant and superstitious
(Grillo 1989: 31, 174).
These were in fact commonplace attitudes in European polities.1 Grillo
(1989: 173–74) states bluntly – and accurately, I think – that “an integral fea-
ture of the system of linguistic stratification in Europe is an ideology of con-
tempt: subordinate languages are despised languages.” This has been true both
where regional dialects are concerned and where the languages of subordinate
ethnicities are concerned. In his study of the rise of language standards and
standard languages, Joseph (1987: 31) suggests that language is particularly sus-
ceptible to what he calls “prestige transfer”:
He further considers that “the power which prestigious dialects hold over non-
prestigious speakers goes beyond what logic and rationality can predict or
account for,” and that the prestige-holding segment of a population can use the
mechanisms of prestige-language standardization to maintain and increase lin-
guistic differences between themselves and speakers of less prestigious speech
forms (ibid.). The histories of several of the national languages of Europe,
very conspicuously those of French and English (Grillo 1989), are histories of
a growing monopoly on legitimacy and prestige by a single dominant speech
1 They were not uncommon in non-European contexts either, for that matter. The Aztecs used
a variety of unflattering terms for the languages of their subject peoples, some of which stuck
and became the name by which the language is still known, at least to outsiders. Derogatory
language names deriving from Aztec labels include Chontal “foreigner”, Popoloca “unintel-
ligible”, and Totanac “rustic” (Heath 1972: 3).
2 One result of this is a tendency among other-language speakers in contact with standard-
language speakers to consider that any feature in which their own language differs markedly
from the standard language must indicate some own-language deficiency. In Scottish Gaelic
the adjective normally follows the noun. Unfortunately for Gaelic speakers, the dominant
standard language to which they compare their own is English rather than French. In East
Sutherland, Gaelic speakers frequently remark that Gaelic “puts the cart before the horse” in
this regard, implying a failing on the part of Gaelic (since carts don’t belong before horses).
That is, they assume that English, which they were taught in school, represents things as
they ought to be. Because the ancestral language is measured against dominant-language
norms, it’s difficult for speakers who have no special training – and often no schooling in
the ancestral language at all – to see in a positive light any unique or highly developed fea-
tures of their own language. Gaelic, for example, has a very rich system of emphatic suffixes
which can attach to nouns, adjectives, many pronouns, and a few verbal forms. Although the
emphatic suffixes lend Gaelic a distinctive flavor and constitute a rich discourse device, I’ve
never heard an ordinary Gaelic speaker so much as mention the emphatic suffixes, let alone
praise them for the subtle effects they make possible in creating discourse tone and express-
ing point of view and social distance.
7 The history of the Osage tribe after sudden oil wealth illustrates some of the problems associ-
ated with abrupt and unexpected wealth.
8 Catalan can be considered a “small” language in the context of Spain since it is spoken by
a much smaller number of people than Spanish and is found chiefly in a single region of
the country. By comparison with minority languages in many other settings, however, it is
extremely well represented. Numbers as such form an uncertain measure of linguistic secu-
rity, of course; some of the distinctly precarious languages of Central India have over a mil-
lion speakers, e.g. Kurux (Abbi 1995), and so did beleaguered Breton, in France, as recently as
1926 (Timm 1973: 289, citing Meillet 1928: 380).
education, identified the “new middle class” in Wales as the chief factor in the
turn-around (1980: 77–78):
an emerging middle class, and any or all of these factors may either precede or
accompany movement towards greater political autonomy.
While these factors appear to enhance the chances of ancestral language
maintenance, their absence need not doom a small language to rapid disap-
pearance. The case of the Arizona Tewa, still in possession of their ancestral
language even though long enclaved among the Hopi, suggests a different sort
of counterpoise to the negative effects of European-derived linguistic ide-
ologies. On the basis of long-term work among the Arizona Tewa, Kroskrity
proposes that in the theocratic Pueblo societies, where political and religious
authority are fused, ceremonial speech has a position analogous to that of the
standard language in a nation-state. The highly regarded ceremonial speech
variety called te’e hi:li ‘kiva talk’ is of critical importance to the Arizona Tewa,
and the rigorous standards applied to its maintenance spill over into attitudes
towards Arizona Tewa generally (Kroskrity 1993: 37–39):
The Arizona Tewa have maintained their ancestral language for 300 years,
despite enclavement within a Hopi environment, despite considerable inter-
marriage with the Hopi, and despite a small population base. There may well
be a variety of elements in their success: they pride themselves on their skill at
languages, for example, and they consider their bilingualism in Hopi, while the
Hopi do not control Tewa, a form of cultural victory (Kroskrity 1993: 23, 218).9
10 There are other cases of great potential interest. English is said to be the language of all
monetary rewards for the Koasati of Texas, who number under 200. The children attend
English-language schools, furthermore, and yet the ancestral language is successfully
transmitted with “monolingual Alabama-Koasati speakers still present in each genera-
tion of children” (Saville-Troike 1989: 215).
Northern Ireland, came forward with the proposal that what they term “mul-
tiple allegiances” be recognized. The Irish prime minister, in a talk before the
National Press Club in Washington, DC (Bruton 1995), pointed out that land
ownership is no longer the basis of real wealth in modern economies, so that
discrete assignment of all land to mutually exclusive nations does not have the
urgency it once had. So long as no flags are run up the pole (i.e. so long as
certain traditional emblems of purely political allegiance are avoided), indi-
viduals with different sets of multiple allegiances should reasonably be able
to co-exist in one region. He is so much persuaded of the power of this con-
cept that he recommends it as a potential solution in other ethnically and lin-
guistically complex regions of Europe such as Latvia and Catalonia. Multiple
allegiances in this sense might be seen as an extension into the sphere of
political organization of the sociologist’s status sets, the totality of all the sta-
tuses one occupies (not always entirely congruently) in one’s social life. In
the political sphere this suggestion is to some extent “post-nationalist” and
to that degree perhaps an escape hatch from the demands of mutually exclu-
sive nationalisms.11 The fact that recognition of multiple allegiances is being
recommended as a solution for otherwise irresolvable nationalist conflicts
precisely in Europe could be especially helpful, since it is the concept of the
nation-state coupled with its official standard language, developed in mod-
ern Europe and extended to the many once-colonial territories of European
states, that has in modern times posed the keenest threat to both the identi-
ties and the languages of small communities. Outside the modern European
sphere of interest the same problem of insistent single-language dominance
coupled with hostility to minority languages has not necessarily arisen.12 The
Ottoman Empire was largely free of it over most of its long history, and the 300-
year coexistence of the Arizona Tewa with the First Mesa Hopi indicates that
both ethnic and linguistic persistence are feasible over long time spans with-
out fatal ethnic hostility on either side. Thailand, with a stable hierarchical
social structure until very recently, is said to have had minimal ethnolinguistic
conflict over the six and a half centuries of its monarchy despite the pres-
ence of a variety of ethnicities within the national borders (Smalley 1994).
In yet another ironic development, there has been a good deal of
consciousness-raising within the European Community recently in connec-
tion with small languages and minority languages. The EC member states have
not been willing to yield to economic pressures and permit the use of some one
or two languages as the Community’s official working languages (English and
French being the chief candidates). Instead they have insisted on the national-
language principle and have accepted the enormous costs of mounting inter-
pretation services and document-translation services for each of the individ-
ual national languages. This unyielding adherence to the national-language
ideology has given rise to unprecedented European support for multilingual-
ism, and in an overspill of protective enthusiasm for smaller languages, even
minority languages within the EC countries have gained a certain increased
recognition and at least a few economic benefits (Coulmas 1992: 116–117). This
recent development shows a language ideology which has previously worked
against small languages beginning to work for them instead: if all nations, no
matter how small, have a right to the use of their own language, then by exten-
sion other small-language populations, with or without a nation-state of their
own, can with some justice claim the right to the use of their own languages as
well. In time some of this change of attitude could conceivably be generalized
into wider European spheres of influence.
Popular opinion in the United States and other European-settled parts of
the world is unfortunately still largely infected with earlier European lan-
guage ideologies of the types discussed above, all unfavorable to the survival of
smaller indigenous languages. The emergence of government level initiatives
to counter some of the negative aspects of nationalism (in the form of the new
“multiple allegiances” discussion), the stirrings of a new legitimacy for small
languages, and perhaps also the growing acknowledgement in recent decades
(in the United States at any rate) of the value for the health of individual and
planet of at least some non-Western, small society forms of religious or spiri-
tual world-view, conceivably offer a small window of opportunity to make the
case for the wisdom of preserving linguistic and cultural diversity.
Still, recent concerns about loss of linguistic and cultural diversity, together
with new recognition of the possibilities of multiple sociopolitical allegiances
and of the legitimacy of ethnic languages and of multilingualism, come very
late in the day for most small languages. Material well-being has been inti-
mately linked to the adoption of dominant languages for a very long time,
and the reality of that linkage is undeniable. It requires enormous social and
psychological self-confidence for any small group to insist on the importance
References
Abbi, Anvita. 1995. Language contact and language restructuring: A case study of
tribal languages in Central India. International Journal of the Sociology of Language
116: 175–85.
Barber, Carroll G. 1973. Trilingualism in an Arizona Yacqui village. In Paul R. Turner,
ed., Bilingualism in the Southwest, 295–318. Tucson: University of Arizona Press.
Bruton, John. 1995. Address to the National Press Club, Washington, DC, March 17,
1995. (Federal News Service, Inc., broadcast transcription.)
Coulmas, Florian. 1992. Language and economy. Oxford: Blackwell.
Decime, Rita. 1994. Un projet de trilinguisme intégré pour les enfants des écoles mater-
nelles de la Valée d’Ayas. International Journal of the Sociology of Language 109:
129–37.
Fishman, Joshua A. 1989. Language and ethnicity in minority sociolinguistic perspec
tive. Philadelphia: Multilingual Matters.
Flinn, Juliana. 1990. We still have our customs: Being Pulapese in Truk. In Joyce Linnekin
and Lin Poyer, eds, Cultural identity and ethnicity in the Pacific, 103–26. Honolulu:
University of Hawaii Press.
———. 1992. Diplomas and thatch houses: Asserting tradition in a changing Micro
nesia. Ann Arbor: The University of Michigan Press.
Gellner, Ernest. 1983. Nations and nationalism. Ithaca, NY: Cornell University Press.
Grillo, Ralph D. 1989. Dominant languages: Language and hierarchy in Britain and
France. Cambridge: Cambridge University Press.
Heath, Shirley Brice. 1972. Telling tongues: Language policy in Mexico, colony to nation.
New York: Teachers’ College Press.
Heath, Shirley Brice, and Richard Laprade. 1982. Castilian colonialization and indig-
enous languages: The cases of Quechua and Aymara. In Robert L. Cooper, ed.,
Language spread, 118–47. Bloomington, IN: Indiana University Press.
Hohenthal, W. D., and Thomas McCorkle. 1955. The problem of aboriginal persis-
tence. Southwestern Journal of Anthropology 11: 288–300.
Joseph, John Earl. 1987. Eloquence and power: The rise of language standards and stan
dard languages. Oxford: Blackwell.
Khleif, Bud B. 1980. Language, ethnicity, and education in Wales. The Hague: Mouton.
Kroskrity, Paul V. 1993. Language, history, and identity: Ethnolinguistic studies of the
Arizona Tewa. Tucson: University of Arizona Press.
Kuter, Lois. 1989. Breton vs. French: Language and the opposition of political, eco-
nomic, social, and cultural values. In Nancy C. Dorian, ed., Investigating obsoles
cence: Studies in language contraction and death, 75–89. Cambridge: Cambridge
University Press.
Langgaard, Per. 1992. Greenlandic is not an ideology, it is a language. In Nelson H. H.
Graburn and Roy Iutzi-Mitchell, eds, Language and educational policy in the north,
167–78. Berkeley: Working Papers of the Canadian Studies Program, International
and Area Studies, University of California at Berkeley.
Meillet, Antoine. 1928. Les langues dans l’Europe nouvelle. Paris: Payot.
New York Times. 1975. France refuses to recognize six children because of their Celtic
names. January 12.
Rumsey, Alan. 1990. Wording, meaning and linguistic ideology. American Anthropologist
92: 346–61.
Sabino, Robin. 1994. Establishing the necessary database for the restoration of the
Cherokee language to the Alabama Cherokee population: A proposal. Unpublished
ms.
Salisbury, Richard F. 1962. Notes on bilingualism and linguistic change in New
Guinea. Anthropological Linguistcs 4: 1–13.
Saville-Troike, Muriel. 1989. The ethnography of communication. 2nd ed. Oxford:
Basil Blackwell.
Seliger, Herbert W., and Robert A. Vago. 1991. The study of first language attrition:
An overview. In Herbert W. Seliger and Robert A. Vago, eds, First language attrition,
3–15. Cambridge: Cambridge University Press.
1 Introduction
with the people of Greenland (greater autonomy) and their language and with
the people of Slovenia (independence) and their language. With only some-
thing over two hundred nations presently in existence, however, and with most
of them intent on maintaining the territorial and political status quo, it seems
clear that most of the world’s 6,000 or so languages, insofar as they survive at
all, will remain the languages of peoples who are minorities within a relatively
small number of nation-states.
Under conditions of steadily expanding communication and transport
networks and of ongoing economic and cultural globalization, genuine geo-
graphic and social isolation can only grow rarer in coming decades. The out-
look for endangered minority languages is linked to the willingness of minority
peoples to sustain the bi- or multilingualism that characterizes many of them
now or to cultivate it deliberately in those cases where the ancestral language
has become the valued possession only of the elderly. Such willingness is as
much a matter of ideology and cultural values as of objective factors like popu-
lation size, a viable economy, or political autonomy. Because minority peoples
are subordinate peoples, furthermore, the ideologies of the dominant group
are fully as important to the outcome as those of the minorities themselves.
than vice versa, though each group would have included bilingual members.
This is the general linguistic situation reconstructed, for example, for the larger
and smaller peoples of the Papua New Guinea Highlands in contact with each
other in the period before contact with people of European descent (Sankoff
1980: 108–09). Where interaction among a variety of peoples was very frequent,
bi- or multilingualism was also more widespread. This seems to have been the
norm, for example, in much of Aboriginal Australia and in parts of sub-Saharan
Africa, and it is still the prevailing condition among some groups in both
settings.
Where a number of languages coexisted interactively within a particular
region, hierarchies of power and prestige often existed among the languages,
reflecting hierarchies of power and prestige among the various groups who
spoke them. One factor in the establishment of such hierarchies was demogra-
phy, as the Highland Papua New Guinea instance suggests: larger peoples had
an obvious advantage, since they were better able to assert their control via
warfare or political domination. Abundance of natural resources could also
play a role. In one interconnected island complex in Micronesia, the resource-
rich high-volcanic island of Yap constituted the center of a tribute system that
involved a great many outlying island groups made up of coral atolls. Distance
from Yap corresponded to place in the social and political hierarchy of islands,
so that the most outlying atoll groups ranked lowest while those closest to Yap
ranked highest. The origins of Yap dominance are not now entirely certain, but
access to resources available only on Yap (e.g. timber, certain spices and foods)
and fear of Yapese sorcery seem to have sustained the system, remembered
and resented even now by islanders from the outlying atoll groups, despite the
fact that it ceased to operate more than a hundred years ago (Flinn 1992: 21–23).
Embrace of the nation-state in the modern era has gone hand-in-hand with
embrace of a one-language, one-nation ideology. This ideological construct is
generally associated with French and German philosophers of the eighteenth
century and linked with policies that took shape in France during the clos-
ing decade of the eighteenth century, after the Revolution of 1789 (Grillo 1989:
22–42; Woolard 1998: 16–17). At that time the longstanding and previously
unremarkable existence within the French polity of substantial subcommuni-
ties who neither spoke nor understood French came to be viewed as unaccept-
able. The unity of the new revolutionary state was henceforth to be expressed
via a common language, replacing the linguistic heterogeneity that in the revo-
lutionary view had served the purposes of a discredited monarchy by prevent-
ing various segments of the country’s population from making common cause
with one another. The Alsatians, the Basques, the Bretons, and the Occitanians
would come to feel their national unity and would express it, according to
revolutionary tenets, by adopting the use of the French language. Certain
characteristically European ideological positions were given expression in the
implementation of this policy. A single language variety associated with peo-
ple of high social position (the king and his court, in this case) was accorded
fixed form and unique authority through standardization, and a monopoly
of legitimacy and prestige was conferred on that single form. In the resultant
linguistic hierarchy, the unstandardized language varieties of politically and
socially subordinate peoples within the state underwent a parallel attitudinal
subordination and were subjected to what has been termed an “ideology of
contempt” (Grillo 1989: 173–74).
1 In the same chapter, expanding on his claim that bilingualism is sustained only so long as
it has practical – chiefly economic – value, Edwards quotes directly from my work (Dorian
1982: 47) in a way that makes it appear as though I were in agreement with his position
(Edwards, 1994: 116). It is the unfortunate omission of a lead-in sentence and a following
sentence, both strongly qualifying the quoted material, that gives rise to that misleading
impression.
While the study of minority peoples and their language choices has made
the pattern of shift from a minority language to a majority-group language
familiar, this pattern is more likely to occur under some circumstances than
others. Paulston notes, for example, that the opposite pattern, sustained group-
wide bilingualism, is unusual only under particular conditions, namely when
the modern nation-state is the setting and when both socioeconomic incen-
tives and access to the dominant language are present:
Environments that favor the maintenance of multiple languages are not diffi-
cult to find, but they are reported largely from non-European cultural contexts
where quite different language ideologies prevail. Since as noted the reporters
are in most cases members of Western societies, there is sometimes a strik-
ing contrast between the expectations of the Western investigator and the
linguistic situation encountered and described. This is the case, for example,
with a 1998 field report by English researcher Roger Blench on two languages
that he searched out during a field trip in Plateau State, northern Nigeria. He
found the Niger-Kordofanian language Horom spoken in one main village by
a maximum of perhaps 1,500 people, while the inhabitants of nearly all the
other villages in the area spoke varieties of a Chadic language called Kulere. He
described the Horom people as “extremely multilingual”, since they reported
themselves to be fluent in Kulere as well as Horom and also in Rindre, another
Niger-Kordofanian language, and still more significantly in Hausa, an impor-
tant and very widely spoken Chadic language of northern Nigeria and the
Niger Republic. Kulere and Rindre are both spoken by larger populations than
is Horom (Rindre by a considerably larger population), while Hausa has many
millions of first-language speakers in Nigeria and the Niger Republic and also
increasingly serves as a lingua franca in West Africa. Blench was surprised, by
his own account, to find that Horom did not appear to be an endangered lan-
guage: children present during his language elicitation session “were able to
produce the required lexical items simultaneously with the adults” (Blench
1998: 10). He suggested that remoteness might have acted as a buffer in the
maintenance of Horom, though he acknowledged that remoteness had not
sufficed to keep Chadic languages in the Bauchi area of Nigeria from disap-
pearing. He remarked, too, on the cultural vitality of the Horom people, whose
traditional religion, pottery, weaving, and music he found likewise to persist
strongly.
The linguistic tenacity of another small and isolated Niger-Kordofanian-
speaking people in Plateau State, the residents of Tapshin village, struck
Blench still more strongly. He gives the following account of their vigorous
Nsur language:
On the face of it, Nsur should be a prime candidate for language loss. All
adults appear to be fluent in Ngas and Hausa and Tapshin is an enclave
within the Ngas, a numerous population speaking a Chadic language, by
whom they are culturally dominated. The number of speakers [of Nsur]
cannot be more than 3–4000. . . . However it was apparent during the
interviews that even young children are learning the language and there
is no evidence of a decline in competence. Even more surprisingly, but no
doubt related, the language is by no means full of loanwords from Hausa
and Ngas . . . (1998: 11).
for urban populations or the educated elite, then adds another layer to the
multilingualism.
The differences among the languages in question are not by any means minor
dialectal features. According to Haiman, Gimi, Siane, and Chimbu are impres-
sionistically as different from Hua as French, German, and Russian are from
English.
Similar patterns of exogamous marriage and resultant multilingualism
among all or nearly all group members have been reported from the Vaupés
region of the Amazon basin and from Burma. A linguist working in Burma tells
of meeting exogamous Kachin in Burma “who can converse happily in at least
half a dozen languages, with native knowledge of three or more”, though he
relates their profound multilingualism more particularly to a low place in the
regional linguistic hierarchy, noting that speakers of another low-position lan-
guage (Lisu) can be equally multilingual (Bradley 2001: 155).
ism from West Africa indicates that the multilingualism in question can be
profound indeed, even to the point of making the concept “native language”
moot. In an unusual dual presentation of the fieldwork experience, linguist
Fiona Mc Laughlin offers her own account of work on noun classes in three
Niger-Congo languages of Senegal – Wolof, Pulaar, and Seereer – side by side
with a parallel account from her principal Pulaar teacher, Thierno Seydou
Sall (Mc Laughlin and Sall 2001). At one point Mc Laughlin declined an offer
from Sall, whom she had found to be a natural linguist with remarkable abil-
ity to analyze the structure of Pulaar, to work on the noun classes of Wolof
with her, in addition to their work on the Pulaar noun classes, because she
categorized him as a native speaker of Pulaar. Sall gives this account of the
interaction:
Sall notes that because he speaks the “deep Wolof” of the heartland he knows
the noun classes of Wolof better than a good many Wolofs who live in towns
and cities, where they mix with non-Wolofs a great deal, and that he therefore
sometimes corrects Wolof people’s Wolof, including their noun classes.
Mc Laughlin, reflecting on her experience with multilingual Senegalese like
Sall in the Sahelian town of Fatick, where she lived and worked for a year and
where Sall lived as part of a large traditional family, revised her notion of what
it meant to be a native speaker of a language:
Mc Laughlin is not the only linguist of European descent to have allowed field-
work practices to be constrained by the notion that an individual has one, and
only one, native language. A similar case is reported from Australia, where
a researcher realized belatedly that he had probably let precious chances to
gather data on an endangered Aboriginal language slip away because he did
not take the opportunity to work with profoundly fluent speakers who were
officially native speakers of some other Aboriginal language (cited in Evans
2001: 255–56). The passage from Bradley quoted above is notable in attributing
“native knowledge” of three or more languages to multilingual Kachin speak-
ers. Bradley does not say whether he did or would accept such speakers as full-
fledged data sources for more than one language, but just such work needs
to be done by way of plumbing the full capacities of profoundly multilingual
individuals.
Apart from the considerable practical benefits bi- or multilingualism may con-
fer (with exogamous marriages and manipulable identities included among
such practical benefits), additional benefits that qualify at least in part as ideo-
logical are recognized among peoples who have traditionally cultivated knowl-
edge of more than one language. Among some such peoples, multilingual skills
are regarded as a sign of intellectual or cultural superiority. In the same general
area of the New Guinea Highlands where Hua is spoken, a variety of Siane
known as Komunku was spoken by people whose neighbors spoke a variety of
Dene. While members of the Emenyo tribe spoke Komunku and often spoke
Dene as well, Dene speakers less frequently spoke Komunku. An anthropolo-
gist who worked with Emenyo in the 1950s and ’60s summed up Emenyo atti-
tudes as follows (Salisbury 1972 [1962]: 56):
The fact that there are more Emenyo bilingual in Dene, than Dene-
speakers who are bilingual in Komunku is not associated with any feeling
among the Emenyo that they are politically less important or that their
language is inferior to Dene. Bilingualism is treated as a desirable accom-
plishment and their command of Dene makes them, if anything, superior
to the Dene.
Where the exogamous multilingual Indians of the Vaupés region are con-
cerned, one observer (Sorensen 1972) described them as having an “instru-
mental and practical” orientation to multilingualism. But he noted at the same
time that multilingualism was actively cultivated, as with the Emenyo, and
cultivated furthermore across an entire lifetime despite universal command
of a regional lingua franca. Children, typically fluent speakers from their early
years of both their mothers’ and their fathers’ languages and also of Tukano,
the regional lingua franca, not only acquired during their adolescence several
other languages spoken in the community’s longhouse, but might go on to
acquire still more in adulthood. Nor did the learning process stop there: “as he
[a longhouse resident] approaches old age, field observation indicates, he will
go on to perfect his knowledge of all the languages at his disposal” (Sorensen
1972: 86). The degree to which expert knowledge of additional languages was
actively pursued indicates an ideological orientation to multilingualism that
ultimately transcends the strictly practical level of which Sorensen spoke.
9 Receding Multilingualism
2 Aikhenvald (p.c., 7 Feb. 02) heard older Salesian missionaries make remarks to this effect in
conversation. She notes that a younger generation of Salesians is now active in promoting
indigenous languages, but for small language groups such as the Tariana the change in atti-
tude and policy most likely comes too late.
by older kinsfolk and other community members was curtailed by this edu-
cational experience, and in addition an influential new reference group was
formed for the young people, one that consisted of English-speaking age-mates
with off-reservation experience. Kroskrity found young people signaling their
allegiance to this new reference group by replying in English when their elders
spoke to them in Arizona Tewa. The young people had become aware of the
economic disadvantages of reservation life and of the possibilities of material
advantage elsewhere, through exposure to mass media and also through their
time in urban boarding schools. In Kroskrity’s view socioeconomic factors were
primary in motivating young people’s increased use of English, but because the
actual economic possibilities for young Arizona Tewa in off-reservation set-
tings are extremely limited, he considered it possible that an unrewarding ven-
ture into off-reservation employment might ultimately provide an incentive
for return to a more traditional reservation life. The future for Arizona Tewa –
maintenance or loss – could depend, consequently, on whether the preference
for English among young people proved to be a life-cycle phase, with a later
reaffirmation of Arizona Tewa traditional values (including the language), or
whether it proved to be permanent (Kroskrity 1993: 103–05).
Post-adolescent decisions to stay in the home area after all, and therefore
also to maintain the ancestral language, have in fact been known to change the
apparent outlook for a small group and their language. In the 1970s a linguist
working in an enclaved village in southern Italy, where a Francoprovençal lan-
guage had been spoken for more than 500 years, predicted on the basis of his
census of bilingual speakers of various ages, and also by his observations of lan-
guage use, that Faetar, the local Francoprovençal language, would be dead by
the year 2000. The villagers themselves expressed the same opinion. Another
linguist, arriving to work in the same village in the 1990s, found the villagers as
bilingual as ever, but she heard them give much the same prediction: Faetar
would be dead in twenty years. Because the earlier linguist had done his census
by age groups, the second linguist was able to establish that adolescents who
were using Italian almost to the exclusion of Faetar in the 1970s were adults
using Faetar among themselves and with their children in the 1990s. The ado-
lescents of the 1990s, however, were again using mostly Italian by preference,
and their language preference was again giving rise to fears that the language
would soon cease to be spoken (Nagy 2000: 128–29). The most serious threat to
Faetar appeared to lie in a gradual shrinkage of village population as more peo-
ple left in pursuit of greater economic opportunity, a depopulation common
to agricultural regions of Italy. Lack of sufficient population to keep the village
viable remains a future possibility, therefore, but lack of adolescent speakers as
such is to some extent an age-graded phenomenon in Faeto. Awareness of such
of nearly all the many minority languages spoken under its rule (Smalley 1999:
341–49).
Acquisition of a second (or third or fourth) language need not imply loss of
a first language. The ideologies and behaviors of two populations are relevant
in this matter. Not only do the villagers of Tapshin continue to speak Nsur, even
though they also speak Ngas and Hausa, but the numerically and culturally
dominant Ngas speakers who surround them apparently have no objection to
the continuing use of Nsur in their midst. The relevance of dominant-group
attitudes becomes more obvious, perhaps, when the Arizona Tewa case is con-
sidered. The Hopi, a somewhat larger population and a well-established one by
comparison with the Tewa who took refuge among them, found it acceptable
for the Tewa not only to learn Hopi but also to continue to use their original
language during the roughly 300 years of their residence on the Hopis’ First
Mesa. Paradoxically, anglophone America, with an overwhelming numerical
dominance and a uniquely secure official language, exerts serious pressure on
the Arizona Tewa (as on all other minority groups) for an assimilation that
includes abandonment of the ancestral language as well as adoption of the offi-
cial language. Hundreds of years of enclavement among a numerically supe-
rior group, and even a degree of Tewa-Hopi intermarriage, did not threaten
the survival of the Tewa language during its earlier history in Arizona; yet it is
possible that boarding-school education in English is now doing just that.
Greatly increased pressure for a shift to dominant-language monolingual-
ism in many parts of the contemporary world seems likely to rest in part on
continuing dissemination, despite he decline of colonialism, of European lan-
guage ideologies: above all the one-language, one-nation ideology associated
with nationalism and the “ideology of contempt” for subordinate peoples, but
also European notions that the languages of peoples who exhibit low techno-
logical development must necessarily be equally limited, while the languages of
peoples who have achieved politically and technologically dominant positions
must necessarily be superior linguistic instruments (Dorian 1998). This self-
serving view justifies the established ascendancy of a few “superior” European
languages in a post-colonial world still linguistically shaped by European colo-
nialism, while simultaneously rendering the displacement of innumerable
“lesser” languages unimportant. Still other elements of European linguistic
ideology have been identified in the pressure for monolingualism, for example
the notion that acquisition and use of an ancestral minority language early in
life is deleterious to full competence in the official language. As one researcher
memorably put it, in connection with the pressure exerted on children from
minority groups in far northern Norway, Sweden, and Finland to make as rapid
a shift to the national language as possible, “it was believed that a child’s head
(especially a minority child’s head) would not have space for two languages”
(Huss 1999: 129). Huss follows other scholars in tracing such attitudes to the
influence of seriously flawed bilingualism research that supposedly detected
lower intelligence and inferior linguistic and cognitive development among
bilingual children, especially the children of late nineteenth-century and early
twentieth-century immigrants to the U.S.3 Certainly such notions have been
communicated to minority-group parents, both among immigrant popula-
tions and among indigenous peoples, with some of those who have gone over
to raising their children with the majority language explaining their decision
in such terms.
Since the 1970s ethnolinguistic vitality studies have attempted to account for
observed cases of shift or maintenance and to predict how likely a minority
population will be to maintain its language or to give it up. Three sets of objec-
tive factors were originally the chief focus of evaluative efforts: status factors
(economic, social, and sociohistorical factors, and status within and without
the group), demographic factors (proportion of the overall population, con-
centration of the minority population, birthrate, etc.), and institutional sup-
port and control factors (use in mass media, education, government services,
industry, religion, culture, politics). If the “surprisingly” vigorous Nsur lan-
guage is taken as a test case, these factors are clearly inadequate for predictive
purposes. The Ngas language was said to be spoken by many more people than
Nsur, and the Ngas were also described as culturally dominant. Whether Ngas
had any institutional support is not evident in Blench’s report, but Nsur almost
certainly did not. By these measures Nsur speakers ought to be in the process
of shift to Ngas, just as Blench expected them to be.
Among the various efforts to refine ethnolinguistic vitality assessment, the
approach with perhaps the greatest potential for uncovering the roots of main-
tenance or shift in a non-Western setting like Tapshin village is the develop-
ment of a Beliefs on Ethnolinguistic Vitality Questionnaire (Allard and Landry
1986). This is not to say that an oral version of the questionnaire as such would
be a suitable instrument for research in Tapshin and its neighboring villages
(an unlikely prospect, since questionnairing is a research procedure that
does not always travel well, culturally speaking), but rather that examination
3 Huss (1999: 129) cites Peal and Lambert, 1962: 1–2 and Skutnabb-Kangas, 1981: 222 in this
connection.
Despite the weight of the many factors that favor language shift, some counter-
currents are observable. One pattern of maintenance involves early shift with
later reversion to traditional cultural and linguistic behaviors, in the reactive
pattern that Kroskrity considered a possibility for Arizona Tewa young people.
Initially, contact with the dominant society, especially educational contact,
appears to open a route to economic advance. But where minority peoples are
visibly distinguishable from members of the dominant group (and have often
acquired a distinctive local version of the dominant language besides), job
scarcity combined with a lingering racism can limit advancement and assimi-
lation. This is said to have happened already among the Kwak’wala Indians of
Vancouver Island: the Kwak’wala young people have shifted entirely to English,
but without reaping the material benefits that the shift seemed to promise
(Anonby 1999: 35). In a study of the outlook for the survival of New Caledonian
vernaculars, Schooling found that Melanesian New Caledonians educated
in French with an expectation of job opportunities that did not materialize
settled back into traditional life patterns in their villages of origin, where kin
networks remained strong and they had kin-based claims on land. Schooling
reported disillusionment among young New Caledonians whose parents had
emphasized the acquisition of French more than that of the local vernacular, to
the point where some young adults were reversing the pattern and consciously
speaking the vernacular with their own children even though husband and
wife might frequently use French between themselves (Schooling 1990: 51–52).
Another pattern favoring maintenance can appear when a group succeeds
in achieving some measure of hoped-for economic advance, whether through
education or through development of resources (e.g. scenic resources for
tourism or sport), before shift to a dominant language is complete. Economic
success can produce enough psychosocial confidence among still-bilingual
speakers to encourage assertion of their ethnolinguistic identity and of the
right to use their heritage language more widely, including in education, as has
happened in Wales, for example.
Also favorable to ethnic self-assertion are wider recognition of linguis-
tic human rights in some parts of the world and increased communication
(especially electronic communication) among widely separated small peoples.
Languages are sometimes viewed as endangered even when nearly all group
members are still speakers. If dominant-group assimilative pressures are seen
to be rising while resistance on the part of a minority group would seem to
require resources that are not in evidence (such as a large population, a strong
ethnolinguistic identity, a viable and locally-based means of subsistence), long-
term survival of the minority-group language can not be taken for granted. If
a shift has already begun – if young children prefer to use a school-acquired
dominant language with one another, for example – the future of the language
begins to look questionable. If lack of ancestral-language knowledge reaches
into the ranks of young parents who are raising children, the outlook becomes
a little darker still.
In a number of shift-prone settings some young parents can be found going
against the tide and taking pains to raise their children in and with the minority-
group language, as has occurred for example in Scotland and Nova Scotia
(Scottish Gaelic), in Finland (Sami), and in Hawai’i (Hawaiian). Among these
parents accomplished learners are often to be found, either individuals of the
relevant ethnic group who were not raised with the ancestral language them-
selves or outsiders who have married into the ethnic group and learned the
traditional ethnic language. Speakers of both sorts value the ethnic language
the more for having had to acquire it effortfully, and with infants of their own
they progress from being dedicated learners to being dedicated transmitters.
It would take a good many fluent parents deeply devoted to home trans-
mission to produce a numerically significant number of fluent new minority-
language speakers, however, and many endangered-language communities
turn to schooling, in particular to immersion schooling, for the relatively rapid
multiplicative effect it can produce: a handful of dedicated and well-trained
teachers, using only the minority-group language in the classroom, can pro-
duce scores of new minority-language speakers over a period of several years.
Such immersion programs have by now amply demonstrated their success,
especially in the cases where primary immersion schooling is followed by sec-
ondary immersion schooling.
References
Blench, Roger. 1998. Recent fieldwork in Nigeria: Report on Horom and Tapshin.
Ogmios: Newsletter of the Foundation for Endangered Languages 9, 10–11.
Blommaert, Jan, & Verschueren, Jef. 1998. The role of language in European nation-
alist ideologies. In Bambi B. Schieffelin, Kathryn A. Woolard, & Paul V. Kroskrity
(eds.), Language ideologies: Practice and theory (pp. 189–210). New York and Oxford:
Oxford University Press.
Bradley, David. 2001. Language attitudes: The key factor in language maintenance.
In Osamu Sakiyama & Fubito Endo (eds.), Lectures on endangered languages: 2
(pp. 151–60). Kyoto: Endangered Languages of the Pacific Rim.
Dauenhauer, Richard, & Dauenhauer, Nora Marks. 1998. Technical, emotional, and
ideological issues in reversing language shift: Examples from Southeast Alaska.
In Lenore A. Grenoble & Lindsay J. Whaley (eds.), Endangered languages: Current
issues and future prospects (pp. 57–98). Cambridge: Cambridge University Press.
Dorian, Nancy C. 1981. Language death: The life cycle of a Scottish Gaelic dialect.
Philadelphia: The University of Pennsylvania Press.
———. 1982. Language loss and maintenance in language contact situations. In
Richard D. Lambert & Barbara F. Freed (eds.), The loss of language skills (pp. 158–
67). Rowley, Massachusetts: Newbury House.
———. 1998. Western language ideologies and small-language prospects. In Lenore A.
Grenoble & Lindsay J. Whaley (eds.), Endangered languages: Current issues and
future prospects (pp. 3–21). Cambridge: Cambridge University Press.
Edwards, John. 1994. Multilingualism. London and New York: Routledge.
Evans, Nicholas. 2001. The last speaker is dead – long live the last speaker! In Paul
Newman & Martha Ratliff (eds.), Linguistic fieldwork (pp. 250–81). Cambridge:
Cambridge University Press.
Flinn, Juliana. 1992. Diplomas and thatch houses: Asserting tradition in a changing
Micronesia. Ann Arbor: University of Michigan Press.
Grillo, R.D. 1989. Dominant languages: Language and hierarchy in Britain and France.
Cambridge: Cambridge University Press.
Haiman, John. 1987. Hua: A Papuan language of New Guinea. In Timothy Shopen (ed.),
Languages and their status (pp. 35–89). Philadelphia: University of Pennsylvania
Press, for the Center for Applied Linguistics.
Huffines, Marion Lois. 1989. Case usage among the Pennsylvania German sectarians
and nonsectarians. In Nancy C. Dorian (ed.), Investigating obsolescence: Studies
in language contraction and death (pp. 211–26). Cambridge: Cambridge University
Press.
Huss, Leena. 1999. Reversing language shift in the far north: Linguistic revitalization in
northern Scandinavia and Finland. Uppsala: University of Uppsala.
Kroskrity, Paul V. 1993. Language, history, and identity: Ethnolinguistic studies of the
Arizona Tewa. Tucson: University of Arizona Press.
MacEachern, Scott. 2002. Residuals and resistance: Languages and history in the
Mandara Mountains. In Brian D. Joseph, Johanna Destefano, Neil Jacobs, & Ilse
Lehiste (eds.), When languages collide: Perspectives of language conflict, language
competition, and language coexistence (pp. 21–44). Columbus: The Ohio State
University Press.
Mc Laughlin, Fiona, and Sall, Thierno Seydou. 2001. The give and take of fieldwork:
Noun classes and other concerns in Fatick, Senegal. In Paul A. Newman & Martha
Ratliff (eds.), Linguistic fieldwork pp. 189–210. Cambridge: Cambridge University
Press.
Nagy, Naomi. 2000. Fieldwork for the new century: Working in Faeto, an endangered
language community. Southern Journal of Linguistics 24, 121–36.
Paulston, Christina Bratt. 1994. Linguistic minorities in multilingual settings: Implications
for language policies. Amsterdam/Philadelphia: John Benjamins.
Peal, Elisabeth & Lambert, Wallace. (1962). The relation of bilingualism to intelligence.
Psychological Monographs 76, No. 546.
Salisbury, R. F. 1972. Notes on bilingualism and linguistic change in New Guinea. In
J. B. Pride & Janet Holmes (eds.), Sociolinguistics: Selected readings (pp. 52–64).
Harmondsworth: Penguin Books. [Reprint of Anthropological Linguistics 4, 1962,
1–13.]
Sankoff, Gillian. 1980. Multilingualism in Papua New Guinea. The social life of language
(pp. 95–132). Philadelphia: University of Pennsylvania Press.
Schooling, Stephen J. 1990. Language maintenance in Melanesia: Sociolinguistics and
social networks in New Caledonia. Arlington: The Summer Institute of Linguistics &
the University of Texas at Arlington.
Skutnabb-Kangas, Tove. 1981. Tvåspråkighet. Lund: Liber Läromedel.
Smalley, William A. 1999. Linguistic diversity and national unity: Language ecology in
Thailand. Chicago and London: University of Chicago Press.
Sorensen, A. P., Jr. 1972. Multilingualism in the northwest Amazon. In J. B. Pride & Janet
Holmes (eds.), Sociolinguistics: Selected readings (pp. 78–93). Harmondsworth:
Penguin Books.
Williams, Glyn. 1992. Sociolinguistics: A sociological critique. London & New York:
Routledge.
Woolard, Kathryn A. 1998. Language ideology as a field of inquiry. In Bambi B.
Schieffelin, Kathryn A. Woolard, & Paul V. Kroskrity (eds.), Language ideologies:
Practice and theory (pp. 3–47). New York & Oxford: Oxford University Press.
For East Sutherland Gaelic (ESG hereafter), the competition comes not from
any more vigorous variety of Gaelic, but from English. In the local context the
two languages, English and Gaelic, are in competition with one another, and I
will therefore speak of ESG as a language approaching extinction, even though
it is actually a regional dialect of Gaelic which is fading from the scene; there
are certainly other forms of Gaelic which will survive the approaching demise
of ESG. As would be expected, the competition between ESG and English has
been a very unequal one. English has long had the unwavering support of the
national state, so that education has been solely in English since the state took
it on, and local people’s experience of military service has also meant the use
of English, in modern times, as has any contact with the court system or local
administrative services. The one social institution which has favored Gaelic in
the 20th century has been the church, with some denominations (all Protestant
in this region) persisting longer in the use of Gaelic than others.
It was the long-continued use of Gaelic in religious life which prevented
the language situation among the ESG-speaking fisherfolk from becoming one
of bilingual diglossia during the first half of the twentieth century, in fact. So
long as Gaelic remained the language of church services, scripture readings,
psalm singing, and praying, it could not be said that English had usurped all
H[igh] language functions or that Gaelic had retreated to L[ow] language func-
tions. By the early 1960s, when I first arrived in East Sutherland to begin study-
ing the dialect and its setting, only one village still had Gaelic church services
available. All of the Gaelic speakers who were of fisherfolk background (which
means all the indigenous Gaelic speakers who remained, by that time) had had
their formative religious experiences through the medium of Gaelic, however.
Active control of religious language was largely restricted to men, since only
males were trained to precent (line out) the psalms during services, and only
men had been expected to lead prayer spoken out loud in household settings,
at wakes, and so forth. Only men could serve as elders and thus take a limited
leadership role in congregational life. (Ministers, too, were exclusively male,
but since there were never any ministers of local origin this offered no verbal
role for speakers of ESG.) As the number of men who could precent the psalms
in the one remaining church with Gaelic services declined, a relatively young
woman was sometimes persuaded to precent when no male precentor was
available (though she did so from her pew, declining to take the precentor’s
seat facing the congregation at the front of the church). She had a strong voice
and was self-taught, but her example at least indicates that women could some-
times take a prominent role in using the rather archaic (and very nonlocal)
Gaelic of religious usage, at least in the fixed and familiar language of the met-
rical psalms.
With the retirement of the last Gaelic-speaking minister in the late 1960s or
early 1970s, Gaelic ceased to play any regular formal and public role for Embo
villagers; it had ceased to play any such role for members of the other two for-
mer fishing communities of East Sutherland long since. Because it seemed
neither appropriate nor congenial to explore merely as an observer the use of
Gaelic in people’s private devotional lives, I never tried to investigate the active
control of religious language among the more than forty fluent speakers with
whom I worked. In their work with me a few men demonstrated some control
of Biblical vocabulary not in everyday use, and I noted also some occasional
use of the language of benediction in formal partings among fluent speakers
not in regular contact with one another (including the closing of tape-recorded
messages). It seems likely that some individuals used Gaelic in private prayer,
and some men were known as good psalm singers. It is possible that some few
men read the scriptures in Gaelic.
For most ESG speakers in the latter half of the twentieth century, and in
particular for most women speakers, Gaelic was the language of hearth and
home. There simply were no public spheres in which their native form of
Gaelic could appropriately be used. The sole remaining public spheres for any
Gaelic at all in these villages, when I began to do research there in the 1960s,
were the church (for Embo village only) and the ceilidh. Ceilidhs were eve-
ning functions at which admission was charged, with performances of vocal
and instrumental music, Highland dance, and sometimes also amusing mono-
logues; they were usually held to benefit some cause or organization. For each
ceilidh there was a fear-an-taighe (master-of-ceremonies), and if possible a
Gaelic speaker served in this role. It was customary for any Gaelic-speaking
fear-an-taighe to use some Gaelic, at least ceremonially (e.g., in welcoming the
audience at the outset and in thanking the performers at the close), during the
course of the evening. I am not aware of any Gaelic speakers native to the three
fishing communities who served as fear-an-taighe, any more than as minister,
however; these were roles performed by outsiders to the local communities,
and consequently the Gaelic used was not the local variety. All the same, the
Gaelic spoken from the platform at ceilidhs exposed local speakers to a rela-
tively formal use of the language in a public function, as church services also
did for those who still had them available. These local uses of a more formal
Gaelic had some personal reality for ESG speakers, unlike the limited use of
Gaelic in broadcasting, but it must be stressed that any variety of Gaelic used
by a minister, a fear-an-taighe, or a broadcaster would inevitably be too differ-
ent from the local East Sutherland variety to serve as a model for local imita-
tion. ESG is radically deviant, from the point of view of the standard language,
and the fit between ESG and any variety used in public spheres is too poor to
allow any use of the other variety by local speakers beyond the odd adoption
of lexical items or turns of phrase. ESG speakers who attempted more than this
(and some few occasionally did) quickly came to grief through their inability
to sustain the performance over more than a few phrases or sentences.1
When a language has retreated to the private sphere exclusively, as ESG has,
there may well be some question about its stylistic range. If it is never used to
make a formal speech, to introduce the speaker at a meeting, to give the vote
of thanks to a guest performer or speaker at a public function, to debate the
agenda of a local organization, to make a motion during a public meeting, to
make an announcement of coming events, and so forth, how much capacity
will it maintain to express stylistic differences at all?
My impression, from participating repeatedly over a fifteen-year period in
some of the daily-life activities of fluent ESG speakers who became my friends,
was that there remained a useful range of styles available to speakers, though
a limited range certainly by comparison with that available to speakers of lan-
guages which are used to more public and formal purposes than ESG. In this
chapter I try to substantiate my impression of the stylistic flexibility of ESG
from tape-recorded but freely spoken data drawn from the single ESG speaker
who was most available to me, my landlady over the entire fifteen-year stretch
of my recurrent East Sutherland fieldwork.
Though it is the best I have available, this body of data is very far from ideal
for the purpose of demonstrating the fullest possible range of styles in ESG. For
one thing, it all derives from a single kind of verbal event, a very lightly guided
variety of interview.2 For another, it derives from a single main participant in
1 I should note that this is just as true of me as of other ESG speakers. Though I’ve often tried,
in unavoidable interactions with speakers of more standard varieties, and especially with
Gaelic intellectuals, to make some active use of my passive and partial knowledge of stan-
dard Gaelic, I’ve been unable to sustain it. Since I have a large advantage over most ESG
speakers in passive literacy and in conscious knowledge of the structure of the standard lan-
guage, my own failures in this respect make me deeply aware of the difficulties of adopting
nonlocal models in active Gaelic use.
2 I term these speech events interviews in recognition of the fact that two people were present,
one of whom operated a tape recorder and sometimes asked questions or made comments.
In actual fact one speaker did the vast majority of the talking, chose the topic a good part of
the time, and was relatively seldom interrupted. Linguists sometimes present themselves
as power figures in any situation involving their use of a tape recorder (Briggs 1986: 89, 120)
and in theory I could of course have terminated the session by turning off the tape recorder;
but the reality of these sessions is that the other participant was the dominant party in most
respects: she controlled the language being spoken far better than I, she chose most of the
that verbal event, with one unchanging lesser participant. These are certainly
not ideal circumstances for plumbing the far reaches of style change. There
are some redeeming factors, nonetheless. My landlady, whom I will call MMK,3
was profoundly illiterate [in Gaelic].4 She was also remarkably uninclined to
attempts at shifting in the direction of more standard Gaelic forms, which was
striking in view of the fact that she had lived for a few years early in her mar-
ried life on the west coast of Wester Ross, where she had heard a good deal
of the Gaelic dialect (more nearly standard and definitely more prestigious)
native to that region. She can be said to represent genuinely local ESG speech
norms. She was also a wonderfully uninhibited speaker. That is, she was not
only unpretentious in sticking strictly to local Gaelic usage, she was also much
more impervious than most local speakers to the scorn of speakers of other
dialects and to the disdain of many English monolinguals for speakers of Gaelic
generally and for speakers of the local Gaelic particularly. She spoke Gaelic
with pleasure and relish, and she was willing to speak it with normal audibil-
ity on the street, something a good many local women avoided. She liked to
talk, in either of her languages, and had a great appreciation for conversation,
gossip, and the recounting of stories. She was, in short, a great talker, harder
to persuade to silence than to prompt to exuberant volumes of speech. For a
linguistic fieldworker she had another splendid attribute: she had no qualms
at all about being tape-recorded. Because of this, and because she and I lived
under the same roof so often, she is better represented in my field recordings
than most other speakers. And though I have perhaps as much material for two
others, no other speaker romped so enthusiastically through as many stories
and reminiscences as she. Consequently it was to the material which she spoke
on tape for me that I turned when I became interested in stylistic variation,
and her freely spoken tape-recorded material forms the basis for this study.5
topics for presentation, she held the floor uninterrupted the majority of the time. Where I
affected the topic it was by asking that she repeat for the tape recording a story which she’d
already told me. Where I affected the form of the story it was generally by interrupting to ask
for more information about something she’d said.
3 She is G2, that is, the second-oldest of my Golspie village sources, in many of my other pub-
lications. That seems an excessively impersonal designation for the present paper, however,
where the individual aspects of her speech are at issue.
4 The adverb reflects the fact that she could not so much as recognize written Gaelic when she
saw it; she used to give me the weekly radio bulletin so that I could tell her which program
titles were Gaelic as opposed to French, because both looked equally unfamiliar to her.
5 I was not interested in register variation as such at any point in my original study of ESG and
made no effort to gather materials which would be useful for that purpose. By the time I was
interested, I was prevented by health problems from returning to East Sutherland to collect
material specifically to that end. I have not been able to visit East Sutherland since 1978, and
most of the fluent speakers with whom I worked (all, in the villages of Brora and Golspie)
have died since then.
6 She provided other tape-recorded material on other occasions, including songs, phonologi-
cally contrastive words and phrases, and two sections of a letter-tape, but these three ses-
sions contain all of her narrative material.
7 This session takes less time on tape than it took in the event, because I stopped recording
each time I myself spoke. The tape was made late in 1963–64, my first year in East Sutherland,
and my Gaelic was not yet comfortable. I evidently preferred not to have my own efforts
at Gaelic interviewing immortalized on the tape, since I stopped the recording whenever I
spoke and restarted it as MMK began to speak.
8 A question from me interrupted narrative 5 before it had reached its natural conclusion.
eight such narratives, two in the first session and six in the second. Under short
titles based on their themes, I offer brief summaries here:
9 The word counts are approximate. ESG is an unwritten dialect and though I have sometimes
contrived to write it in Gaelic orthography for publication purposes, I do not normally, or
willingly, do so. All of my ESG tape transcriptions are in quasi-phonemic renderings, and in
many cases I followed phonological boundaries rather than those of the traditional orthogra-
phy. For example, the expression bha aid ‘they were’ has a single long vowel with no rearticu-
lation in ESG: [va:ǰ]. I always wrote it as a unit in my field notes, and I reckoned it as a single
word in making the word counts for MMK’s narratives here. There are so many such cases,
all involving short words, mostly of high frequency, that I did not trust myself to take these
phrases apart with any consistency while counting. Consequently the word counts given
here are all lower than they would be if someone more comfortably literate in Gaelic than I
were to translate the texts into written ESG, following the word-division conventions of the
standard written language.
all the villages give her presents when she marries locally, but when the
marriage is dissolved within a short time, she has the gifts auctioned off.
8. The defective Gaelic RECORD (214 words). The instructional record which a
new boarder is using to help him learn Gaelic seems to MMK’s ears to be
teaching him incorrect Gaelic.
on loanwords from English. One literate male Golspie speaker replaced the
nearly universal /phɔlisxən/ ‘policemen’ with /ɫuxkh ən ɫə/ ‘people of the law’,
for example, when taping a narrative for me, and a male Embo speaker who
needed to use the word /thramph/ ‘tramp’ for his narrative carefully prefaced
it with /mər ə xanu aǰ/ ‘as they would say’. In view of all this I expected to see
some change in MMK’s handling of loanwords as she moved from the sober,
didactic telling of her opening story to more relaxed stories in the later session.
In analyzing the use of loanwords in MMK’s eight narratives, I looked at the
following features: (1) the sheer number of recognizably English loanwords rel-
ative to the total number of words in the story, (2) phrasal switches to English,
(3) use of loanwords in markedly Gaelic or markedly English fashion. Within
the third category I looked in particular at the application of the initial con-
sonant mutations so characteristic of Gaelic12 to loanwords, in environments
where a native Gaelic word would necessarily show mutation, and at the use of
inflectional morphology with obviously English loanwords. Where inflectional
morphology was concerned, the issue turned out to be whether an obviously
English loanword was made plural by means of a Gaelic plural allomorph or
an English plural allomorph, since only the plural proved to be variable; all
other inflections used with loanwords were uniformly Gaelic. Switches to a
wholly English phrase within the Gaelic narratives were also tallied, since it
seemed likely that self-consciousness about dependence on English lexicon
would extend a fortiori to phrasal switches. Table 17.1 shows MMK’s handling of
obviously English material in her eight narratives.
There is no indication in MMK’s narratives that she is trying to limit her use
of obviously English loanwords as such, since all her narratives have them; nor
that a low percentage of such loanwords correlates with the relatively low-key
delivery of her first two narratives, since narratives 4 and 5 have a lower percent-
age than even the first, deeply serious, narrative. The handling of the English
material is conservative in the first two narratives overall, however. There are
12 The initial consonants of nouns, verbs, and adjectives can be altered to show grammati-
cal distinctions, as the sole sign of the category in some instances but in conjunction
with suffixal morphology in others. In addition, mutations are sometimes an obligatory
though grammatically nonsignificant feature of certain constructions. Some ESG exam-
ples: /maru a, xə̃ nʹax/ ‘Kill him, Kenneth!’ versus /varu a khə̃ nʹax/ ‘He killed Kenneth’:
/hũnig mi ə pra:r/ ‘I saw her brother’ versus /hũnig mi ə vra:r/ ‘I saw his brother.’ The initial
mutations are pervasive in all of the Celtic languages, and ESG is no exception. Some con-
sonants are simply not susceptible to mutation, however, and of course some loanwords
appear in environments which do not call for mutation, so that this criterion of loanword
adaptation does not apply universally to the handling of every loanword which appears
in the texts, despite the high frequency of initial mutations.
TABLE 17.1 The Handling of Obviously English Loanwords and Phrases in MMK’s Gaelic
Narratives, First or Last Instances of Significant Features Indicated by Dividing Lines
1964 1 3.4 0 of 4 0 of 3 0 –
2 4.7 0 of 2 – 1 1
1968 3 5.3 5 of 13 0 of 2 1 1
4 2.8 – – 0 –
5 3.2 0 of 8 – 1 0
6 7.4 0 of 6 1 of 3 2 0
7 3.8 2 of 10 1 of 7 4 0
8 6.5 0 of 8 – 0 –
13 The sibilant plural of this noun was not counted as an English plural inflection for pur-
poses of Table 17.1; since the noun has no singular, there is no independent base to be
she uses two English-based loanwords, /khru/ ‘crew’ and /anti/ ‘aunt’, without
self-consciousness, but stops immediately after uttering the third, /wikhεnd/,
breaking off her narrative to say to me in English, “What would I say for week-
end?” As I murmur that weekend seems fine to me, she carries straight on with
her story, but it’s just after this that she uses the phrase Golspie holiday and
then immediately offers a Gaelicized version of the phrase. She interrupts her-
self only once again, in these narratives or the later ones, with doubts about her
use of an English word or phrase. In narrative 5, Class, where the acceptability
of her Golspie Gaelic is the specific focus of the story, she uses the loanword
/khwεsčən/ ‘question’ in starting to describe the teacher’s classroom request
for the Gaelic version of an English sentence: /agəs hurd a rəm:əs ə gwεsčən –
/ ‘and he said to me the question, – ’. She then stops and repeats the offending
noun in self-mockery, immediately reformulating the clause with the Gaelic
verb ‘ask’. The borrowed noun was well integrated phonologically, with the
initial mutation (nasalization) appropriate to a masculine noun of its phono-
logical class after the definite article, but it is evidently not suitable for use in
preparing to quote the arrogant Gaelic teacher, even though she uses English
loanwords freely and unselfconsciously later in the same story. This instance
of corrective self-consciousness in 5 (Class) is spontaneous, while the first nar-
rative remains unique in its deliberate introduction of an English loanword,
self-consciously framed with the Gaelic phrase ‘or I could say’.
The handling of obviously English loanwords as an index of self-consciously
careful style was the first differentiating feature which I looked for in the eight
narratives, but it seemed to me that the narratives might also be expected
to differ in terms of features which contributed to their relative liveliness as
stories, if there was a stylistic continuum of any sort among them. As indica-
tors of liveliness I tallied seven features: the number of simple direct quotes
or direct-quote interchanges; the number of those which were in fact inter-
changes rather than simple direct quotes; the number of interchanges which
consisted of more than the minimum two turns; the number of direct quotes
which used /(h)ɔrs/, the more vivid of the two quotative past-tense verbs of
ESG;14 the number of instances of doubled-up quotatives (akin to English
“I sez, ‘_____’, sez I”); the number of strong interjections (i.e., interjections other
than the routine ‘oh’, ‘well’, ‘ach’, or ‘och’); and the number of uses of ‘adventurous’
Table 17.2 Incidence of ‘Liveliness Features’ in MMK’s Gaelic Narratives, First Instance or First
Major Increase Indicated by Dividing Lines
1964 1 1 0 – 0
2 1 0 – 0
1968 3 6 3 – 11
4 1 0 – 1
5 2 2 1 4
6 1 0 – 0
7 1 1 1 7
8 2 0 – 0
Doubled-up Strong ‘Adventurous’
Quotatives Interjections Language uses
# # #
1964 1 0 0 0
2 0 0 0
1968 3 1 1 0
4 0 0 3
5 1 0 0
6 0 2 1
7 0 1 2
8 0 3 1
language (i.e., profanities and indelicate terms for body parts.)15 Table 17.2
presents the results of this tally.
As Table 17.2 indicates, none of the liveliness features of MMK’s narratives
except simple direct quotation appears in the first two of her stories (and only
15 MMK is by no means the most profane or indelicate ESG speaker I’ve heard, but she is
more likely to use profane or indelicate language on tape than most. This probably reflects
both her ease with tape-recording and her particular ease with me as her long-standing
boarder. [See Chapter 19 in this volume for an excerpt from narrative 7 that includes two
instances of “adventurous” language as that term is used here.]
one of those in each story), but four liveliness features occur for the first time
as of the third story. That third story is certainly a very lively one, but even so
it has only one strong interjection and no instances of indelicate or profane
language. The fourth story, a very brief afterthought-anecdote connected with
narrative 3 (Bather), is wholly indelicate in subject matter, and once she has
told that story other ‘adventurous’ usages appear more freely. Of the subse-
quent stories only 5 (Class) is without them.
The combined evidence of Tables 17.1 and 17.2 seems to indicate that MMK
is indeed using a relatively restrained style when she begins the earlier tap-
ing session, which produces the first two of the eight narratives, and that she
opens the later taping session with a more relaxed and casual style. Although
MMK is still sufficiently self-conscious about English material, at the opening
of the 1968 session, to pause and reformulate a phrasal switch to English back
into Gaelic, the third narrative marks the first failures of consonant mutation
with an obviously English loanword in environments where mutation would
be obligatory in a native Gaelic word. Narrative 3 also provides the first inter-
changes, in the use of direct quotes, and the vivid quotative /(h)ɔrs/ not only
appears, but appears in large numbers. The first instance of doubled-up quota-
tive verbs likewise characterizes this narrative. She uses no ‘adventurous’ lan-
guage, but she does for the first time use a strong interjection. Though all but
two narratives after the third have phrasal switches to English, no narrative
after that one shows the reversal of such a switch. Only 5 (Class) among the
subsequent narratives has neither a strong interjection nor any ‘adventurous’
language, but 5 does have a number of lively features associated with direct
quotation (interchanges, use of /(h)ɔrs/, and the other instance of doubled-up
quotatives).
Seven of the eight narratives involve MMK personally; only 4 (Sins) does
not. She is a particularly central actor in the early part of narrative 3 (Bather),
where she is the person who realizes that the bather is in trouble and makes
the initial effort at rescue; and in the first half of 5 (Class), where she is the
local Gaelic speaker who draws the teacher’s scorn by speaking the Golspie
variety of Gaelic. But she is also an actor in five of the six others. (For example,
she is present when the glass shatters in 1, and it is she who runs to buy eau
de cologne as an antidote to the terrible odor from the second body when it
eventually washes up on the shore; she is among the older children who are
supposed to be looking after the little boy in 2 when he wanders off.) Narratives
3 and 5, in which MMK plays a very central role, are especially lively stories, but
so is 7 (Coach), in which she plays a far less crucial role: she takes a dislike to
the Gaelic coach and states her intention to urge the local Gaelic committee
not to engage the coach for the Golspie choir’s tuition, but she is not a major
figure in most of the story. Personal involvement motivates all of the narratives
except 4 (in subject matter an addendum to 3), but degree of involvement does
not fully account for degree of liveliness in the narratives.
It seems reasonable to look at the third narrative as the first in MMK’s more
casual style, which she then maintains to the end of that second taping session
(to which the eighth narrative, Record, provides the close). Strictly in terms of
Tables 17.1 and 17.2, there is nothing to distinguish the first two narratives from
one another except for the absence of any phrasal switches to English in the
opening story. For the listener, all the same, there is one characteristic of the
first narrative which is uniquely its own, a feature I would term an artistry of
repetition which MMK shows nowhere else in this taping session (or for that
matter in the second and third). It lends an almost Biblical flavor to her Gaelic
here, especially since she keeps her lexicon and her syntax very simple. In set-
ting the scene for the breaking of the glass, for example, she relates that she
was sitting with elderly relatives of her mother’s: “Listening to their songs. And
their singing.” She repeats the possessive here, making two phrases out of what
could easily be one (i.e., “their songs and singing”). She does the same sort of
thing several times again in speaking of the rescue attempt and the victims,
after the capsizing of the fishing boat: “And my father was on one, and my uncle
was on the other one” (instead of “on the other,” i.e., of the boats going out to try
to rescue the men in the water); “And they caught the old man, but they didn’t
catch his son”; “And the police came, and they all came”; “but one boot was off
him and the other boot was on him.” In each case the italicized second instance
of a word could have been avoided, and in more ordinary style probably would
have been. The example involving the repetition of the verb ‘catch’ is especially
notable, because that particular verb is regular, so that repeating it produces
identical-sounding verb forms in rapid succession. The more common and less
marked verb here would have been ‘get/find’. But ‘get/find’ is suppletive and in
the preterite is /h/-initial; because initial /h-/ is unstable in this context and
would have disappeared in the second clause, using ‘get/find’ would have pro-
duced less similar-sounding verbs in the positive preterite (first clause) and the
negative preterite (second clause).
One other phrasal oddity appears in this short story, one which resembles
the other overly complete repetitions but is actually still more striking, when
MMK says, “And the old man was drowned, and his son.” Gaelic is a verb-initial
language, and the passive MMK uses to start the sentence would be inflected for
third person plural possessive if she had meant “the old man and his son” to act
as a compound subject of the passive.16 She inflected it for third person singu-
lar masculine possessive, making it clear (as do the suprasegmentals) that “and
his son” is intended to be a second but separate third person singular subject,
agreeing with the third person singular masculine passive (or with an elliptical
second such passive). The phraseology is unusual, and is the more effective
for that reason. In none of the subsequent narratives does MMK use phrasal
repetitions in this way. When she uses repetitions in these other stories, they
are simply expansions and afterthoughts, as is the case in this example from
narrative 7 (Coach), “MacLeod was in the army along with Bob. Along with my
husband. He was in India along with my husband.” There are opportunities to
use the same sort of repetition effect seen in narrative 1 again, but MMK does
not take them. In 3, Bather, for example, a narrative which opens with a near-
tragedy resembling the tragic scene of the first narrative, she tells of pulling an
almost drowned woman from the sea: “And her mouth was full of green foam.
And white. And I caught her, and I was pulling her in, but she was too heavy,
and I couldn’t get her in.” In this passage MMK does not repeat the noun ‘foam’,
to produce “. . . of green foam. And white foam”; and instead of repeating the
verb ‘pull’, as she perfectly well could here, she makes a switch from ‘pull’ to
‘get’ when speaking of retrieving the body from the sea. She is using language
quite differently from the way she used it in the first narrative of the earlier
session, although she is describing a similar scene.
The handling of English loanwords and phrasal switches to English and the
use of liveliness features constitute two sorts of dimensions along which MMK
is able to vary the degree of formality versus casualness in her narratives. By
observing the variations along those dimensions, the listener can distinguish
between the two opening narratives in her first taping session, which are both
relatively formal, and the six subsequent narratives, from the second taping
session, which are more casual. By attending to the high degree of purposeful
phrasal repetition in the first narrative, the listener can recognize in it a more
crafted formality than in the second narrative, even though the second narra-
tive, too, is less casual than those which follow.17
MMK is not the speaker whom I would have chosen if I had been setting
out specifically to explore the fullest range of stylistic variation which an ESG
speaker in Golspie village could muster. I would have opted instead for one of
the three males with whom I worked in Golspie, since I would have expected
them to control some of the lexicon and phraseology of religious usage in ESG,
giving them an outer limit of formal style more extreme than MMK’s. Neither
17 Tannen (1987: 576) recognizes the repetition of a word, phrase, or longer syntactic unit as
part of the “poetics of talk.”
would I have chosen to use two recording sessions with myself as interviewer/
audience as my test case, in exploring MMK’s stylistic range or anyone else’s.
I would have tried to set up situations in which the individual was speaking
to a variety of persons with whom s/he was on more intimate footing or less,
in settings of greater or lesser familiarity, and to a variety of purposes. There
is one advantage to having less than ideal material to work with, however: if
variation in style is identifiable under such conditions, then it becomes that
much the more certain that stylistic variation can reliably be claimed for this
very “domestic” language which is already close to extinction.
It is important to establish the presence of a moderately broad stylistic
range in terminal ESG, because it indicates that natural languages, even when
restricted by decline in functions and domains to the private sphere, and hence
to speech situations which are on the whole informal ones by comparison with
those characteristic of languages in vigorous use across an entire society, can
still be spoken by the fully fluent in ways appropriate to their various stylis-
tic needs. The evidence of MMK’s narratives, all drawn from two same-setting
speech events, does not bear out the “monostylistic” label which Dressler
applies to dying languages, for example, in a recent reference work:
While it is true that a speaker like MMK uses her Gaelic almost entirely for “rou-
tine interactions at home,” it is not true that her Gaelic is “monostylistic.” She
can vary it according to her intention in telling a story, so that a serious story
which is told for purposes of impressing and convincing shows quite differ-
ent stylistic features from a hilarious story which is told for its entertainment
value above all. No doubt her Gaelic is “inadequate for certain speech situa-
tions, domains, and functions.” She probably could not, in Gaelic, pray aloud,
welcome a guest speaker to the Women’s Rural Institute, or repeat the multi-
plication tables. These limitations have a great deal to do with MMK’s range
of activities, however; I doubt that she could easily do any of these things in
English, either, or would undertake to do so willingly. She would (and did, once
I replaced the interviewer whose western dialect she couldn’t understand) do
an interview for broadcast on the radio, and I’m certain that she would have
had no difficulty whatever in lodging a protest in Gaelic about traffic safety for
children, or other such daily-life matters, with the village authorities, assuming
only that she had a speaker of a Gaelic mutually intelligible with hers to lodge
it with. That is, the limits of her Gaelic style range reflect to a considerable
degree the limits of her range of activities and not any drastic poverty of her
Gaelic, which is rich and fluent by local standards.
One of my male sources in Golspie was highly political and quite prepared
to engage in serious political discussions in Gaelic. He was not in late adult-
hood a churchgoer, but though I never heard him pray in Gaelic I suspect he
was capable of it because of the religious upbringing which young males of
his day experienced. He was an intelligent and resourceful speaker in both
Gaelic and English, intellectually inclined despite limited education, and
absolutely fearless in verbal interactions regardless of any differences in social
status between him and his interlocutor. Listening to Gaelic speakers like this
man, and listening to any group of ESG speakers moving from topic to topic
and mood to mood, was more than sufficient to convince me that their Gaelic
should be termed polystylistic rather than monostylistic:
One can locate styles within [both the East Sutherland English and
the East Sutherland Gaelic of the Brora, Golspie, and Embo bilinguals]
which are appropriate to formality or informality, to vulgarity, to humor,
to anger, and the like. Speakers differ in their ability to perform vividly
in these styles, but they can certainly shift in the appropriate directions.
(Dorian 1981: 85)
topic and purpose in her narrations and also with the increasing ease of the
interview situation over two tape recording sessions. She is a testimony to the
versatility of speaker and language both, late in the life of each.
I am grateful to Edward Finegan for encouraging me, both in general and with specific
prompts and questions, to persist in working through my imperfect corpus to locate
evidence of stylistic variation, and to Edward Finegan and Douglas Biber for help-
ful comments on the original draft of this paper and for suggestions on the relevant
literature.
References
Briggs, Charles L. 1986. Learning how to ask: A sociolinguistic appraisal of the role of the
interview in social science research. Cambridge: Cambridge University Press.
Dorian, Nancy C. 1978. East Sutherland Gaelic: The dialect of the Brora, Golspie, and
Embo fishing communities. Dublin: Dublin Institute for Advanced Studies.
———. 1981. Language death: The life cycle of a Scottish Gaelic dialect. Philadelphia:
University of Pennsylvania Press.
Dressler, Wolfgang U. 1982. “Acceleration, retardation, and reversal in language decay?”
In Robert L. Cooper, ed., Language spread. Bloomington: Indiana University Press.
321–36.
———. 1988. “Language death.” In Frederick J. Newmeyer, ed., Linguistics: The
Cambridge survey. Vol. IV. Language: The socio-cultural context. Cambridge:
Cambridge University Press. 184–92.
Dressler, Wolfgang U., and Ruth Wodak-Leodolter. 1977. “Language preservation and
language death in Brittany.” International Journal of the Sociology of Language 12:
33–44.
Hymes, Dell. 1984. “Sociolinguistics: Stability and consolidation.” International Journal
of the Sociology of Language 45: 39–45.
Labov, William. 1970. The study of nonstandard English. Champaign, IL: National
Council of Teachers of English.
Obler, Loraine K., and Lise Menn, eds. 1982. Exceptional language and linguistics. New
York: Academic Press.
Tannen, Deborah. 1987. “Repetition in conversation: Toward a poetics of talk.” Language
63: 574–605.
In the latter half of the twentieth century the sole remaining bilingual pop-
ulation in East Sutherland, on the far northeast coast of mainland Scotland,
consisted of former fisherfolk and their descendants in the villages of Brora,
Golspie, and Embo. Gaelic-English bilingualism was nearing its end in this
region, since no young people were acquiring Gaelic, and productive skills in
Gaelic were restricted to the upper portion of the age spectrum. In spite of the
recessive state of Gaelic in this locality, older speakers fluent in Gaelic con-
tinued to display well developed individual style differences in speech events
that they engaged in with some frequency, such as joking, teasing, and telling a
story. I propose to consider here one aspect of narrative style displayed by some
high proficiency speakers of East Sutherland Gaelic, namely the use of direct
quotation in story telling, and in particular code choice in direct quotation.
In the speech communities to be looked at here, the first of these two code
choices is influenced above all by interlocutor, provided that all parties to the
conversation are bilingual (Dorian, 1981: 76–77). Each bilingual has certain
conversation partners with whom she or he normally speaks the ancestral
language (the East Sutherland variety of Scottish Gaelic, in this case) and others
with whom English is the norm. For fully fluent bilinguals, Gaelic is typically
used with many or most contemporaries (some once fluent and mostly
younger bilinguals, both male and female, decline to use the Gaelic they were
raised with) and with all or nearly all bilinguals older than themselves. With
speakers younger than themselves, fully fluent bilinguals establish Gaelic or
English as the habitual language of conversation on an individual basis. Gaelic
is more likely to be chosen as the normal conversational medium with kinfolk
not too much younger in age. Gaelic may also be the preferred conversational
medium with other younger speakers frequently met with, for example younger
members of families who live very near by or younger work-mates. Among less
than fully proficient younger bilinguals (semi-speakers), Gaelic is used almost
exclusively with certain kinfolk, typically a select few from the first- or second-
ascending generation.
Politeness norms add certain constraints, in terms of fluent bilinguals’ code
choices. If a monolingual English speaker is present (other than a young child
1 Many such “interviews” were more like story-telling sessions than the rubric suggests, and
one or two were more like conversations among the fluent speakers present. In all but one
case I was very well known on a long-term basis to all of the other speakers present (and in
that one case to two of three others), so that the stiffness that might be expected in a single-
encounter interview, the kind conducted by a stranger among strangers, did not appear. I
retain the label “interview,” all the same, in recognition of the fact that tape-recording
constrains at least potentially all speech events on which it intrudes, and of the fact that I
retained the privilege of interposing questions during the course of the recording session and
did so with greater or lesser frequency according to the flow of the session or to my need for
background information.
2 The census is taken by parish, and neither Golspie village nor Embo village is coterminous
with a parish. The whole village population figures given here for Golspie and Embo are esti-
mates given by courtesy of the Scottish Registry.
3 Speakers are coded by village and age. E stands for Embo. G for Golspie, and within each vil-
lage, 1 represents the oldest speaker among those with whom I worked.
the laird was a well-to-do incomer and neither he nor any of his family spoke
Gaelic.
(E4, 1970)4
Appropriate as a code switch of this kind may seem to the change of interlocu-
lors here, realistic reflection of the original is not a compelling consideration
for most East Sutherland narrators when they include direct quotations in their
Gaelic stories. The unusual change of codes in this story, rendering the conver-
sational interchange between the narrator and his second (English-speaking)
interlocutor, was surely influenced by just who that interlocutor was. In order
to hold this particular conversation the narrator had to go up to what is still
4 See the list of abbreviations that follows the body of the paper for grammatical identifications.
known in the Highland countryside as an Taigh Mòr, “the Big House,” where
the laird and his family resided, and he went essentially as a suppliant, since
it was the laird who made available the allotments that the fisherfolk used for
their potatoes and oats. Furthermore, Embo householders owned their houses
but not the land the houses stood on; for the latter they owed feu duty to the
laird, which they had to trek to the Big House to pay. In this region, subject
for centuries to anglicized and anglicizing nobility and gentry, there were no
Gaelic speakers to be found at the Big House, and both the place and its occu-
pants were thoroughly identified with English in the minds of local bilinguals.
Certainly this Embo narrator did not routinely quote all conversations with
English monolinguals in English. In another narrative recorded in the same
year he quotes the skipper of a Lowland herring fishing boat on which he had
hired out as if the skipper had spoken in Gaelic, for example as the Lowlander
assuredly had not.
Because conversational speech to and from monolinguals is in the usual
case freely rendered as if it had been spoken in Gaelic, during the presentation
of Gaelic narratives, it is often impossible to tell on the basis of anything in
the Gaelic of the narrative itself whether a speaker being quoted is or is not
actually a speaker of Gaelic. In favorable cases there may be a clue in the
storyline or in the setting, even though the quotation itself gives no clue, but
in others there are no grounds for a firm conclusion. The same elderly Embo
bilingual quoted above tells in another reminiscence of someone coming to
the door of his home with a telegram:
(E4, 1970)
(3) hurd mi hĩãn ri pε̃n m____ , te: [ən] danəs a ǰe: šə̃ n
said I self to Mrs [name] what the.devil is that.one that
‘I said myself to Mrs. M__, “What the devil is that one
(G2, 1968)
The quotation’s form indicates that the other woman in fact produced the
phrase “I don’t know” in Gaelic, since the final two words of the Gaelic phrase
/an al fis akəm/ are rendered in nonlocal dialect forms (though not the first
two, which appear in strictly local forms).5
In several other cases where the narrator makes a still more explicit statement
about the language used in the original, identifying it as English, the statement
constrains the language selected for the quotation in the bilingual’s Gaelic
narrative only briefly, or not at all. The Golspie speaker quoted in (3) above
told with some indignation on another occasion of meeting two fisherfolk
descendants in a shop, neither of whom used Gaelic, and of challenging them
about their choice of language:
(G2, 1964)
5 Macaulay (1991: 183–191) distinguishes between mimics and translators, among his Scots-
speaking narrators who use direct quotation. The latter reproduce in their own Scots speech
remarks originally produced in other forms of English, where that suits the tone or purpose
of the narrative, whereas the former enjoy mimicking other forms of English when quoting
non-Scots speakers. In these terms the East Sutherland Gaelic speakers who served as my
sources are nearly all translators, when telling a story in Gaelic, and even the efforts of E4
and G2 to reproduce the language of the original are not very successful. The laird’s daughter
is quoted in English, to be sure, but her English shows the same strong Gaelic accent with
which E4 spoke that language, and Mrs. M.’s Gaelic remark in G2’s rendering appears with
only two words out of four in nonlocal form.
a h igəs . . . va a k eax
that is at.him . . . was he at shouting
that he has, he was shouting
va n´ ĩrĩn´ adəs
was the daughter at.you-EMPH
“Your daughter was
(E33, 1967)
The conversation between the farmer and the speaker’s father continues, in
her narrative, rendered entirely in Gaelic.6
There are a few cases in which the narrator’s conversational partner is
quoted so consistently and extensively in Gaelic that I actually began to
wonder whether, without my having realized it, an interlocutor known to me
purely as an English speaker was just possibly a Gaelic speaker after all, perhaps
originally an incomer from a more westerly area where Gaelic was more widely
spoken.7 In one such case four direct-quotation conversational interchanges
between the bilingual narrator and a woman whom I had always believed to
be an English monolingual had already occurred, with both the narrator and
her conversation partner represented as speaking in Gaelic, before a direct
quotation in English suddenly appeared as a conversational response from
the putative monolingual to a direct quotation in Gaelic from the bilingual
narrator:
6 The short initial shout of outrage from the English-speaking farmer, with its doubling for
narrative effect, may appear here in English because it had become a stereotyped piece of
mockery used by the Embo children to plague this particular farmer. There was at any rate a
certain repertoire of verbal mockery, regularly shouted out most unkindly by the Embo chil-
dren after certain individuals, for example, a mocking rhyme called after the man who came
to the village selling meat from his cart and a frequently made mistake in her Gaelic called
after a female incomer who was trying to acquire Gaelic.
7 Because the number of local Gaelic speakers was so small, and also because local Gaelic
speakers are referred to by means of a distinctive set of by-names, it was possible to rule
out the woman in question as a previously overlooked member of the local Gaelic speech
community. In this case I also checked subsequently with others and made certain that the
conversational partner here was, as I had supposed, an English monolingual.
xanu ɔ____.
say-COND [name].’
O____ would say.’
(G2, 1970)
The reason for the sudden realism of this particular quotation (followed by two
additional English sentences from O. in continuation of the same exchange),
within a narrative in which conversation to and from O. appears in Gaelic both
before and after this interchange, is straightforward enough in the local con-
text. There is no colloquial equivalent locally for the English word damn, and
female bilinguals are in any case more inclined to reproduce English profan-
ity in English than they are to reproduce any other quoted English material
in the original. An Embo woman, for example, telling in Gaelic of a drunken
supervisor who included a swearword in ordering his workers to leave shelter
to perform an outdoor job in vile weather, similarly quoted his profane remark
in English (E33, 1964). The same narrator who quotes O.’s profanity in English
here was variable in quoting her own profanities, giving them twice in Gaelic
(once realistically, since her interlocutor was also a Gaelic speaker [see (3)
above]; but once unrealistically, with the interlocutor her own monolingual
son). On a third occasion she began realistically in English, representing a dia-
log with an English monolingual, and then switched in mid-quote to Gaelic, in
keeping with the language of the rest of this long story:
ən d uər u bokh: am
INTERROG PRET got you the.bag at.me
did you get my bag?” ’
(G2, 1970)
Just as profanity can trigger a switch to a more realistic code choice in rendering
a conversational original, so can other remarks the negative or positive force
of which the narrator especially wishes to convey. This same Golspie Gaelic
speaker quoted in English the cutting English remark of a Gaelic teacher who
rejected her local Golspie version of a Gaelic phrase he wanted translated
(“That’s not in the language!”; G2,1968), and a nonagenarian Embo speaker
reproduced in English the callous remark (from the point of view of one who
had previously dealt only with doctors who made house calls as a matter of
course) of a new doctor who asked instead that she find someone to bring her
to his office (“I have a very busy surgery”; E17, 1996). In the same fashion the lat-
ter speaker quoted in English, with obvious disgust, the memorably unpleas-
ant abuse hurled at an unfortunate cat by its monolingual owner (E17, 1994).
In a happier connection this same elderly Embo speaker switched into
English to reproduce a remark that had resonated exceptionally positively
with her. She was deeply gratified to find that a journey she had made at some
physical cost to herself to visit a younger but terminally ill relative in the
Lowlands was as cheering to the ill woman as the speaker had hoped it would
be. She quoted from a phone conversation between herself and the relative
after the visit, rendering their conversation as if it had been in Gaelic until she
came to the remark that had been so particularly gratifying to her, at which
point she switched to the English in which the conversation had actually taken
place:
(E17, 1994)
iǰər anə šɔ
at all in here
here at all.” ’
(G2, 1970)
the sometimes faulty English typical at one time of East Sutherland bilinguals,
whose stronger language was Gaelic:
tε xãnu i riš
what say-COND she to.it
what she would say for it [i.e., for /khiščh ǝ ɫɔi/ in English].’
(E27, 1967)
(E29, 1974)
The point of this story is that the narrator feels compelled to use a language
that no one else in his office knows when he speaks on the phone to someone
who is a habitual Gaelic conversational partner of his. But when he quotes
the remark made by his fellow-workers about the unintelligibility of his phone
conversation, he nonetheless quotes them as if they had made the remark in
Gaelic.
This same speaker makes a similarly odd-seeming code choice in quoting
his own monolingual daughter’s frequent requests, at high-school age, that he
teach her Gaelic:
(E29, 1974)
Abbreviations
Acknowledgements
I’m indebted to Susanne Romaine and Robin Sabino for helpful criticisms of
the original draft of this paper. Responsibility for remaining shortcomings is
my own.
References
Introduction
and that seemed to suggest, when reporting on other figures in the novel. Some
of the same implications could be introduced in that way, but such turns of
phrase would quickly become repetitious in a way that ongoing use of a single
verbal tense-and-mood choice does not. A different tone would result, and a
stylistic economy that contributes to the novel’s power in German would cer-
tainly be lost.
Who would read Breitbach in English (or Dostoyevsky, or Ibsen, or Tagore in
English) if s/he had the requisite language skills to read the work in the origi-
nal language? Languages have their individual voices, created equally by the
means which they deploy and by the details of the deployment. As George
Steiner says: “Each human language maps the world differently. . . . Each
tongue . . . construes a set of possible worlds and geographies of remembrance”
(1992: xiv). Semantic mappings across the lexicons of any two languages are
often so obviously different that the rankest beginner confronts them imme-
diately and sharply; more subtle differences continue to emerge and confront
the second-language learner for years, as familiarity increases.
Linguists have the preoccupations of their special field, naturally enough.
The properties of a language that fascinate and please them are those that
are unusual from a specialist’s point of view: a prominent role for one of the
grammatical devices less widely encountered among the world’s languages, or
a region’s (infixes, say); the appearance of a syntactic property that had been
thought not to exist (languages with both classifiers and gender as separate
categories; see Aikhenvald 2003); the occurrence of one phenomenon without
another phenomenon believed to be routinely co-occurrent with it (massive
grammatical restructuring despite near-absence of lexical borrowing from the
language that provides the model for the restructuring; see Aikhenvald 1996).
All of these things are of understandably high interest to anyone who studies
linguistic properties, but dear as they are to the linguist they are no more likely
than much commoner features to form the basis of the very particular effects
that native speakers are able to achieve with their languages. A grammatical
feature unfamiliar to speakers of English, Spanish, or other Western European
languages, such as the partial-reduplication prefixation that expresses inten-
sification in Turkish (beyaz ‘white’, bembeyaz ‘extremely white’; yalnĭz ‘alone’,
yapyalnĭz ‘absolutely alone’; etc.), attracts the instant attention of an English
speaker who learns Turkish as a foreign language. But the same English speaker
who’s struck by intensification prefixation in Turkish may never have noticed in
his or her own speech the subjunctive marked by absence of the present-tense
indicative 3rd-person suffix -s in the finite verb of the subordinate clause in
sentences such as I suggest she try again, I’d prefer that he not go. Sophisticated
native English speakers, much given to lamenting the disappearance of the
I would like here to look at East Sutherland Gaelic (ESG) from the point of view
of expressive uniqueness, highlighting an expressive resource that ESG speak-
ers, all Gaelic-English bilinguals, have available to them in their Gaelic but not
in their English.
The Gaelic dialect at issue here is a variety spoken in the second half of the
twentieth century by a dwindling population of fisherfolk and their descen-
dants in three villages of coastal East Sutherland in Highland Scotland. ESG
to suit ESG, rather than as in standard written Scottish Gaelic): phòs a /fo:s a/
‘he married’, vs. pòs a /pho:s a/ ‘marry him!’; tha cat agham /ha khath am/ ‘I have
a cat’, vs. tha cait agham /ha khačh am/ ‘I have cats’. Some of the mutations are
obligatory and yet do not carry grammatical information, which doesn’t trou-
ble native speakers in the slightest but can seem an unnecessary and unkind
complication to learners. Despite the fact that consonant mutations are rela-
tively unusual as a high-frequency grammatical device, there is relatively little
expressed via consonant mutation in ESG that is not either fully matched as
a grammatical category by some grammatical element in English, as is true
of the past tense and the plural, the two grammatical categories expressed by
consonant mutations in the examples given above. (The major exceptions are
direct address in the form of the vocative case, which is marked only supra-
segmentally – e.g. by pitch- and stress-contours and by timing – in English but
by consonant mutation as well as by suprasegmentals in ESG; and grammati-
cal gender, which is marked by consonant mutation for one class of nouns,
provided the definite article is present, but can be marked by other devices in
various other grammatical environments).
By comparison with the attention lavished on the consonant mutations of
Scottish Gaelic in most grammars, the emphatic suffix, the chief feature to be
discussed here, is only briefly mentioned in most treatments of Gaelic dialects.
One reason for the disparity in treatment is natural enough: the consonant
mutations affect a large number of different consonants and appear obligato-
rily in many different environments (and optionally in still others), whereas
the emphatic suffix takes a limited number of forms and can be suffixed to
only a limited number of elements, while its use is largely optional. Still, there
is most likely another reason as well. The emphatic suffix serves above all to
create discourse effects, rather than to express grammatical categories, and
both traditional grammars and linguistic descriptions show a tendency to con-
centrate on grammatical elements whose domain is the sentence. Some of the
expressive force of the emphatic suffix can be seen within the sentence, or
across one or two sentences, but to see its most striking effects it’s necessary to
look at longer stretches of discourse.
Scottish Gaelic grammar also recognizes a set of three unstressed and post-
posed enclitic particles with deictic force, expressing roughly ‘this’, ‘that’, and
‘yon’. The first two of these are recognizably present in ESG, but only one of
them is productive, a proximal deictic which takes exactly the same phono-
logical form as the dominant allomorph of the emphatic suffix, namely -(e)as
/-əs/ ‘this’. The deictic force of proximal -(e)as is clearly recognizable when it’s
applied to nouns with temporal reference, e.g. an t-seachdan-as /ən ǰaxkanəs/
‘this week’, but otherwise the deictic force is less obvious, often undiscernible.
The phonological merger of all these elements – emphatic suffix, emphasiz-
ing particle, and deictic enclitics – blurs their distinctiveness in ESG, and it
appears that the -(e)as suffix can now combine their semantic force to some
extent.
In view of the ESG merger of the emphasizing elements (emphatic suffix
and emphasizing particle) with the deictic enclitic, it would be more accurate
to speak of an emphatic-deictic suffix for the dialect. But since the emphatic
function is considerably the more prominent in contemporary usage, the
rubric ‘emphatic suffix’ can be used for economy’s sake to cover the single pho-
nological outcome.
The emphatic suffix appears at highest frequency with the personal pro-
nouns and with the prepositions that conjugate for person. Among the latter
the pronominal forms of the preposition aig ‘at’ are most frequent, since conju-
gated forms of aig, in conjunction with the verb ‘to be’, serve in the absence of a
verb ‘to have’ to express possession (‘I have a brother and a sister’ is expressed
as ‘a brother and a sister are at me’). As an example of emphatic suffixation, the
ESG forms of the personal pronouns and of the prepositional pronoun aig are
given here with their emphatic equivalents (as they might realistically be writ-
ten if this were a written speech variety, and with the forms from the village of
Embo cited wherever there are inter-village differences):
* Although the form normally used in ESG is am, speakers appear to be aware of a fuller
underlying form and occasionally someone produces a form such as agam.
** In the case of the two pronouns that end with voiceless final consonant, aic and ac, the
emphatic form shows variation, appearing sometimes just as suffixed -s and sometimes as -(e)as.
Apart from its appearance with the personal pronouns and the conjugated
prepositional pronouns, the ESG emphatic-deictic suffix can attach to certain
other pronominal forms (e.g. an té ‘the one’, used in standard Gaelic for femi-
nine nouns only and likewise in ESG where humans are in question, but used
in ESG for both female and male animals and for all inanimate nouns that have
purely grammatical gender) and to a noun or an adjective when either is the
final element in a noun phrase with a possessive pronoun: mo phiùthaireas
/mə fyu:rəs/ ‘MY sister’, an taigh móras /ən dε mo:rəs/ ‘their BIG house’. Note
that by contrast to English, where the voice emphasis falls on the possessive
pronoun in the case of ‘MY sister’, it’s the noun that takes the emphatic suffix
in Gaelic; the possessive pronoun, mo in this case, is always unstressed and
cannot combine with other elements. In ESG just one verb form can add the
emphatic suffix, namely the first person singular of the conditional: rachain-
neas air ais /raxĩn´əs er aš/ ‘I WOULD GO back!’ There are occasional occur-
rences with numerals: bheir mi na dhà-as seachad /ver mi nə γa:əs šaxəd/ ‘I’ll
pass THESE TWO on’.
Of particular interest here are uses of the emphatic suffix over more extensive
stretches of speech to create discourse effects. I offer below some examples
from recordings of ESG interviews, narratives, and taped “letters” to show
typical discourse effects. They begin with an instance that could be paralleled
quite effectively in English by use of supra-segmentals alone, since emphasiz-
ing the equivalent words by stress and pitch in the English translation creates a
similar effect. In the later examples, however, the number of emphatic suffixes
used goes beyond what could appropriately be matched by voice emphasis
in English; the number of sentence elements that can take voice emphasis in
English without semantic or affective distortion, over an extended stretch, is
limited by comparison with the number of sentence elements with emphatic
suffix attached that can comfortably appear in an extended stretch of ESG.
Abbreviations in Transcriptions:
* The conditional suffix and the 3 singular imperative (subjunctive) suffix are both /-u/. But in
the independent conditional (that is, in the conditionals not governed by certain particles and
conjunctions), the suffix /-u/ combines with obligatory verb-initial consonant mutation,
whereas no initial consonant mutation appears in the subjunctive.
Lines in Transcriptions
1. orthographic rendering of East Sutherland Gaelic; words with emphatic
suffixes are capitalized and boldfaced
2. phonemic rendering
3. morpheme-by-morpheme gloss; a period within the gloss signifies a com-
plex morpheme the constituents of which can not be designated by seg-
menting the surface structure
4. English translation; words that correspond to Gaelic words with emphatic
suffixes are capitalized and boldfaced.
3. you / your party vs. me / my party: ‘I’ve got a bone to pick with you!’
------------------------------------------------ --------------
(a) a serious matter
------------------------------------------------ --------------
(Source: Embo female, aged 42 at the time of the recording in 1967.)
(b) joking
------------------------------------------------ --------------
(Source: Golspie female, aged 69 at the time of the recording in 1964.)
Note that in the case of the oppositional emphatics of 3(a), lines 2 and 3, and
3(b), line 3, both parties are represented by pronouns that show the emphatic
suffix, whereas in English one pronoun in each line would certainly receive
voice emphasis but probably not the other. This sort of symmetrical emphasiz-
ing of the pronoun forms representing both speaker and hearer is frequent in
the use of ESG emphatic suffixes, highlighting the interactional dimension of
the material.
To see the emphatic suffix come into its own most fully, in ESG, the best
place to look is a narrative told with plenty of feeling. In the tape-recording
from which example 4 is taken, the narrator tells of a rivalry between choirs
from the villages of Golspie and Brora. A Gaelic choir from each village will
be competing for a cup at the provincial Gaelic music festival, and the singers
(none of whom speaks the standard language in which competitive singing is
done, and many of whom don’t speak or understand Gaelic at all) will need
the services of a Gaelic coach to help them prepare. The narrator of the story,
a bilingual Golspie woman, doesn’t take kindly to the woman who has come to
coach the East Sutherland choirs, and the force of her opposition to the woman
is felt not just in the words she uses but also in the number of emphatic-deictic
elements that appear in the narrative.
Quite generally speaking, the narrator from whose recorded story example 4 is
drawn is a highly expressive speaker. In most of her interviews and stories, not
just this one, she makes use of a particularly rich array of interjections, and she
doesn’t shy away from using mild profanities and other indelicate lexical items,
even when she’s being tape-recorded. The pitch- and stress-contours in her sto-
ries tend to be greater than average, and she’s inclined to hilarity when there’s
the least shade of impropriety or absurdity in whatever matter she relates. In
English as well as in Gaelic her stories are lively, then, but in her Gaelic arse-
nal she has some weapons not available to her in English. One of them is the
emphatic suffix, supplemented on occasion by emphatic use of fhéin ‘self’.
There are 27 clauses in the narrative stretch offered as example 4 above.
Twelve of them, or almost half, include an emphatic suffix; in two instances
there are two emphatic suffixes in a single clause. (There are also two instances
of emphatic use of fhéin.) The tone is set immediately, when the narrator
speaks of seeing the Gaelic coach come into the hall and refers to her as an
té-eas ‘that one(-female)’. Two sentences earlier the narrator had referred to
the same woman as am boireannach seo ‘this woman’ and had then also used
two unemphatic pronominal forms in referring to her; but as she begins the
because the emphatic suffix in Gaelic doesn’t require any particular pitch or
stress prominence, it can be employed multiply within a single clause and
repeated frequently over a series of clauses without making the narrative
sound overwrought and without distracting attention from the unfolding of
the story line. Gaelic makes available a separation of voice emphasis and focus
that English, with only supra-segmental features to indicate emphasis, can
not provide. The speaker in example 4 uses emphatic forms of the personal
pronoun four times in leading up to direct quotations, as she reconstructs a
conversation within her narrative. By keeping the emphatic suffixes coming,
she keeps the discourse tone (continuing intense interest in the objection-
able Gaelic coach) constant over a long stretch. None of these personal pro-
nouns has voice stress in the Gaelic, and it’s precisely the absence of stress
that makes it possible for a series of emphatic personal pronouns to play their
role in maintaining discourse tone without diminishing the salience of the
quotations that they precede. In addition, symmetrically placed emphatic suf-
fixes can highlight speaker-addressee or subject-object oppositions in ESG (see
especially example 3(b) above) in a way that multiple contrastive occurrences
of voice emphasis in English can not, at least without distortion.
For the written language there is of course also the advantage that the dis-
course tone of the spoken Gaelic text persists, thanks to the visible presence
of the emphatic suffixes, while the discourse tone created by voice emphasis
is lost in formal written English. (It can be evoked in casual written English by
means of underlinings and exclamation points, liberally resorted to by some
people in their private correspondence by way of a substitute for the missing
suprasegmentals.)
As is evident in example (4), where the use of fhéin ‘self’ is seen to supplement
the discourse effect of the emphatic-deictic suffix, languages not only offer
distinctive resources but offer the possibility of combining them in distinc-
tive fashion. Gaelic speakers (and Irish speakers, too), besides combining the
emphatic use of fhéin with use of the emphatic-deictic suffix -(e)as, are known
for their frequent use of “clefting” to allocate emphasis: a “dummy subject”,
it, leads off the sentence, linked by a form of be to material highlighted by its
postponement. So prevalent is clefting in both Scottish and Irish Gaelic that
in its frequent carry-over into Highland and Irish English it’s become a stereo-
typed feature (e.g. “It’s grand stories he’s telling!”). And once again, speakers
can combine this device with the emphatic suffix or fhéin to produce a particu-
larly strong effect. The following two examples drawn from narratives recorded
from an octogenarian Embo man in 1970, combine, respectively, clefting with
emphatic suffixes and clefting with fhéin:
a’s a’ bhaileas.
as ə val-əs
in the village-emph
in this village.
In Irish, clefting and the emphatic suffixes are used to the exclusion of supra-
segmentals to mark focus and emphasis, according to Cotter’s analysis of the
Irish of radio broadcasting:
The Irish language does not use pitch prominence in the intonation con-
tour in the way that English speakers do, but uses instead syntactic reor-
dering through what could broadly be called clefting, and the so-called
‘emphatic suffixes’ (Cotter 1996: 48).
draw on them to powerful effect as well. Here are the opening lines of Gaoir na
h-Eorpa (‘The Cry of Europe’) by the late Sorley Maclean, considered by many
the finest of modern Scottish Gaelic poets (from MacAulay 1976: 74–75):
In this first verse of a poem evoking what was for Maclean the dark Europe of
the 1930s after Franco’s victory in the Spanish civil war, three emphasizing par-
ticles appear in the Gaelic (bheòil-sa ‘your mouth’, latha-ne ‘our day’, and phòig-
sa ‘your kiss’), setting the lovely girl of spirit and beauty against the darkness
of contemporary Europe. The emphasizing particles are an important element
in establishing the contrast in Scottish Gaelic, but though the translation is
Maclean’s own, nothing is available to him in written English to create a similar
effect. If the English version of the poem were to be read aloud, furthermore,
and voice stress were applied to your and our, the effect would be distorting,
not enhancing.
Among the ESG examples offered above, the expressive potential of the
Gaelic emphatic suffix is especially evident in the fourth. The deployment of
the emphatic suffix seems a pretty straightforward matter in example 1, and
perhaps also in examples 3(a) and 3(b). It’s less obvious (at least to me) why
some pronominal forms appear with the emphatic suffix in examples 2 and 4
while others don’t. (Why not gheibheadh aid Gàidhlig bhoidheas ‘they would
get Gaelic from him’, for example? And why does the high-frequency expres-
sion ’s am bidh ‘whatever, at all’ [literally ‘in the world’], never pick up an extra
degree of intensification and become ’s am bidh-eas?) Already in example 2,
but even more so in example 4, we reach the realm of skilled-native-speaker
stylistic choices. It’s beyond me both as linguist and as learner to account fully
for the motivating factors behind the native-speaker choices, in texts such as
these, where stylistic choices were obviously made not to use the emphatic
suffix in some potential environments, as well as to use it in others.
At the same time, it’s well within my capacity both as linguist and as learner
to recognize and relish certain stylistic effects from the speakers’ deployments
of the suffixes, and to celebrate the liveliness and coherence of tone that the
emphatic-deictic suffix brings to their Gaelic discourse. The use made of the
emphatic-deictic suffix by the speaker in example 4 is not subtle, and it’s easy
to imagine sophisticated story tellers who might introduce suffix-bearing forms
less frequently but more slyly and strategically, to more cunningly designed
effect. But the particular way the speaker in example 4 uses the emphatic suffix
is well suited to her personality and speaking style. The grammatical environ-
ment surrounding the central figure, the an té-eas ‘that woman’ of the narra-
tive who then also appears as the direct focus of nine emphatic suffixes, creates
a consistent discourse environment in which that high-focus figure becomes
an unsurprising target for ‘a kick in the rear’ or an invocation of the devil. The
Gaelic passage has a well-sustained narrative tone in which the ESG emphatic-
deictic suffix plays a substantial and – for all my morpheme segmentations
and glosses – never fully translatable part. Sorley Maclean, practiced transla-
tor of his own poetry that he was, fared no better when he came to render the
Gaelic of Gaoir na h-Eorpa into English; he had to forego in English the special
contrastive effect that the emphasizing particles had given his opening verse in
Gaelic. The very distinctiveness of a language’s most particular features creates
the insoluble problem any would-be translator (or celebrator) faces, of course:
belonging uniquely to that language, they are essentially untranslatable.
References
the proficiency continuum by age alone. Neither did the onset of progressively
greater deviance appear at exactly the same individual along the continuum
for each feature which was tested, yet over the rather large number and type of
features which were investigated a quite stable ordering of the twelve imper-
fect speakers’ relative conservatism proved attainable. The East Sutherland
Gaelic semi-speakers with whom I worked also did not themselves fall neatly
into line via correlation between age and proficiency, but on the other hand
there was perfect age/proficiency grading across the three groups of differen-
tially skilled speakers: older fluent speakers represented a relatively conser-
vative norm for the dialect; younger fluent speakers (who appeared only in
the smallest, most cohesive, and least anglicized village where the dialect was
dying most slowly) were all younger than the older fluent speakers and showed
identifiable deviations from that norm which were not however noticed by
the speech community, of which they were fully fluent, skillful members;
semi-speakers varied considerably among themselves as to their skills, and
greater proficiency did not always correlate with greater age, but within each
village where they were studied they were all invariably younger than older
fluent speakers and younger fluent speakers (Dorian 1981: 117). As in the case
of Schmidt’s study, the large number of features investigated made it possible
to group the semi-speakers into stronger and weaker proficiency categories of
quite good generality overall.
Lack of knowledge about any prior conservative norm for the speech form
has occasionally led researchers to compare the dying speech form with a
standard language, where one exists. This is usually both diachronically and
descriptively precarious and can result in findings which are unprovable at
best and seriously misleading at worst. Most often one simply does not know
what to make of such findings; certainly no very useful conclusions can be
drawn about degree and direction of change, since the differences might be
explainable in terms of very early dialect differentiation rather than subse-
quent deviation from the norm. If we had no independent knowledge about
the Alemannic dialect of Old High German, for example, and tried to explain
the forms and sounds of Swiss German dialects in terms of changes away from
standard German, our discussion would be impossibly flawed. By good fortune
we have early attestation of Alemannic Old High German, and we also know a
good deal about, the development of a standard German as a rather late com-
promise form based on East Middle German, with its roots in the Prague and
Wittenberg chanceries, serving subsequently as the dominant element in the
language of Martin Luther’s influential Bible translation (Waterman 1966: 117).
Because of lack of knowledge about earlier local norms, Trudgill (1976–77)
was in this sort of unfavorable position where Arvanitika, a form of Albanian
certain negative attitudes toward the local language are equally shared
by fluent and terminal speakers. Surfacing of these attitudes is almost
predictable from knowledge of the discourse context such as naturally
emerging language focused conversations. The whole set of linguistic
attitudes could be summarized as follows: (1) Arvanitika is a bastard lan-
guage not worth saving; (2) localities other than the speakers’ own speech
community always use a deeper or less hellenized variety regardless of
actual geographical distances or of objective dialect differences; and
(3) children should not be instructed in Arvanitika since its learning
interferes with a proper acquisition of competence in Greek (1983: 293).
minority language is the normal daily speech of the parental and grand paren-
tal generations. It is noted for example by Huffines (1980: 52) for Pennsylvania
German, by Pulte (1973: 426) for Cherokee, by Denison (1971: 166–67) for
German and Friulian in Italy, by Timm (1980: 30) for Breton, and by Dorian
(1981: 104) for Scottish Gaelic .
The majority population in which the threatened speech form is enclaved
may have negative attitudes towards the minority language, shared by some
but not all members of the threatened speech community. This was true for
the English- and Dyirbal-speaking area studied by Schmidt (1983: 24–26) and
for the German- and Hungarian-speaking area studied by Gal in Austria (1979:
106–07). In such areas the range of attitudes and the ambivalences found
within each segment of the population can make particularly slippery foot-
ing for the investigator. Generally speaking it is certainly true that negative
attitudes, whether as wide-spread throughout the bilingual population as was
the case with the Arvanitika speakers studied by Tsitsipis or as variably present
among both monolingual English speakers and bilingual English-and-Dyirbal
speakers as was the case in the population studied by Schmidt, complicate the
investigator’s task. In my own work in East Sutherland I met with obstacles
ranging from fluent native speakers’ initial recommendations that I should go
elsewhere so as to study the “right” Gaelic and attempts on the part of dia-
lect speakers to modify their local Gaelic in the direction of biblical Gaelic or
whatever they knew of western dialects, to incredulous questions from English
monolinguals about the possibility of doing any serious work in the most iso-
lated and in-bred of the Gaelic-speaking villages, where the natives had (in
some circles) the reputation of being backward and slow-witted.
Rivalries among remaining speakers, on the other hand, were a very minor
problem in East Sutherland. A few people who disliked each other cast asper-
sions on each other’s language loyalty, suggesting that the other individual was
“too proud” to speak the native language, but false allegations of incompe-
tence were non-existent. If anything, a reverse problem occasionally cropped
up: people were credited with rather more ability than they actually had. The
reasons for overestimation seemed to be two: strong enthusiasm for Gaelic on
the part of an imperfect speaker could lead people in relatively little direct
contact with that individual to assume more ability than actually existed; and
near-perfect receptive control of the language in combination with excellent
knowledge of the sociolinguistic norms for the community could make quite
limited active use of the language, with heavy reliance on fixed phrases and
high-frequency collocations, sufficient for fairly unremarkable interaction
with more fluent speakers.
In some communities where a language is nearing extinction, familiarity
with the ancestral tongue may have special value for the few remaining speakers
since it qualifies them, and them alone, to perform certain special rites or ser-
vices. This in turn entitles them to particular respect as a link with a more
intact ethnic past. If, in addition to the high prestige already associated with
their rare language skills, linguists or anthropologists appear on the scene and
create a new and also prestige-enhancing audience for their abilities, the value
of their linguistic performances can rise sharply and kindle a certain amount
of competition. In the absence of a speech community large enough and vital
enough to permit the investigator either to become a skilled speaker him- or
herself or to obtain convincing community consensus on the relative abilities
of the remaining speakers, the investigator can find it all but impossible to
determine which of the few speakers available are the most reliable and most
skillful. The claims of friendship, plus lack of frequent direct interaction, may
make the testimony of others among the surviving speakers less than perfectly
dependable when they assess the skills of the leading contenders for linguistic
laurels. Among the Delaware centered around Dewey, Oklahoma, for example,
with a total fairly dispersed speaker-population of only about eight individuals
in a community which valued traditional activities with a Delaware-language
component, the appearance of investigators from outside the community who
also placed a very high premium on language skills complicated the situa-
tion and gave rise to evaluation difficulties of this kind (S. Roark-Calnek, pers.
comm.).
Because of the complex social conditions typical of communities in which
languages or dialects are dying, the possibility that potential informants will
understate their own abilities is probably as great as the possibility that they
will overstate them. It is true that some individuals apparently relish the role
of sage and language expert for the prestige it confers and adopt it without the
linguistic competence to support it. Swadesh gives a particularly vivid exam-
ple for Chitimacha, where he describes the bizarre performance of an elderly
woman who produced without a minute’s hesitation “a most remarkable mix-
ture of Indian words and invented vocables, for the most part based on French
or English with some twist or change” when asked for Indian vocabulary; he
felt that she “must have been at least partly conscious of inventing forms in
order to maintain her role as one who knew much ancient lore” (1948: 231).
On the other hand some individuals, coming to adulthood in a region where
the ancestral language is negatively viewed and well aware of the low prestige
attached to native-speaker status, may choose to disclaim knowledge of the
language. By reason of strong friendships within a particular kin network I was
able to test quite extensively one individual who disclaimed speaker status and
to establish that her ability to generate sentences in East Sutherland Gaelic
was actually slightly superior to that of a near relative almost exactly her age
who eagerly claimed speaker status and used Gaelic at every opportunity. The
disclaimer never voluntarily used Gaelic at all, with the exception of rare occa-
sions when she wanted to keep something secret from her monolingual son,
and she had not identified herself as a Gaelic speaker on the national census
form (Dorian 1981: 109 and MS.). Altogether untestable of course are the asser-
tions of individuals in terminal speech communities about the “inner speech”
of their daily lives. While Melchers cites the experience of the Faroese scholar
J. Jakobsen investigating the Norse-derived speech of the Shetlands late in the
19th century, she certainly received one highly contradictory piece of testi-
mony herself:
Jakobsen was told that the ‘terminal speaker’ of Norn was a native of Unst
who died about 1850, which he found credible, whereas reports of later
speakers should be regarded with scepticism. These speakers probably
used exceptionally many Norn words but did not actually speak the lan-
guage. However, during my own visit to Shetland in 1979, a native of Yell
told me that he ‘did all his thinking in Norwegian’! (Melchers 1981: 256).
Since the 100-word list was administered as the last item in the interview,
after the language-attitude questions, speakers were often particularly
concerned to give ‘correct’ . . . responses for the Spanish test items. Thus,
many of the ‘missed’ items would doubtless be hispanisms in everyday
speech (1981: 216).
Gal, given her interest in shift between Hungarian and German and between
levels of style within each code, used a variety of strategies to elicit speech
variety. Her own identity as a native speaker of standard Hungarian made use
of Hungarian to her by bilinguals a matter of courtesy, but she was able to
counter this effect by taking advantage or norms of local accommodation to
monolingual German speakers and reinterviewing many of the same people
with a local monolingual German student along (1979: 66). She was also able,
during the second half-year of her residence in Oberwart, to obtain through
the close involvement of her participant observation, permission to tape ongo-
ing daily activities such as card-playing, dinner conversation, and housework
within eight households – sometimes even in her own absence, with the tape-
recorder left running. Whatever the effect of the tape-recorder, the relative
freedom of this style of recording (as opposed to direct interviewing) did pro-
duce the hoped-for wider range of style variation (op. cit.: 67). Subsequent vis-
its to the same community resulted in still further removal of the constraints of
formal interviewing as her status as what might be called an “inside outsider”
developed; young people were even willing to stand in for her, in a sense, by
taking her tape-recorder along in visits to relatives and friends, and she was
increasingly able simply to be present during conversations which did not
focus on her or even always direetly include her (MS.: 13).
My own greatest assets in East Sutherland were firstly that I was not British,
so that in spite of having English as my mother tongue I could not be placed
anywhere within the British class system; and secondly that I had never stud-
ied Scottish Gaelic at all, but rather learned it in East Sutherland by linguistic
field techniques, so that my Gaelic was the same as that of my sources. I had
learned a bit of Inner Hebridean Gaelic before my arrival in East Sutherland,
but it was quickly and permanently displaced by the local variety. And since the
local Gaelic was extremely distinctive, truly a badge of local identity, I became
something of an honorary East Sutherlander simply by becoming a speaker.
Hill and Hill (1981), Gal (1979 and MS.), Schmidt (1983), Ralph Cooley with
the Delaware of Andarko, Oklahoma (Yoder and Cooley MS.), and I, among
others, all worked in settings where more than one speaker was often present,
yet where the degree of “naturalism” might still vary markedly and the suitabil-
ity of the multiple-source technique might also be more or less great according
to the particular purpose of the session(s).
Hill and Hill found interviews with groups of people highly desirable for
the parts of their interview which were aimed at eliciting casual speech, but
distracting in the lexicostatistical word-list section, where as they report:
Gal’s ability just to “hang around”, after her identity and welcome in the com-
munity were well established, was a crucial counterbalance to her native-
speaker use of standard Hungarian. Schmidt was able to overcome the skin
color barrier after about two months and gain acceptance within two groups of
female young people; but it proved impossible for her as a young white woman
to record male speakers of Young People’s Dyirbal in a natural context because
her very presence would prompt a switch to English. Among the young women
she found that the popularity and general use of cassette recorders made
recording much less problematic than she had anticipated, and she carried
her recorder in a shoulder bag on group activities such as camping and fish-
ing trips; during story-telling sessions around the camp-fire speakers were not
even necessarily aware of whether the recorder was on or off (1983: 185–86).
Like Hill and Hill, I found that some types of elicitation work were rendered
useless where one of a pair (or group) of source people was particularly eager
and was consequently unable to refrain from breaking in with a response to a
question directed to someone else. In one especially disappointing case I had
to omit a conservative older fluent speaker from a select sample in which I
would very much have liked to include him because of the impossibility of
arranging elicitation sessions without a quicker and more eager member of
his household present. At the same time, like Gal, I found that the many occa-
sions and settings in which I was simply present and able to listen, or some-
times even make notes, without affecting the Gaelic conversation going on
evolve a strategy by which to capture both the discussion and the translation;
but my chosen informant was in effect in charge, with my own role reduced to
that of recording engineer, whereas Cooley continued to co-preside over his
group sessions by starting with a single-speaker text and recruiting the group
of friendly relatives as discussants. He was justifiably pleased with his results,
and indeed his method might very profitably be adopted by other fieldwork-
ers, assuming that a similarly friendly and cooperative group of “discussants”
could be assembled, each of whom would be agreeable to minute dissection
of a text which he or she had provided in a prior one-to-one setting with the
investigator.
In direct elicitation, whether with single or multiple source-people, special
problems are likely to arise. The most obvious are anxiety on the part of the
source people and artificiality in the responses. Much of the investigator’s suc-
cess in overcoming these obstacles will depend on his or her personality and
manner, in the first case, and on his or her ability to gain enough exposure
to spontaneous speech so as to spot artificiality in elicitation and reject it, in
the second. Since it is difficult to speak for others in these respects, and the
problems are seldom fully and explicilty discussed in the professional litera-
ture, I will deal here with my own field experiences. They were rather different
among the East Sutherland Gaels and among the Berks County Pennsylvania
Dutch, which at least makes one direct comparison possible.
In my East Sutherland work I was fairly young when I began, still a graduate
student and a mere youngster to most of the rather elderly population with
which I was to work. For my own part I had grown up in a household with
resident grandparents and felt at ease in dealing with the elderly. I tended to
like them, and they me; the tradition of Highland hospitality helped, most defi-
nitely, but the acceptance went beyond it. My “target population” was strongly
stigmatized, but I had not been socialized into such attitudes, of course; my
own personal bias was an anti-aristocracy and anti-gentry one, which only
endeared me to the ordinary people, who by and large held the same opinions.
Serendipity, so rightly invoked in the title of Miller’s 1971 paper, played its
usual prominent role. My place in one or two segments of the community at
large was aided by two accidental factors. A local family (the parents incomers
but long-time residents), one of whose daughters was widely acquainted in
Gaelic-promoting circles because of her profession, had lost another daugh-
ter to disease only a few years before, and by their account I resembled the
deceased daughter. My friendship with this family became close and long-
standing, almost certainly aided by the accident of physical resemblance to
the much mourned daughter. (See Yoors [1967: 76–82] for an account of his
sudden total acceptance by a Romani family when his close companion, their
son, was killed in a traffic accident.) In another incomer family, west coast
Gaelic speakers even more widely acquainted in Gaelic circles, my acceptance
was probably hastened by the fact that I was very musical: they took responsi-
bility for training me, as they did so many others, to take an active role in the
Gaelic musical life of the district. Music was also an initial passport to accep-
tance and eventual sponsorship in another key family, one of whom was an
East Sutherland Gaelic speaker, though of crofting rather than fisherfolk back-
ground; he held a position of authority in the village richest in local Gaelic
speakers and paved the way for my work there through his personal influence.
He was himself a Gaelic singer, and music formed the basis of our friendship
initially.
In their book People Studying People, Georges and Jones (1980: 50–51) discuss
in somewhat unfavorable terms scholars who pass themselves off as “students”
instead of stating outright that they are fieldworkers, researchers, anthro-
pologists, linguists, folklorists, or what-have-you. I found that in stating that
I wanted to learn the local language and write a study of its sound system I
was automatically assigned that label by my potential sources; although I never
myself used the word, they would speak of my coming for my “lessons” long
after my sophistication in the local language had reached a point where I was
able to posit the probable existence of complex structures I had never sponta-
neously heard and actively plan strategies to elicit them.
Serendipity again came to my aid in carrying out direct e1icitation tests.
All of my sources had television sets, and quiz shows were both popular and
frequent in the programming. These quiz shows were low-key and decorous
affairs by the standards of American television: contestants pushed buzzers to
signal their preparedness to tackle a question, and the contestants as well as
the audiences were quietly excited and eager, with delight at correct answers
expressed in a restrained if readily apparent way. As my rapidly increasing
number of source people grew steadily more accustomed to the task of provid-
ing on demand translations in Gaelic for English sentences, I found that per-
son after person likened him- or herself to one of these quiz show contestants,
often expressing an actual wish for a buzzer to push because the answers to
my questions (as opposed to those on the TV shows) were easily come by and
would have made the respondent feel like “the brain of Britain” (after the title
of one of the shows). Translation from English to Gaelic (or vice versa) was a
common activity in any case, because kinship and friendship networks typi-
cally included both bilinguals and monolinguals and a remark made or heard
in one language would very often be repeated later in the day in the other lan-
guage for an interlocutor with whom the original language of the remark was
not the one regularly used. It took very little effort on my part to make expert
informants of the bilinguals with whom I worked, since they were already
skilled translators and the role of the TV quiz show contestant was one with
which they so readily identified. My skills as an elicitor were put to the test
primarily in gauging which informants could be pushed for more conservative
or elaborate alternative structures without becoming confused or discouraged,
and in finding ways to couch my test sentences so that they would be most
likely to produce the desired results.
Among the secular (non-Anabaptist) Pennsylvania Dutch the situation
was quite different. Clannishness was just as great, but hospitality to outsiders
was a less prominent ethnic value. I had entree to the community in question
through the family of a former student, and my network of speakers was devel-
oped by exploiting that connection. I wanted to work with multiple-generation
families rather than with geographically-bounded or occupationally-defined
groups, and therefore I needed the cooperation of entire large farnilies. The
assent of at least one or two strong-willed central figures in the kin network
was the key to success. Since my original entree was through a former student
and I was more than a decade older than when I began my East Sutherland
work, I had my status as a “professor” to reckon with. Before actually meet-
ing me, one or two people were quite alarmed at the prospect of being inter-
viewed by me, and in one of these cases the individual was an elderly woman
with a heart condition who had to be carefully prepared for my appearance.
She was too anxious to use her faulty English, and I knew that my standard
German would be off-putting, since it represented the language of church ser-
vices. I finally interviewed her in her farmhouse kitchen, a daughter-in-law and
a granddaughter-in-law comfortingly present and preparations for cooking
going on as a partial diversion, and our interview was “macaronic”: I asked my
questions in English and she answered them in Pennsylvania Dutch. She never
spoke any English at all during my visit. By contrast, an equally elderly woman
of strong character and high intelligence marshalled her entire vast family for
me and virtually orchestrated their cooperation. They became so reconciled
to my activities among them that I was encouraged to attend a huge wedding
anniversary celebration at which the entire family would be assembled and
commandeer one by one any strays whom I had not yet had a chance to inter-
view. Only one grandson notorious for exceptional shyness managed to resist
the family pressure for participation in the undertaking.
Again I found that direct elicitation und translation tasks were no obsta-
cle (although this is most definitely not the experience of all investigators of
obsolescing speech forms: Mohan [MS: 54–55] found translation-style elicita-
tion utterly counterproductive). I did not, however, attempt, as I did in East
Sutherland, to acquire control of the local dialect. Knowledge of standard
German, together with plenty of experience of the local dialect, was perfectly
adequate for fieldwork, and outsiders who had acquired Pennsylvania Dutch
were a much comrnoner phenomenon, if not in this day, then in well remem-
bered times, so that becoming a speaker of Berks County Pennsylvania Dutch
would not have gained me the sort of instant quasi-local identity that the acqui-
sition of East Sutherland Gaelic guaranteed. Nor is Berks County Pennsylvania
Dutch as different, relative to other varieties of Pennsylvania Dutch, as East
Sutherland Gaelic is, relative to other varieties of Scottish Gaelic; this again
reduces its value as a marker of distinctive “insider” identity.
I did, on the other hand, do fairly extensive pilot studies with family mem-
bers and family friends of my former student, so that I had quite a good under-
standing of the features of the local dialect before I began my more focussed
study with the two large multi-generational networks. I also spent much more
time and effort on explaining what it was that I was doing, since a number
of the middle-aged members of these networks were well educated and in a
good position to understand the purpose of the study. No one ever referred to
my interviews or elicitation sessions as “lessons”, under these circumstances,
and of course I was not actively acquiring the dialect in any ease. I also never
moved into residence in Berks County, but rather made day trips to the area,
although I was welcome to stay overnight with family or relatives of my former
student and did so on one occasion. The networks I was concentrating on were
somewhat dispersed, in any case, and it would not have been a simple matter,
as it was in East Sutherland, to choose a location central to the study. The par-
ticularly strong-willed matriarch in fact effectively summoned home the more
outlying members of her vast family for special times, so that I could interview
them at her house. Given the size of her family, I could not possibly have com-
pleted the study without her active interest and cooperation, although several
of her children also took an active role in recruiting siblings or their own chil-
dren and in providing settings for interviewing.
Although the published literature on language death and language obso-
lescence deals relatively little with the nitty-gritty of field settings and field
methods, some striking differences in working conditions can be discerned.
One is the ability or inability to work in a community without sponsorship. I
went to East Sutherland because the Linguistic Survey or Scotland asked me to,
but I had no sponsors whatever in the area itself and was warned that I might
not find enough speakers to work with, or gain acceptance among any I did
find and might consequently have to change locations. I found it exceedingly
difficult psychologically to tackle a strange community without any advance
source of entree, but it nonetheless proved perfectly possible in fairly short
order to gain entree without the sponsorship of someone of standing among
identities anyway, and experts that they are in local matters, they will usually
he right; but at least the author will have respected the native reluctance to
name names.
On the whole anthropologists have been quicker than linguists to recognize
the importance of presenting information on the conditions of their work, the
obstacles they encountered, and the reasons (voluntary or involuntary) for
the choices they made among sites, sources, and methods. In an era when
even the “hard” sciences are increasingly acknowledged to be strongly affected
by the human element (see for example Broad and Wade’s volume Betrayers of
the Truth: Fraud and Deceit in the Halls of Science [1982) for some of the nega-
tive aspects of the situation, and Thomas’ charming chapter on “Endotoxin” in
his book The Youngest Science: Notes of a Medicine Watcher [1983] for some of
the positive aspects), linguists would do well to follow this lead for the light it
is certain to shed on the craft and skill with which they ply their profession and
equally on the obstacles and limitations they face in their efforts to advance
the boundaries of their discipline. In the sensitive realm of language death
studies this may be more difficult to accomplish than usual, but the attempt
must be made all the same. This paper is intended among other things as a
contribution to such an undertaking, however restricted in scope.
References
Broad, William, and Nicholas Wade. 1982. Betrayers of the truth: Fraud and deceit in the
halls of science. New York: Simon and Schuster.
Denison, Norman. 1971. Some observations on language variety and plurilingualism. In
Social anthropology and language, ed. by Edwin Ardener, 157–83. London: Tavistock
Publications.
Dixon, R. M. W. MS. Reassigning underlying forms in Yidiɲ – a change during language
death.
Dorian, Nancy C. 1981. Language death: The life cycle of a Scottish Gaelic dialect.
Philadelphia: University of Pennsylvania Press.
Gal, Susan. 1979. Language shift: Social determinants of linguistic change in bilingual
Austria. New York: Academic Press.
——. MS. Phonological style in bilingualism: The interaction of structure and use. To
appear in Meaning, form, and use in context, ed. by. D. Schiffrin. Georgetown:
Georgetown University Press.
Gambhir, Surendra. 1983. Diglossia in dying languages: A case study of Guyanese
Bhojpuri and standard Hindi. Anthropological linguistics 25: 28–38.
Georges, Robert A., and Michael O. Jones. 1980. People studying people: The human ele-
ment in fieldwork. Berkeley: University of California Press.
Haas, Mary R. n.d. Tunica. Handbook of American Indian languages 4. New York:
J. J. Augustin.
——. 1968. The last words of Biloxi. International Journal of American Linguistics 34:
77–84.
Hill, Jane, and Kenneth Hill. 1977. Language death and relexification in Tlaxcalan
Nahuatl. International Journal of the Sociology of Language 12: 55–69.
——. 1981. Regularities in vocabulary replacement in Modern Nahuatl. International
Journal of American Linguistics 47: 215–26.
Huffines, Marion Lois. 1980. Pennsylvania German: Maintenance and shift. International
Journal of the Sociology of Language 25: 43–57.
Melchers, Gunnel. 1981. The Norn element in Shetland dialect today: A case of ‘never-
accepted’ language death. In Tvåspråkighet, ed. by Eva Ejerhed and Inger Henrysson,
254–61. Umeå: Umeå University.
Miller, Wick R. 1971. The death of language or serendipity among the Shoshoni.
Anthropological linguistics 13: 114–20.
Mithun, Marianne, and Reginald Henry. MS. Notes on moribundity: Incipient obsoles-
cence of Oklahoma Iroquois.
Mohan, Peggy. MS. A language implodes: The death of Trinidad Bhojpuri.
Pulte, William. 1973. Cherokee: A flourishing or obsolescing language? In Language in
many ways, ed. by William B. McCormack and Sol Wurm, 423–32. The Hague:
Mouton.
Salzman, Zdeněk. 1969. Salvage phonology of Gros Ventre (Atsina). International
Journal of American Linguistics 35: 307–14.
Schmidt, Annette. 1983. Young people’s Dyirbal: An example of language death from
Australia. Australian National University Master’s thesis.
Swadesh, Morris. 1948. Sociologic notes on obsolescent languages. International
Journal of American Linguistics 14: 226–35.
Thomas, Lewis. 1983. The youngest science: Notes of a medicine-watcher. New York: The
Viking Press.
Timm, Lenora A. 1980. Bilingualism, diglossia and language shift in Brittany.
International Journal of the Sociology of Language 25: 29–41.
——. 1984. The segmental phonology of Carhaisien Breton. Zeitschrift für celtische
Philologie 40: 118–92.
Trudgill, Peter. 1976–77. Creolization in reverse: Reduction and simplification in the
Albanian dialects of Greece. Transactions of the Philological Society 1976–77: 32–50.
Ttsitsipis, Lukas D. 1983. Language shift among the Albanian speakers of Greece.
Anthropological linguistics 25: 288–308.
Voegelin, C. F. and F. M. Voegelin. 1977. Is Tübatulabal de-acquisition relevant to theo-
ries of language acquisition? International Journal of American Linguistics 43:
333–38.
methods course was offered during the summer sessions, and a number of fac-
ulty members were themselves either experienced fieldworkers or well-trained
Middle English lexicographers who also needed to exercise careful patience in
amassing and accounting for primary data. Among the senior faculty, Kenneth
Pike was a major presence, and his prowess as a fieldworker was legendary. His
occasional “monolingual demonstrations” made a vivid impression on all of us.
A speaker of some language unknown to Pike was produced, sworn to speaking
his or her mother tongue exclusively, while Pike, armed with a few props such
as two sticks and a leaf, asked questions using only Mixteco (a Mexican Indian
language he spoke fluently) and miming. Pike wrote everything the speaker of
language X said on a blackboard and after half an hour performed an instant
grammatical analysis on the material. This was awe-inspiring to watch and
no doubt created a certain fieldwork mystique among linguistics graduate
students.
During my graduate studies Old English had been a delight to me. Here was
English as I thought it ought to be, a fully Germanic tongue without the overlay
of Romance and Latinate vocabulary that seemed chiefly to serve the causes
of euphemism and hypocrisy (“prevaricate” indeed, if the lady had lied!). But
somehow a dissertation on Old English struck me as an improper use of my
training. I’d been given to understand that a linguist could use the field method
techniques we had been taught anywhere, with any language. And since the
techniques could be applied anywhere at all, why not go where my interest was
highest and try them out on a language I’d wanted to learn more about since
childhood, namely Scottish Gaelic?
2 Somewhere Ho!
Good advice is a boon when you’re contemplating fieldwork and the prob-
lem of funding it. Mine came from Eric Hamp, famed Celtic scholar at the
University of Chicago, who suggested that I link my fieldwork to the needs of
the Gaelic Division of the Linguistic Survey of Scotland by offering to write a
dissertation on whatever dialect the Survey director considered most in need
of study. Hamp predicted that funding sources would see a study linked to an
established project as well-focused and worthy of support, as one soon did.
I was content to go wherever the Survey directed me and work on whatever
project they proposed, since I was off to the country of my choice to work on
my top-choice language.
In correspondence the Survey director had indicated that a phonological
study of the Gaelic spoken in any one of three different Highland locations
would be highly suitable from the Survey’s point of view. He seemed at the
time to be leaving the final choice to me, but soon after my arrival in Edinburgh
he handed me a list of names and wished me well for my work in eastern
Sutherland – not the location I had mentally picked for myself among the
three, and a daunting distance away on the map, almost as far to the north as
one could go without dropping off the mainland altogether.
At the time I wasn’t familiar enough with Gaelic dialectology to understand
the motivation for the director’s choice. He, however, knew that the Gaelic
spoken by the fisherfolk of eastern Sutherland was a dialect of the extreme
periphery, and that in classical fashion it differed notably from more central
dialects. Furthermore, it was certainly understudied. The speakers whose
names appeared on the director’s list had provided answers to a vast question-
naire made up almost entirely of isolated lexical items used by the Survey to
track historical phonological development across the whole of Gaelic-speaking
Scotland. The questionnaire was well designed for its limited purpose, and the
fieldworkers were skilled at their jobs, but most of Scotland’s local Gaelic dia-
lects were otherwise poorly known, especially those of the northern and east-
ern mainland. The director’s final words to me reflected the extreme scarcity
of solid information about the Gaelic of East Sutherland in the early 1960s. He
warned me that I might find no speakers left in the three coastal fishing com-
munities he was sending me to, in which case I was to come back and he would
give me another assignment; and he urged me to find out, if I did locate speak-
ers, whether it was really true that the Gaelic of eastern Sutherland lacked
preaspiration of voiceless stops and affricates (preaspiration being a striking
phonological feature of most Scottish Gaelic dialects). Far from finding no one
to work with, I soon had an informal census of local Gaelic speakers running
to more than 200 people, and the absence of preaspiration, so difficult for the
director to credit, proved to be one of the most obvious general features of
the whole dialect area, with implications for other parts of the phonological
system.
The 200 or so local Gaelic speakers still available in East Sutherland did not
include many of the people whose names were on the Survey’s list. Survey field-
workers had moved through eastern Sutherland in 1953 and 1957, and most of
the elderly speakers who had served as their sources, or had been mentioned
to them as possible additional sources, had died before I reached the area in
1963. After one man who had survived turned out to be lively enough at 86
to make it advisable to keep a table between us at all times, I abandoned the
Survey list and searched out my own sources.
Whether I was relying on Survey sources or not, my work was still neces-
sarily tied to the Survey’s interests. My funding had been granted on the
understanding that I would target my research to their needs, and beyond that
they had given me the use of a Survey van for the year. Sutherland has been
thinly populated since early in the nineteenth century, when most of the ten-
antry of the great Highland estates were summarily evicted, often with con-
spicuous brutality, in order to “clear” the land for sheep farms. There was some
distance between villages, and because of the low population public transport
was much scantier than is typical of most of Britain. It was impossible to get
from one village to another and back again on the same day without private
transport, and since the Survey wanted me to cover three villages, the van was
quite simply a necessity.
Probably the three-village assignment should have alerted me to complexi-
ties lying in wait for me in Sutherland, but I only supposed that the Survey
director thought I might have to comb through three villages to find enough
people to work with. Once I was on location the inconvenient truth of the mat-
ter broke over me very quickly: my work was not the relatively simple job of
describing a uniform fisherfolk variety of eastern Sutherlandshire Gaelic, but
the very much more complicated job of describing each of three slightly dif-
ferent local varieties of fisherfolk Gaelic, one for each village. In the standard
field methods fashion that I had been taught, I began my work by eliciting
commonplace vocabulary that was likely to be monosyllabic, or at least short.
“What do you say for ‘garden’?” I asked. “/yεs/,” said an elderly lady in Brora, the
northernmost village. “/l´εs/,” said her counterpart in Golspie, seven road-miles
to the south. “/l´es/,” said a woman in Embo, ten road-miles south of Golspie.
These were small enough differences, but there was worse to come. ‘Bone’
proved to be /khrẽ:ũ/ in Brora and Golspie, with plural /khrã:vən/; in Embo it
was /khrãĩ:/ with plural /khrã:n/. Even when it came to a word as central to the
lives of all these fisherfolk descendants as ‘sea’, they didn’t agree: the word was
/mur/ in Brora and Embo, but /mwir/ in Golspie. Things were no better when I
moved from single words to connected material. My original Brora and Golspie
sources gave ‘if you don’t plant oats’ as /mər khur u khɔrkh/, but the equivalent
in Embo was /mə khur u khɔrkh/ or /mə gur u khɔrkh/, using a different form of
the conjunction ‘if . . . not’ (the first word in each example) and one with vari-
able effects on the initial consonant of the following word. The fact that the
Gaelic of these three fishing communities – so similar in their historical origins,
so close to one another (especially by sea, once the chief communicative link),
and so nearly identical in all economic and social aspects – differed in each
locality had immediate consequences for my work. Every word or sentence I
gathered had to be checked across all three villages, lest there prove to be local
differences. And since there often were such differences, which then had to be
checked for possible individual idiosyncrasies, it wasn’t good enough to have
people of that description were the ones least likely to have been influenced
by any non-local usages they might have been exposed to. Luckily for my future
work, it proved difficult in East Sutherland to isolate the speakers who best
met the Survey’s criteria from their usual well-peopled social contexts. I was
working with them in their own homes, and in several households there were
Gaelic-speaking spouses or siblings who were younger than the speaker I had
specifically come to work with, plus occasionally a grown son or daughter who
spoke some Gaelic. Answers of their own popped out eagerly from some of
these others when I put questions to the older speaker. Being young, polite,
and deeply grateful to all the families who let me into their homes and toler-
ated my interminable questions, I considered it proper to write down whatever
was offered. So I recorded these extraneous responses, too, and found myself
confronted yet again by uncomfortably diverse data. I wasn’t getting reliably
identical responses, even though my sources in these cases were not just from
the same village but from the same household.
The material from younger family members didn’t find much place in my
dissertation, since that document was also in effect my report to the Survey,
and Survey standards excluded material from such sources. But it was in my
notebooks, as was a small amount of material from the very few elderly Gaelic-
speaking crofters (sub-subsistence agriculturalists) whom I unearthed in the
rural districts round about the three villages when the Survey director handed
me another assignment: gathering Gaelic place names for the place-name spe-
cialists of the School of Scottish Studies in Edinburgh. Supplied with bundles
of oversized map segments, I quickly covered the areas that were well known
to my fisherfolk friends, after which I dutifully headed off into the countryside
to follow up on uncertain reports of occasional elderly crofters who still spoke
Gaelic. I found three, each one the lone surviving Gaelic speaker of his dis-
trict, and with their help dotted the highly detailed maps with Gaelic names
for cleft, knoll, hillside, rivulet, and so forth: ancient indigenous place names
certain to be lost all too soon. I worked on a bit with one crofter after the place-
name task was complete, enjoying the visits to his particularly pleasant family
and intrigued by the obvious small differences between his Gaelic and that of
the nearest fisherfolk. I was interested, for example, in a number of initial con-
sonant clusters with a prominent bilabial second element (as in the Golspie
word /mwir/ ‘sea’) that were typical of the Gaelic of the fisherfolk communi-
ties. I knew these to be unusual in terms of western Gaelic dialects, and now I
found that they were absent even in the crofter Gaelic once spoken very near
at hand.
Tying my fieldwork to the Linguistic Survey of Scotland’s interests had a
good many consequences. Fieldwork funding and the loan of a car were obvi-
It doesn’t seem to be necessary to like the people one is studying very much in
order to do productive fieldwork. When Malinowski’s diaries were published
posthumously (1967), it appeared that he had not had a great liking or respect
for the Trobriand Islanders (Van Maanen 1988: 36), and Erving Goffman told
me, when I had a chance once to ask him directly, that he had not particularly
liked the Shetland Islanders about whom he wrote so illuminatingly in The
Presentation of Self in Everyday Life (1959). But liking the people you work with,
as I did, certainly makes the fieldwork experience more enjoyable, and for
some personalities and casts of mind it may be an important factor in deter-
mining the course of future research, since it enhances the appeal of returning
to work in depth in a particular fieldwork site.
Like many another sojourner in the Highlands, I was astonished by the
generosity of people in whose midst I appeared as an unannounced stranger.
People were often slow to believe that I could be interested in their local
Gaelic, since they had heard nothing but negative comments about it all their
lives, both from English monolinguals and from speakers of more conserva-
tive westerly Gaelic dialects. But once convinced of my interest, most people
showed an almost unlimited willingness in helping me learn about it. Payment
was out of the question, since the very mention of it proved offensive, and the
small hostess gifts that I learned were acceptable at each of my visits, seemed
completely inadequate thanks to people who were giving up whole afternoons
or evenings to answering my questions and were regularly pressing great quan-
tities of tea and baked goods on me besides. Even after I left that first year, five
of my sources carried on answering my questions, putting long lists of phono-
logically relevant lexical items and short sentences onto tape for me so that I
could consult this material during the ten months that I had spectrographic
equipment at my disposal while working on my dissertation. All of the tapes
that arrived proved to be the spoken equivalents of letters as well, with added
messages giving me news and good wishes in Gaelic, and sometimes including
general Gaelic chats among my friends. These were not people easily forgotten.
Furthermore, I genuinely liked the East Sutherland variety of Gaelic from an
aesthetic point of view, especially the Scandinavian-sounding tonality of its
longest vowels and the unusual sonority of its many uninterrupted multivowel
sequences. I was also acutely aware that my year’s fieldwork had been barely
adequate even to the single task of describing the phonology of this distinctive
and little-known Gaelic variety. I lived frugally while I worked on my disserta-
tion, saved money from my fellowship, and left for Scotland again five days
after defending the dissertation.
Hard pressed though I’d occasionally felt, as I made my perpetual swings
from village to village and fanned out into the countryside with the place-
name survey maps, my limited connections with the Linguistic Survey and the
School of Scottish Studies taught me very quickly how precious and how frag-
ile the store of human knowledge and experience among the dwindling Gaelic
speakers of East Sutherland was. The material most coveted by the place-name
experts, for example, was not the Gaelic place names of East Sutherland itself,
but the far rarer Gaelic place names the fisherfolk knew for ports farther down
the east coast of Scotland, where Gaelic had not been spoken for centuries.
The uniqueness of such knowledge, and the finality of this chance to capture
it while some Gaelic-speaking fisherfolk still remained on the east coast, was
impressed on me, and it stirred the incipient cultural conservator in me.
Phonologically I had certainly encountered phenomena that were curiosi-
ties for a Scottish Gaelic dialect, from those initial consonant clusters with /w/
as second member to word-final geminate consonants in unstressed syllables
(these last difficult to hear until I met a few of them before vowel-initial words
within the same noun phrase). That is to say, I already had evidence, by the
end of one year’s work, that East Sutherland Gaelic (ESG) was unusual in more
respects than the absence of preaspiration, and I suspected that more surprises
might come my way if I spent more time with the dialect. Another reason for
my return trips to East Sutherland, in 1965 and after, was that I’d begun to feel
a responsibility to document this unusual variety of Gaelic that clearly had a
short life-expectancy; family transmission had ceased in this area, and there
were no longer any young speakers. It didn’t hurt, either, that the place was
beautiful (even if one could hardly say the same for the climate) and that most
of the people I worked with personified a fieldworker’s dreams.
Originally I worked mostly by elicitation, which by good fortune my sources
found congenial and easy (not by any means always the case in fieldwork).
My field methods training had stressed elicitation, but the conversational limi-
tations of my Gaelic were a more important reason for relying on it. Asking
people to produce stories addressed to a tape recorder felt uncomfortable
in purely social terms. The tape recorder provided no social reinforcement,
such as smiles at humorous bits or nods in response to rhetorical questions,
and until I was comfortable enough in the language to supply these ordinary
human responses while the story was in progress, I was reluctant to put a
microphone in front of people and ask them to tell stories into it. Knowing
I should gather texts to exemplify the Gaelic of the three villages, I had done
a few taping sessions at the end of my first year, but with my halting Gaelic
I found them extremely awkward. It wasn’t until 1967–68, when I was finally
comfortable enough in the local Gaelic to make a reasonable conversational
partner, that I did a more significant amount of taping; but by then I was work-
ing hard on grammar, which again made elicitation (translation tests) the tech-
nique of choice. I needed to cover a lot of grammatical territory, and since my
sources had proved to handle elicitation with extraordinary ease and even with
pleasure – several said it made them feel like the brainy, rapid-fire responders
on a popular TV quiz show – elicitation was an efficient way to go about it.
As it turned out, elicitation had an unanticipated benefit. The social context
in East Sutherland, and especially in Embo, where there was a larger pool of
speakers, continued to favor fluid work sessions with more than one family
member present. I tended to ask for a good many examples of any structure I
was exploring, and during the course of a session a variety of individuals might
give their versions of a particular structure. Over time it became apparent that
people closely connected with one another were far from unanimous about
how certain grammatical niceties were to be expressed. Because I spent a fair
amount of purely social time in some of these households, I also heard sponta-
neous usages that strengthened an impression of ongoing grammatical change
in certain constructions. Eventually it seemed important to check on this, and
I embarked on batteries of translation tests designed to elicit key constructions
from across the widest age-range of speakers available. It also seemed useful to
go back, as I did with pleasure and interest, to some of the imperfect speakers
whom I’d been sorry to drop from my speaker sample earlier on.
Most of the linguistic variation that was being investigated in the 1960s and
1970s was phonological, as in large part it still is today. In ESG, for whatever
reason, there was relatively little phonological change in evidence, but a good
deal of grammatical change was underway. I had been much impressed by
Labovian studies demonstrating correlations between phonological change
and social factors such as age, ethnicity, social class, and sex, so I looked long
and hard at one clearly advancing phonological change, substitution of [ᶕ l]
for the more traditional velarized lateral [ɫ]. But only age seemed to have any
bearing on how general the use of [ᶕ l] for [ɫ] became (some younger speak-
ers in Embo having begun to use [ᶕ l] even in word-initial position). When it
came to grammatical change, the same was true: age played a clear role in the
extent to which an observably advancing change appeared, but no other social
correlates emerged.
The age differences led me to wonder whether, in excluding younger people
as sources, descriptive linguists, who typically insisted on working only with
the most traditional speakers, were missing an opportunity to find out just
what sorts of changes might be likely to occur as a small and highly localized
speech form went out of use. I made a point of enlarging my speaker sam-
ple again, this time in Embo, where Gaelic was still widely used and speakers
ranged in age from the eighties to the low forties, or even to the upper thirties,
if I included some individuals who spoke Gaelic imperfectly with certain older
relatives. The results of translation tests presented to Embo’s broad age-range
of speakers showed, among other things, that case distinctions were progres-
sively weakening and that one traditional form of the passive was being aban-
doned (though it was leaving its trace in changes introduced into the other
traditional passive (Dorian 1973)). Certainly it was gratifying to find the sort of
age-graded changes I’d anticipated when I started probing for these and other
grammatical changes. But I was struck, at the same time, by the moderation of
many of the changes I looked at. Gender signaling via pronoun reference, for
example, was notably weakening, but Gaelic has a number of gender-signaling
devices and one or two of the others weren’t showing comparable weakening.
It was true that a particularly conservative passive construction was fading out
of use, but the passive itself was still fully expressible in ESG, even among the
stronger of the imperfect speakers. The hyperabundance of plural and gerund
allomorphs in ESG was diminishing, but it wasn’t anywhere near the logical
extreme of one universally applied suffix, either for plural or for gerund. It was
very far from it in fact: even the imperfect speakers still showed plenty of vari-
ety in each case (Dorian 1978b). The limited nature of grammatical “decay”
in ESG, even with the dialect’s ultimate extinction in sight, seemed to me as
significant a finding as the presence of age-related grammatical change, and
I tried to give it equal attention.1
1 In retrospect this seems even more important than it did at the time, since three linguists
working with geographically and structurally very different languages have lately found strik-
ing evidence of grammatical elaboration among the final speakers of obsolescent languages:
Rob Pensalfini (1999) in Jingulu, an Australian Aboriginal language; Alexandra Aikhenvald
(in press) in Tariana, an Arawakan language of the Brazilian Amazon; and Silvia Dal Negro
(1998) in Pomattertitsch, a Walser dialect of northern Italy. Obsolescence processes clearly
needn’t be an unremitting progression into collapse and decay.
the speakers I worked with had experienced virtually identical social and eco-
nomic conditions: all lived in a few densely populated streets, in houses of the
same general structure and in households sharing identical labor patterns; all
were poor and burdened by the same social stigma; all spoke Gaelic in the
home; all came from a highly conservative Protestant religious background.
Almost no one lived in the fisherfolk streets who did not fish for a living, and
after the school years, finished by age 14 in nearly every case, contacts with
non-fisherfolk were limited and almost entirely commercial. Even religious
life was socially segregated, since there were separate services in Gaelic and in
English, with the former attended chiefly by the fisherfolk.
Variationist studies have long since demonstrated that the social features of
large urban populations in particular, and even the generally smaller number
of social distinctions within rural populations, find expression in significant
patterns of similarity and difference in the use of phonological and grammati-
cal features. I had gone to the Highlands expecting to find the same sort of
phenomena there as well, yet years had passed and I had had nothing of this
sort to report on. The very socioeconomic uniformity just described might play
some role, of course, and if asked about my lack of findings that’s certainly
what I would have pointed to. Yet there wasn’t any shortage of variability. Just
the opposite, in fact – there was rampant variability.
Faced with this problem, I realized that at last I stood to reap the rewards of
the three-village assignment set me by the Survey. Because I had always worked
in all three villages and had regularly documented their distinct usages, I knew
the purely geographical dimension of ESG variability intimately. I could there-
fore subtract that form of variation, as well as the strongly age-related varia-
tion I had already looked at, and focus on the intra-village and intra-speaker
variation that remained. I had recognized this sort of variation early, because
it turned up among my sources in puzzling ways. Among my early sources,
an Embo brother and sister were unusual in having no other siblings, and in
both having married within the home village and lived there lifelong. They
also happened to live in adjoining houses as adults and to have a good deal
of daily contact. Yet although they claimed they had never noticed it, their
speech habits were mysteriously different: the sister, the elder by four years,
used /stε/ by preference for adverbial ‘in’, the brother /sčax/; the sister favored
/tə(nə)/ for conjunctional ‘when’, the brother /nə(rə)/; the sister used /mwĩç/
for the locational form of the adverb ‘out’, the brother /mwĩ/; the sister used
monosyllabic /hãn/ more often than /hãnig/ for ‘came’ and /hũn/ as well as
/hũnig/ for ‘saw’, while the brother used only the disyllabic forms of each. Since
they were close in age but of opposite sex, the most obvious hypothesis was
that these were sex differentiated usages in Embo. But that simply wasn’t the
case, as even the most minimal checking quickly showed. The problem, in fact,
was that there were no apparent social explanations for this very prevalent
kind of variation: socioeconomic background was uniform; age didn’t play the
obvious role here that it did in the identifiable changes in progress (although
decreasing age could be shown to correlate with a trend toward the favoring
of certain variables in several instances); sex could usually be eliminated as
a factor; and there was no clustering of favored variants among people who
had lived in the same street. I had variation in plenty; what I didn’t have was
an explanation for it in terms that variationist studies would have predicted.
This is a fascinating conundrum the full dimensions of which I’m still track-
ing, in fact, especially since learning how to tape-record from the phone (with
the permission of those on the other end, needless to say). The still growing
database so far supports certain conclusions to which I was inclined in 1994,
when on the (mistaken) assumption that I wouldn’t be able to expand my data-
base much, I wrote about the matter (Dorian 1994). The most fundamental
of these was that social homogeneity need not imply linguistic homogeneity.
Where the two do not correlate, it seems by the East Sutherland evidence that
three conditions may play an important part. First, some circumstance must
lead to the emergence of an array of variants. The terrible upheaval of the nine-
teenth-century evictions, in the fisherfolk case, with some degree of popula-
tion mixture occurring at that time, may account for some of the variation in
East Sutherland, and processes of language change for a bit more (decay of for-
mer grammatical distinctions, for example). Second, some circumstance must
prevent particular variants from acquiring a link with particular social features
among groups within the population of speakers. In the fishing communities,
small population size and density of interaction, plus a notably uniform socio-
economic background, presumably play this role. Third, some circumstance
must impede local speakers’ access to any standard-language norm that may
exist for the language and keep them from developing normative judgments
in connection with local variants. In the fishing communities the aberrance
of the local dialect (which made importation of church-Gaelic norms or more
mainstream-dialect norms unworkable) and Gaelic illiteracy (women) or very
limited literacy (most men) have this effect.
One critical question that the high degree of intra-village and intra-speaker
variability in the fisherfolk communities raises is this: if ESG currently repre-
sents the only clear-cut case of such prominent but socially unmarked vari-
ability, as it appears to, is that because these former fishing communities are
genuinely unusual, or is it because the way fieldwork is normally practiced, and
to what ends, has precluded recognition of similar cases? There is evidence
I kept quiet. Partly because I couldn’t bring myself to objectify the situa-
tion so quickly, and partly because of Hapiya. . . . He had given us some of
his sacred medicine knowledge, a bit of his own life, his own breath. . . .
I also kept quiet because we had something difficult to tell Hapiya. We
were starting up new fieldwork, and this time it was far from the
Southwest, in Guatemala. It was hard to find the words to explain to him
why we would study elsewhere.
People don’t see themselves as objects of study. Finding that others do see them
that way produces strong reactions. More often than not the reactions are neg-
ative, as some eloquent Native American responses to anthropologists’ studies
have demonstrated (Deloria 1969); but occasionally a sense of validation and
self-worth is roused instead. Social bias against the people who became the
East Sutherland fisherfolk arose shortly after 1800, at the time of their invol-
References
———. 1994. Varieties of variation in a very small place: Social homogeneity, prestige
norms, and linguistic variation. Language 70: 631–96.
Goffman, Erving. 1959. The presentation of self in everyday life. Garden City, NY:
Doubleday.
Malinowski, Bronislaw. 1967. A Diary in the strict sense of the term. New York: Harcourt,
Brace and World.
Milroy, Lesley. 1980. Language and social networks. Oxford: Basil Blackwell.
Ó Dochartaigh, Cathair. 1983. Review of Dorian 1978a. Scottish Gaelic Studies
14: 120–28.
Oftedal, Magne. 1956. The Gaelic of Leurbost, Isle of Lewis. (A Linguistic Survey of the
Gaelic Dialects of Scotland, 3: Norsk Tidsskrift for Sprogvidenskap, supplementary
vol. 4) Oslo: Aschehoug.
Pensalfini, Rob. 1999. The rise of case-suffixes as discourse markers in Jingulu – a case
of innovation in an obsolescent language. Australian Journal of Linguistics 19:
225–40.
Tedlock, Barbara. 1992. The beautiful and the dangerous: Encounters with the Zuni
Indians. New York: Viking.
Van Maanen, John. 1988. Tales of the field: On writing ethnography. Chicago: University
of Chicago Press.
Any researcher who leaves a record of his or her work assumes a number of
responsibilities, chief among them responsibility for the record’s accuracy. But
when an endangered speech form is involved, as in the chief case I will discuss
here, what might be called the ‘last-chance’ responsibility comes strongly into
play: right now may be the one and only chance to create a record of the speech
form in question, and right or wrong, what the late-stage fieldworker puts on
the record is likely to stand.
A number of difficulties, some of them more immediately obvious than oth-
ers, may lie in the way of a researcher who appears at a late stage in the his-
tory of a receding language, hoping and intending to leave a reliable linguistic
record. If all of the remaining speakers are elderly, it may be that neither the
current researcher nor future researchers will have an opportunity to gather
additional material for confirmation or refutation of the original record. This is
unfortunately a very common circumstance in research with at-risk languages.
If the sampling procedure used by the researcher is inadvertently skewed in
some fashion, a source who is unrepresentative of the speech community may
come to represent the community in the official record of that community’s
speech. This happened, for example, with the record for Golspie village in the
five-volume Survey of the Gaelic Dialects of Scotland (Ó Dochartaigh 1997),
where an overly literate man served as the sole source for the village, supply-
ing written-language substitutes for some local dialect forms (see Dorian 2010,
Chapter 9). If the local language is no longer used in ordinary conversation,
the researcher may feel obliged to question the naturalness or completeness
of such speech as can be retrieved for the record. Haas raised these questions
in her work with the last speaker of Tunica, a man who had had no fluent con-
versation partners for many years, even though he himself appeared to repre-
sent a high degree of fluency (Haas 1941). If there are no longer any speakers
who know how the language was used in connection with certain traditional
practices, it may be impossible to gain a sense of the full semantic range of cer-
tain lexical items or expressions. Jocks (1998) describes the dimensions of the
semantic-range problem particularly well, coming to it as an adult learner of
Mohawk. If the speech community tolerates, or even embraces, a considerable
amount of familial or idiosyncratic variation in the ways that Collins (1998)
describes for Tolowa and Kroskrity (2002) for Western Mono, the researcher
may unknowingly take the forms he has recorded to be much more generally
representative of a local speech form than they actually are. It should of course
be acknowledged that misapprehending some aspects of the data and failing
to appreciate the full semantic content of some of the recorded material are
not problems unique to working with receding languages. But these problems
are heightened in late-stage fieldwork, because of limited opportunity to sup-
plement or correct the record.
Impetus for receding-language fieldwork may come entirely from the out-
side, with researchers arriving to look for languages reported still to survive,
as with David Bradley’s quest for remaining speakers of Ugong in Thailand
(Bradley 1989), or it may arise from speakers’ own concern for the future of
their speech form, as was true for Faetar in Italy (Nagy 2000) and for Rama in
Nicaragua (Grinevald 2006). In the latter case sources are readily identifiable,
but in the former case the researcher may have to hunt for elusive speakers. In
particularly favorable cases the researcher may find speakers who have devel-
oped their own sense of mission about leaving a record of their language and
are glad to work with a linguist to achieve that goal. In northeastern Australia
the last Warrungu speaker, Alf Palmer, told researcher Tasaku Tsunoda, “When
I die, this language will die. I’ll teach you everything I know, so put it down
properly” (Tsunoda 2005: 98). But as James Collins found in working with a
thin scattering of Tolowa speakers in northern California, the linguist’s nar-
row focus on contrastive forms and their distributions can be a very long way
from what the remaining speakers have in mind when agreeing to a joint effort
to record their language. “Simply put”, writes Collins, “they were interested in
words, not grammar” (Collins 1998: 260; see also Grinevald 2001: 295). Such dis-
crepancies suggest the potential for conflicting objectives in any joint work
involving academic researchers and community members and for discor-
dant notions on the part of the two parties about the responsibilities of the
researcher in the wake of that work.
1 A fourth constituency not discussed here, some sort of funding agency, may or may not be
involved. For prolonged fieldwork in distant locations, institutional financial support is a
Some results of the scholar’s activities may serve all three constituencies – for
example, a clear, user-friendly, and accessible dictionary of the language. But it
can also happen that the interests of the various stakeholders are at odds and
can not easily be reconciled. In that unfavorable case, serving one set of stake-
holders well may mean serving another set poorly or not at all.
If fieldwork with a single speech variety extends over some years, the sheer
passage of time almost guarantees that the researcher’s position with regard
to her responsibilities will undergo some change. In my own case, 45 years of
work with an isolated and unusual variety of Scottish Gaelic provided plenty of
opportunity to reconsider the focus and scope of those responsibilities, thanks
both to blunders that I made as a young researcher and to changing circum-
stances in the scholarly world and also in the ethnic community. Discussing the
blunders in particular offers a way of focusing on potential ethical dilemmas in
linguistic fieldwork, in particular fieldwork with a rapidly receding language.
My orientation as a young scholar in the early 1960s was typical for the time,
I believe, in that I considered my research to be undertaken in the interest of
other scholars and my responsibility to be primarily to the scholarly commu-
nity. While I certainly felt a strong connection to the people I was working with
from the very beginning, my orientation at the time conformed to the pattern
that Himmelmann has described as typical of twentieth century structuralist
linguistics (Himmelmann 2008: 341): I did not take the results of my fieldwork
to be of any particular interest to the people whose language I was studying.
The variety of Scottish Gaelic that I worked with had about 200 speakers in
1964; it currently has three less-than-perfectly-fluent speakers (four, if I count
myself). It was a dialect of the extreme Highland periphery and was as atypi-
cal as peripheral dialects often are; it had been recorded up to then only in the
form of lexical entries in a Gaelic dialect survey. The coastal East Sutherland
Gaelic speakers had been fisherfolk, an occupation that created a separate
Gaelic-speaking workforce that for some generations had needed only rela-
tively limited English for commercial transactions outside the community and
for part-year occupational involvement in the national herring fishery. Separate
residential areas for fisherfolk had permitted community members to main-
tain their home and neighborhood use of Gaelic well into a period when other
population segments in the surrounding region had become monolingual in
English. But the fishing had come to an end after World War II, and by the
1960s most local speakers were elderly; only a few were under 40, and no chil-
dren were acquiring the distinctive local dialect. The end of this speech form
necessity, but receding languages may also be encountered much closer by. I worked for
example with secular Pennsylvania German speakers who were within manageable driving
range of my home institution without requiring funding for that research.
2 I specify “locally resident” because it’s possible, though by now highly unlikely, that one or two
fluent speakers from the East Sutherland diaspora survive in New Zealand, say, or Canada.
3 The sensitive subject is by-naming, a practice similar to nick-naming but with such promi-
nent elements of mockery that a great many by-names are offensive. As a result, by-names,
unlike nicknames, are not used in direct address, even though they are universally used in
reference.
Sutherland fisherfolk, and I wanted to give some control over the book to the
couple whose tape-recorded reminiscences supplied nearly all of the direct
quotations in the book. I sent them the full manuscript before publication, so
that they could ask to have anything they objected to taken out, and I removed
two items of somewhat personal information at the wife’s request. But even
so, when the book came out the wife (by then a widow) was distressed by two
other matters that she hadn’t noticed on her first reading. This was extremely
painful for me as well as for her, since it was the outcome I had most especially
been trying to avoid; it took the kindly intervention of her children to convince
her that she was not in fact negatively exposed in a public way, as she feared.
It was certainly an instructive experience for me. I had been working in these
communities for fully a decade and a half by the time the oral history appeared,
yet of the four matters that proved sensitive for one of the principals, only one
had actually struck me as likely to be problematic when I was writing the book.
At bottom the issue is by now a familiar one: whose are the materials the field-
worker has gathered, and for what purposes can they legitimately be used?
Different constituencies may have quite different responses. In providing
material to an outsider who is a scholar, community members may seem to be
agreeing to share their knowledge with the wider world. But since local speak-
ers seldom fully understand what scholars do with the materials they gather,
the agreement is more apparent than real. In recent years this problem has
surfaced most conspicuously in connection with the development and expan-
sion of the internet. We recognize very easily that as little as 20 years ago no
one could have foreseen the ease with which specialized information would
circulate via the internet and how broadly access to it would be gained. What
we do not as easily recognize is that for most of the twentieth century schol-
arly publications of any kind were just as unimaginable to most of the local-
language speakers who provided researchers with their source material as the
internet was to all of us until recently: local people usually had no experience
of academic books or journals and no idea who used them, so that there was
no realistic possibility that they could envisage where their materials would
appear and how they would be made use of.
The impossibility of making the purposes of our fieldwork understandable
to our sources is the bedrock dilemma of researcher responsibilities. If we can’t
convey our intentions and our goals to those whose knowledge we propose to
tap into, then we can’t obtain truly informed consent for the work we’re doing.
In the case of the fisherfolk couple who were the central figures in my oral
history, two people fully literate in English, I described the project to them in
advance, gave them time to discuss their participation between themselves,
negotiated a written statement that gave me permission to use their tape
recordings and them a guarantee that I would protect identities,4 sent them
the manuscript before publication, and removed at their request two items
identified at that stage as overly personal. But because the publication process
was unfamiliar to them, it was not obvious to them, as I had assumed it would
be, that the manuscript stage was the only chance there would be to have items
removed. In just the same way, there would have been no possibility of making
intelligible or persuasive to local people a scholar’s experience that the locally
sensitive examples used in my early sociolinguistic paper would be intellectu-
ally interesting and yet socially neutral to the academic readers that that paper
was intended for.
A particularly poignant example of the gulf created by incommunicable
purposes appears in Barbara Tedlock’s description of her and her husband’s
final visit to their long-standing Zuni teacher, Hapiya, in New Mexico. For
the first time ever, Hapiya treated them with Zuni medicine, an act of great
friendship and trust because the Zuni were well aware that outsiders were usu-
ally scornful of their medicine. Much as the anthropologist in her wished to
explore this unexpected medical treatment with Hapiya, Tedlock felt unable
to do so because of the extremely awkward moment that lay ahead: she and
her husband faced the necessity of telling Hapiya that they were leaving to
work with another people altogether, in Guatemala. She writes (Tedlock 1992:
287): “It was hard to find the words to explain to him why we would study
elsewhere”. Indeed it must have been. “Friend” is a graspable concept in most
societies, but “scholar”, with its considerably lesser personal commitment, is
much less so. Lesley Milroy describes a dilemma similar to Tedlock’s that arose
from her Belfast (Northern Ireland) fieldwork:
4 They later agreed to allow their own real names to appear in the published book.
How are people who are not part of the scholarly world to understand a con-
cept like “the observation period”, and what would they think of it if it could
be explained to them?
In all of the cases mentioned here, the researcher is the one who is altogether
out of line with normal social expectations. How could an ordinary speaker
of Tolowa fathom the researcher’s request for endless repetitions of a single
Tolowa speech form in the interests of establishing a particular phonologi-
cal contrast? What does someone for whom language entails social interac-
tion make of a linguist’s insistence on eliciting complete paradigms? And how
could any ordinary person, after the sharing of knowledge and experience that
extended fieldwork requires, view the published use of sensitive material or the
abrupt cessation of long-continued personal contact as anything but evidence
of hypocrisy and exploitation? The scholar’s ability to ‘walk away’ after pro-
longed and intense connection is simply unnatural in terms of ordinary social
expectations. Even under the less intimate field conditions of some current
team-based documentation projects, the contrast between a relatively short
period of intense attention and interest from affluent and powerful outsiders
and the subsequent complete disappearance of the supposedly interested out-
siders must be a source of confusion and disappointment to the groups under-
going the experience.
Increasingly, in recent years, solutions have been consciously sought for
avoiding or mitigating this moral dilemma: for bridging the gulf between a
connection that can have an intensity, intimacy, and duration that is typical
of some degree of friendship, and a connection that from the point of view of
the researcher nevertheless requires at least some degree of the detachment of
scholarly observation. Fieldwork ethics have consequently become the subject
of much discussion and a considerable literature (Grinevald 2006). One way
of bridging the gulf, increasingly adopted today, is for the researcher to enlist
community members as co-researchers, offering training and/or co-authorship
to any community member who might wish it and co-equal researcher sta-
tus to those who accept the invitation (England 1992: 34; Grinevald 2003: 60).
12 For example, the younger daughter of the couple who served as my central oral-history
sources originally questioned my statement that fights had regularly broken out between
the young fishermen and their non-fishing peers at the site of a particular house that
marked the beginning of the fisherfolk residential section of her village; but as it hap-
pened that information came directly from her own father.
tain topics and the potential offensiveness of the material touched on here and
there in our sessions, and I was also deeply aware that in recordings made in
the privacy of their home the wife in particular had touched on family events
and personal feelings that she, a very private person, would not have wished to
have open to all comers.
This situation raised the most difficult question I had yet faced where con-
flicting responsibilities were concerned. On what basis can an outsider with
unrecoverable material of high scholarly interest withhold that material from
a birthright member of the ethnic group in question – in this case also a close
relative of the principal sources? The relative and I held very different views
of my responsibility. I remembered the distress of the widowed wife when she
had felt overly exposed by the publication of the two items she had not asked
to have removed in time, and I considered myself honor-bound to protect my
sources’ privacy. He believed that there was no longer any basis for such sensi-
tivities and that the important thing was to put all of these invaluable record-
ings into the historical record. There was validity to both points of view, but
while I had no difficulty seeing the force of his point of view, he had some
difficulty acknowledging the force of mine. The compromise solution I arrived
at was to consult a daughter who was still living in Scotland about the tapes,
which were mostly in English, and to send them to her so that she could make
the decision about public access to her parents’ recordings. She, too, found a
good deal of the material too personal for general access, but in the end she
selected one or two recordings with lower personal content and made them
available to her relative for wider historical use.
This experience made me still more aware than I had been already of the
growing historical and linguistic value of the recordings I possessed, as did an
inquiry from a Scottish scholar who wanted to listen to Sutherland’s east-coast
Gaelic in order to compare it with the Gaelic still spoken by a small number
of bilinguals on Sutherland’s north coast. Although at this point health prob-
lems had kept me from visiting Scotland for more than two decades, I was still
in touch with family members in most cases and was able to write to them
or speak to them by phone and ask permission to archive tape recordings I
had made with their parents, grandparents, aunts, or uncles, promising careful
editing in the cases where I knew the tapes contained material of some sen-
sitivity. Permission was granted in all cases, and I proceeded to edit and copy
some of my tapes and send them to two archiving bodies in Scotland. There
was still the possibility that as an outsider I might include some objection-
able material without realizing it, but the passage of time and the less socially
charged climate in present-day East Sutherland reduced the likelihood of seri-
ous offense.
The local mode of dealing with the severe social biases of earlier times
seems generally to be denial of the strength of the prejudice and mild roman-
ticization of the very lifeways that were once the focus of serious discrimina-
tion. This seems to have happened even in the case of the one social group that
had still lower standing than the fisherfolk: the tinkers, “travelling” people who
spent much of the year walking the roads, selling small, mostly second-hand
household items and acquiring others for sale. In recent years the migratory
routes that the tinkers took through Sutherland-shire glens have actually been
celebrated by reënactment (The Northern Times, May 18, 2007); by contrast,
the strong, earlier bias against the tinkers among local villagers appears very
clearly in the recollections provided by my oral-history sources (Dorian 1985:
96). Even though revised memories may have prefigured the direction local
attitudes would ultimately take, I was no more able to predict the rapidity of
this attitudinal change than I was able to detect exactly which matters would
be painfully sensitive to the principal figures in my oral-history book.
6 Conclusions
Perhaps the most instructive aspect of linguistic research with the East
Sutherland fisherfolk is that this fieldwork presents the same fundamental
ethical questions as linguistic fieldwork in much more unfamiliar-seeming
settings. By contrast with the Dupaningan Agta people discussed by Laura
Robinson (this issue), East Sutherlanders in their first-world setting seem ordi-
nary and culturally familiar.13 Working in the East Sutherland setting presents
the fieldworker with relatively few difficulties of a cultural or procedural sort.
Neither dramatically different lifeways, such as those of a hunter-gatherer
group, nor excruciating procedural difficulties such as those Colette Grinevald
faced in obtaining the letter-of-consent required by her funding agency for
work with the Rama people of Nicaragua (Grinevald 2006: 361–63), confront
the fieldworker in East Sutherland. Yet despite the seeming ordinariness and
familiarity of East Sutherlanders and their setting, the challenges of avoiding
exploitation or breach of confidence and of achieving informed consent are,
as I’ve tried to demonstrate, still very much the same – and just as difficult to
overcome.
13 This is actually somewhat less the case than it would seem; see Dorian 2010.
References
Bradley, David. 1989. The disappearance of the Ugong in Thailand. In: Dorian, Nancy C.
(Ed.), Investigating obsolescence: Studies in language contraction and death.
Cambridge University Press, Cambridge, pp. 33–40.
Collins, James. 1998. Our ideology and theirs. In: Schieffelin, Bambi B., Woolard,
Kathryn A., Kroskrity, Paul V., (Eds), Language ideologies: Practice and theory. Oxford
University Press, New York, pp. 256–70.
Dobrin, Lise. 2008. From linguistic elicitation to eliciting the linguist: Lessons in com-
munity empowerment from Melanesia. Language 84: 300–24.
Dorian, Nancy C. 1981. Language death: The life cycle of a Scottish Gaelic dialect.
University of Pennsylvania Press, Philadelphia.
———. 1985. The tyranny of tide: An oral history of the East Sutherland fisherfolk.
Karoma Publishers, Ann Arbor, MI.
———. 2010. Investigating variation: The effects of social organization and social setting.
Oxford University Press, New York.
England, Nora. 1992. Doing Mayan linguistics in Guatemala. Language 68: 29–35.
Foley, William A. 2003. Genre, register and language documentation in literate and
preliterate communities. In: Austin, Peter K. (Ed.), Language Documentation and
Description, Vol. 1. Hans Rausing Endangered Languages Project, London, pp. 85–98.
Golla, Victor. 2001. What does it mean for a language to survive? Some thoughts on the
(not-so-simple) future of small languages. In: Sakiyama, Osama, Endo, Fubito (Eds),
Lectures on endangered languages 2: From Kyoto Conference 2000. Endangered
Languages of the Pacific Rim, Kyoto, pp. 171–77.
Grinevald, Colette. 2001. Encounters at the brink: Linguistic fieldwork among speakers
of endangered languages. In: Sakiyama, Osamu, Endo, Fubito (Eds), Lectures on
endangered languages 2: From Kyoto Conference 2000. Endangered Languages of
the Pacific Rim, Kyoto, pp. 285–313.
———. 2003. Speakers and documentation of endangered languages. In: Austin, Peter K.
(Ed.), Language documentation and descripton, Vol 1. Hans Rausing Endangered
Languages Project, London, pp. 52–72.
———. 2006. A view from the field: An Amerindian view, worrying about ethics and
wondering about informed consent. In: Saxena, Anju, and Borin, Lars (Eds), Lesser
known languages in South Asia: Status and policies, case studies and applications of
information technology. Mouton de Gruyter, Berlin, pp. 339–70.
Haas, Mary R. 1941. Tunica: Introduction. Handbook of American Indian languages,
Vol. IV. J. J. Augustin, New York, 9–11.
Himmelmann, Nikolaus P. 2008. Reproduction and preservation of linguistic knowledge:
Linguistics’ response to language endangerment. Annual Review of Anthropology
37: 337–50.
Jocks, Christopher. 1998. Living words and cartoon translations: Longhouse “texts” and
the limitations of English. In: Grenoble, Lenore A., Whalen, Lindsay J. (Eds),
Endangered languages: Language loss and community response. Cambridge
University Press, Cambridge, pp. 217–33.
Kroskrity, Paul V. 2002. Language renewal and the technologies of literacy and postlit-
eracy: Reflections from Western Mono. In: Frawley, William, Hill, Kenneth C.,
Munro, Pamela (Eds), Making dictionaries: Preserving indigenous languages of the
Americas. University of California Press, Berkeley, pp. 171–92.
Ladefoged, Peter. 1992. Another view of endangered languages. Language 68: 809–11.
Matras, Yaron. 2005. Language contact, language endangerment, and the role of
the ‘salvation linguist’. In: Austin, Peter K. (Ed.), Language Documentation
and Description, Vol. 3. Hans Rausing Endangered Languages Project, London,
pp. 225–51.
Milroy, Lesley. 1987. Observing and analyzing natural language: A critical account of
sociolinguistic method. Basil Blackwell, Oxford.
Nagy, Naomi. 2000. What I didn’t know about working in an endangered language com-
munity: Some fieldwork issues. International Journal of the Sociology of Language
144: 143–60.
Newman, Paul. 2003. The endangered languages issue as a hopeless cause. In: Janse,
Mark, Tol, Sijmen (Eds), Language death and language maintenance: Theoretical,
practical and descriptive approaches. John Benjamins, Amsterdam/Philadelphia,
pp. 1–13.
The Northern Times, May 18, 2007. Summer walkers in the straths – a Mackay Country
experience. The Northern Times Company, Golspie (Sutherland).
Ó Dochartaigh, Cathair (Ed). 1997. Survey of the Gaelic dialects of Scotland. Dublin:
School of Celtic Studies.
Penfield, Susan D., Serrato, Angelina, Tucker, Benjamin V., Flores, Amelia, Harper,
Gilford, Hill, Johnny Jr., Vasquez, Nora. 2008. Community collaborations: Best prac-
tices for North American indigenous language documentation. International
Journal of the Sociology of Language 191: 187–202.
Rhydwen, Mari. 1998. Strategies for doing the impossible. In: Nicholas Ostler (Ed.),
Endangered languages: What role for the specialist? Proceedings of the Second FEL
Conference. Foundation for Endangered Languages, Bath, England, pp. 101–06.
Shipley, Bill. 2000. The Maidu story. Jorge Hankamer Webfest, http:ling.ucsc.edu/jorge/
index.html.
Tedlock, Barbara. 1992. The beautiful and the dangerous: Encounters with the Zuni
Indians. Viking, New York.
Terrill, Angela. 2002. Why make books for people who don’t read? A perspective on
documentation of an endangered language from Solomon Islands. International
Journal of the Sociology of Language 155/156: 205–19.
1 Introduction
Among those who are strongly supportive of efforts to revitalize receding lan-
guages, as well as to document them, it is well recognized that the barriers to
success are enormous. Fishman’s 1991 volume discusses many of those barriers,
and voluminous contributions to the literature on this topic have expanded
the discussion. In what follows here, I focus initially on potential problems in
the linguistic fieldwork setting and the language revitalization setting, arising
from different positions with regard to relatively private versus relatively pub-
lic language use, acquisition, and transmission. Efforts to lessen such problems
and the prospects for lessening them further are discussed thereafter, with par-
ticular focus on the increasing professional involvement in these efforts.
for some sort of written consent from the community in which the documen-
tation is undertaken, so that the project will also be as broadly consensual as
possible (Grinevald 2006). Serious questions still arise about the degree to
which individuals or communities can provide informed consent for proj-
ects that are essentially uninterpretable in terms of the local cultures (Dobrin
2008), but the enterprise is at any rate openly and often publicly undertaken.
Researchers are also increasingly prepared to share recordings, films, or some
printed outcome of this kind of work with the host community (Terrill 2002);
in some cases this is an objective of the project from the outset.
Prior to the late 1990s, multiple-researcher documentation projects were
probably still few in number by comparison with more traditional fieldwork
projects in which a lone researcher went “into the field” to make a record
of a language, usually with more descriptive than documentary objectives.
Videotaping was also relatively uncommon, undertaken more at the initia-
tive of the individual researcher than in response to professional expecta-
tions. Some climates and some field settings favored working out of doors, so
that the researcher’s activities were open to a great deal of public scrutiny. In
other climates and settings, however, the researcher disappeared within some
enclosed structure and conducted the research work in considerable privacy,
usually on a more or less dyadic basis. In open-air settings much that passed
between sources and researcher was audible to others and self-censorship on
the part of the local speaker(s) was probably automatic. Self-censorship was
less likely in the privacy of a home or a sequestered workroom, and in addition
the non-judgmental ear of someone not connected to the community by blood
or marriage could produce a freedom of expression neither party originally
anticipated, especially if the work continued over a longer period of time.
The very intimacy of such sequestered fieldwork encourages a trust that
over time reduces inhibitions. This was certainly the case in some of Milroy’s
work in Belfast, where she describes the work she did with one family in the
following terms: “Much of the conversation had focused on the disastrous
and pathetic effect upon the family of the civil unrest in Belfast and its func-
tion was plainly cathartic; many recordings resembled therapy sessions more
closely than sociolinguistic field tapes” (Milroy 1987: 90).
My own experience in fieldwork done prior to the 1990s was that long-term,
sequestered fieldwork produced personal and at times very uninhibited con-
tent some of which was quite unsuitable for general-access archiving; it would
have been equally unsuitable as the basis for printed materials that might pro-
mote revitalization, such as story books and autobiographical sketches. The
more spontaneous and lively a speaker’s recordings were, the more unsuitable
they were likely to be for such purposes. In effect they represented a sharing
Yet stopping the session to re-negotiate permission to record in each such case
would have been fatal to any degree of naturalness (Milroy 1987: 89).
Because of the elusiveness of a truly informed consent from people to whom
the methods, purposes, and products of research are bound to remain to some
extent obscure, the researcher’s best efforts may not be enough to ensure ade-
quate consent, as one of my own fieldwork experiences can illustrate. In 1976
I embarked on an oral history project in one of the three fishing communities
where I had been doing fieldwork intermittently since 1963. I conceived of this
as a way of making a record of what was by then a way of life that lingered only
in the memories of the elderly and also as a way of making some return to the
community for their unstinting generosity in sharing their language with me
over the years. I approached a husband and wife in their seventies with whom
I had already done linguistic work and asked them whether they would be will-
ing to act as the central sources for an oral history; they were ideal for the role
because each of them had had experience of work connected both with the
local line fishing and with the national herring fishery. I gave them time to
discuss the project between themselves, and when they agreed to participate
I wrote an informal agreement into one of my field notebooks, stating that
they would allow me to use the material they recorded while I would be care-
ful to protect identities; this we all signed. We proceeded with this work during
the summers of 1976 and 1978, and when I had a manuscript ready I posted it
to them, so as to give them an opportunity to have anything they objected to
removed. This was to be their story, after all, and I wanted it to reflect their lives
in a way that felt both accurate and acceptable to them. I was also aware that as
an outsider to the community I might not, even after what was by then 16 years,
be able to identify exactly what would or would not be objectionable to local
sensibilities. Sadly, the husband died while the manuscript was in the mail; but
at his widow’s request I removed two items of somewhat personal information
from the text. In spite of all my precautions, however, when the book came out
the widow was distressed by two other items she had overlooked at first read-
ing. She now wanted these removed as well, which of course was not possible
at that point, and it took the kindly intervention of two of her children to per-
suade her that she was not, as she feared, overly publicly exposed.
The flaw in my earnest efforts to make the finished oral history completely
inoffensive was that the publication process itself was unfamiliar to my
sources: it was not obvious to the surviving spouse, as I had supposed it would
be, that the manuscript phase of the book was the only stage at which any-
thing she objected to could be taken out. This was a woman literate in English
whose lifestyle in a familiar-seeming, first-world environment was not at that
point exotic or unusual in any obvious way, yet her genuinely informed con-
sent turned out to be much more difficult to achieve than I had imagined. It
is easy to see how much harder it would be to achieve well-informed consent
in a cultural context that differed more radically from the researcher’s own.
Clearly a good deal of responsibility falls to the researcher when it comes to
the protection of privacy.
language has been used mainly in home and neighborhood settings, transfer-
ring it from intimate settings into a formal and public setting like the school can
produce an inhibiting sense of inappropriateness, for teachers and students
alike. If the schools have a long history of excluding local minority languages,
children from minority language homes may be particularly uncomfortable
about encountering their home language in the classroom and being asked
to use it in that environment. Indigenous community members recruited as
language teachers might be expected to have no difficulty in using their lan-
guages in the school setting, yet school use of the local indigenous language
may be problematic for them as well. In Tlaxcala, Mexico, eight individuals
who were candidates for positions as indigenous language teachers in a revi-
talization initiative were observed during parts of a teacher-training program
mounted between August and December of 1999. Although the candidates
were all speakers of Mexicano (otherwise known as Nahuatl, an indigenous
Uto-Aztecan language), the observer reported that such use as they made of
Mexicano during their course participation was symbolic, except for the one
occasion when they were being tested on their indigenous language skills
(Messing 2003: 82). Despite the avowed revitalization purpose of the training
course, the classroom was “a formal context, . . . without sufficient intimacy
and solidarity between speakers to warrant more use of Mexicano” (ibid.).
It should be noted, however, that the reverse of this development is also
known to happen: minority-language speakers who become teachers of their
indigenous language may adapt well to school use of the language while fail-
ing to use it at home with their own children, promoting public sphere use but
neglecting private-sphere use (Hinton 2009). Parent-child transmission, usu-
ally the swiftest and most complete route to mastery of the target language,
is then replaced by purely school-based transmission, which Fishman in his
benchmark study of reversing language shift convincingly depicts as ineffec-
tive (Fishman 1991: 368–70). Schooling is very often the chosen locus of revi-
talization efforts, all the same, precisely because it moves revitalization from
the less accessible realm of personal motivation into the public and potentially
more maneuverable realm of educational policy-making.
Once the classroom becomes the setting for revitalization efforts, the
requirements of formal teaching present certain foreseeable problems for
speakers of private-sphere forms of the minority language. In the interests
of promoting literacy and broader use, coinage and codification make their
appearance, and their adoption changes the school-promoted version of the
language into one that differs from the variety spoken locally. Coinage of new
terms is needed so that speakers can deal with topics that are either seldom
discussed in the minority language or are discussed by drawing on many loan-
1 In the matched guise technique, listeners hear one or more speakers gifted at variable ren-
derings of certain key linguistic features read a text that differs only in the way those features
are produced. For any given speaker, rate of delivery, tone of voice, and so forth are kept con-
stant, and listeners are not told that the speaker is the same in more than one case. Listeners
are asked to evaluate the speaker of each version of the performed text in terms of various
subjective responses. In this case the speaker’s rural or urban origin, general age group, gen-
eral place of residence within Wales, and likelihood of having received Welsh-medium edu-
cation were evaluated; respondents also assessed the likelihood that the speaker might hold
certain specified jobs.
language beyond the classroom, into some level of more general social use,
and also on the degree to which native speakers prove willing to accept such
second-language speakers into their conversational networks.
Neither development is a reliable outcome of revitalization undertakings,
but hope of such an outcome motivates many language-support movements,
and the demonstrated attainability of more modest goals (younger people
with at least some active knowledge of the heritage language and some famil-
iarity with its lexicon and structure, as in the case of Tolowa individuals who
have passed through the Tolowa language program in northern California;
see Collins 1998a: 264) fuels continuing community support for school-based
programs.
mode and subsequently also with their language, likewise expressed no desire
for the revitalization of their local speech form (Dorian 1987).
Matras takes strong exception to “salvation linguists”, as he calls them, who
see themselves as stepping in to “save” an endangered language. He points out
that communities do not take a single attitude or speak with a single voice,
and that not every community wants its language rescued by a linguist (Matras
2005: 227). Yet it is not uncommon today for speakers of a receding language to
take the initiative themselves, asking to have a linguist study and record their
language (Nagy 2000; Grinevald 2005) or hiring a linguist to help in creating
or refining a writing system, putting together a dictionary, or preparing teach-
ing materials, as do some indigenous Australian and North American groups
(Wilkins 1992; Debenport 2009).
Hinton describes vividly the yearning for recovery of ancestral languages
that brought leaders of the Native California Network together with linguists
in 1992 to find effective ways to preserve or restore their languages (Hinton
1994: 221–22). The success of the Indonesian workshops mounted by Florey
and her colleagues likewise indicates that local speakers wish both for outside
expertise and for greater expertise of their own in furthering documentation
and revitalization.
Dixon, in his 1991 paper, pointed out that communities often remain
unaware of the risk to their language until it has grown too late to change the
situation (Dixon 1991: 231). This seems an odd observation initially, but it is
not so counter-intuitive as it seems. Fluent speakers who have reached middle
age still have available a generation older than themselves who regularly speak
their language. They themselves make ample use of the language, and they
may simply fail to register the degree to which young people and children are
using some more widely spoken language instead. As Kulick demonstrated
for the Papua New Guinean village where he worked, they may also not reg-
ister the degree to which they themselves are failing to use the local language
with their children (Kulick 1992). It is only as they become the older generation
themselves that some local-language speakers look around and realize that
there are no speakers coming along behind them: they are the last remaining
speakers, and unless heroic measures are taken their language will disappear
with them. At this point the attention of specialist outsiders may rather sud-
denly be seen as useful where it was not before.
Grinevald (2006) has pointed out that interventionist agendas developed
among linguists in very particular contexts, namely those in which patterns
of language loss had already reached extreme proportions: North and South
America and Australia. While in Europe some regional languages were acknowl-
edged to be used less than others – hence the terminology EBLUL, European
Bureau of Lesser Used Languages, for the body that represents them – and in
South Asia and Africa some languages were recognized as spoken only by cer-
tain tribal groups in a context of widespread multilingualism, in the Americas
and Australia the context was massive indigenous language loss, already far
advanced and in prospect even more severe. In America and Australia, where
dwindling numbers of indigenous-language speakers confronted this prospect,
indigenous communities were voicing acute concerns about their languages
and beginning to agitate for revitalization; the Americanist and Australianist
linguists with whom these matters were raised were among the first to express
a sense of professional obligation to intervene on behalf of small and reced-
ing languages (Grinevald 2006: 340–41). It was no accident that the linguists
and activists who contributed to the 1992 issue of the journal Language that
famously raised the issue of language endangerment within the U.S. profes-
sional community all worked with American or Australian languages (Hale,
Krauss, Watahomigie, Yamamoto, Masayesva Jeanne, and England 1992).
Linguists have long had their own purely professional reasons for wishing
to document and describe receding languages (more typically to describe than
to document, although Himmelmann 2008 argues powerfully for the greater
value of documentation). Matras articulates the chief such reason at the end of
the article cited above when he speaks of “the urgent task of securing a diverse
linguistic sample corpus for the sake of future generations of students of lan-
guage” (Matras 2005: 248). But Woodbury represents a growing voice in the
profession when he writes that “it is becoming less and less viable for linguists
to think of the stakeholders in language documentation to be constituted only
of a vaguely-conceived scientific posterity” (Woodbury 2003: 39), and the same
may be said where language description and analysis are concerned.
The unhappy fact is that whatever linguistic professionals do, it will be inad-
equate. The complexity and richness of language and its cultural context are
such that they escape all our efforts to capture them. Our descriptions will
prove to be more incomplete, our analyses more imperfect, and our documen-
tations more limited than we imagine. Twenty years from now a new genera-
tion of linguistic professionals will wonder how we could have failed to raise
the questions that interest them most or document the kinds of linguistic
behavior that have come to preoccupy them. Whether we scatter ourselves
across the globe in what promises to be a last-minute attempt to record more
receding languages, or embed ourselves deeply in particular small-language
contexts in an attempt to record and understand them more fully, much will
elude us. But that is no reason for not trying. The forces arrayed against the
survival of small languages are formidable, and our efforts to provide support
for them are likely to prove inadequate. But that is also no reason for not trying.
References
Flores Farfán, José Antonio. 2001. Culture and language revitalization, maintenance,
and development in Mexico: The Nahua Alto Balsas communities. International
Journal of the Sociology of Language 152: 185–197.
Florey, Margaret. 2008. Language activism and the “new linguistics”: Expanding oppor-
tunities for documenting endangered languages in Indonesia. In Language
Documentation and Description 5, Peter K. Austin (ed.), 120–135. London: Hans
Rausing Endangered Languages Project.
Grinevald, Colette. 2001. Encounters at the brink: Linguistic fieldwork among speakers
of endangered languages. In Lectures on endangered languages 2, Osamu Sakiyama
and Fubito Endo (eds), 285–313. Kyoto: Endangered Languages of the Pacific Rim.
———. 2005. Why the Tiger Language and not Rama Cay Creole? Language revitaliza-
tion made harder. In Language Documentation and Description 3, Peter K. Austin
(ed.), 196–24. London: Hans Rausing Endangered Languages Project.
———. 2006. Worrying about ethics and wondering about “informed consent”:
Fieldwork from an Americanist perspective. In Lesser-known languages of South
Asia, Anju Saxena and Lars Borin (eds), 339–370. Berlin: Mouton de Gruyter.
Hale, Ken, Krauss, Michael, Watahomigie, Lucille J., Yamamoto, Akira Y., Craig, Colette,
Masayesva Jeanne, LaVerne, and England, Nora C. 1992. Endangered languages.
Language 68(1): 1–42.
Himmelmann, Nikolaus. 2008. Reproduction and preservation of Linguistic knowledge:
Linguistics’ response to language endangerment. Annual Review of Anthropology
37: 337–350.
Hinton, Leanne. 1994. Flutes of fire: Essays on California Indian languages. Berkeley:
Heyday Books.
———. 2009. Language revitalization at home. Paper delivered at the First International
Conference on Language Documentation and Conservation, Manoa, Hawaii, March 14,
2009. Accessible at http:/scholarspace.manoa.hawaii.edu/handle/10125/5961.
Jones, Mari C. 1998. Language obsolescence and revitalization: Linguistic change in two
sociolinguistically contrasting Welsh communities. Oxford: Oxford University Press.
King, Kendall A. 2001. Language revitalization processes and prospects: Quichua in the
Ecuadorian Andes. Clevedon: Multilingual Matters.
Kulick, Don. 1992. Language shift and cultural reproduction: Socialization, self, and syn-
cretism in a Papua New Guinean village. Cambridge: Cambridge University Press.
Ladefoged, Peter. 1992. Discussion note. Another view of endangered languages.
Language 68: 809–811.
MacCaluim, Alasdair. 2007. Reversing language shift: The social identity and role of adult
learners of Scottish Gaelic. Belfast: Cló Ollscoil na Banríona.
Matras, Yaron. 2005. Language contact, language endangerment, and the role of the
“salvation linguist”. In Language Documentation and Description 3, Peter K. Austin
(ed.), 225–251. London: Hans Rausing Endangered Languages Project.
* Both the Author Index and the General Index were assembled with the invaluable help of four
student interns from the Tri-College Linguistics Department: Madeleine Booth, Swarthmore
College; Zhiyin Ding, Bryn Mawr College; Gregory Nisbet, Haverford College; and Micah
Walter, Haverford College. I’m grateful to the students for their interest and energy and to
the Tri-College Linguistics Department for sponsoring their internship, which entailed a trip
from Philadelphia to Maine.
Denison, Norman 174, 176, 207, 220, 377, 388 Haiman, John 293, 306
De Vos, George 193, 200, 231, 233 Haas, Mary R. 1, 12, 27, 137–138, 143, 145, 372,
Dixon, R. M. W. 20, 26, 373, 388, 436, 439, 441 375, 389, 409, 422
Dobrin, Lise 417n8, 422, 426, 436, 441 Hale, Ken 20, 27, 440, 442
Drapeau, Lynn 261–262 Hamp, Eric P. 92, 209, 221, 258, 260–262,
Dressler, Wolfgang 8, 26, 67, 90, 92, 311, 326, 392
328 Hansen, Magnus Pharao 234n1, 244–245
Eades, Diana 16, 26 Harper, Gilford 417n5, 423
Edwards, John 19, 26, 288–289, 289n1, 306 Haugen, Einar 259, 262
Ellis, Peter Berresford 210, 221, 233, 256, Heath, Shirley Brice 175–176, 265, 268n1,
262 278n12, 282
England, Nora C. 27, 440, 442 Henry, Reginald 372, 389
Evans, Nicholas 1, 16, 26, 295, 306 Heyd, Uriel 255, 262
Hill, Johnny Jr. 417n5, 423
Fennell, Desmond 228, 233 Hill, Jane H. 12–13, 27, 67, 92, 143, 144n7, 145,
Fenyvesi, Anna 115, 134 149, 155, 168, 254, 262, 372, 379, 381, 389
Fillmore, Lily Wong 114, 162, 165, 218, 221 Hill, Kenneth C. 67, 92, 254, 262, 372, 379,
Fishman, Joshua A. 65, 147, 158, 165, 194, 201, 381, 389
210, 221, 235, 244–245, 249n1, 251, 255, 258, Himmelmann, Nikolaus P. 20–21, 27, 411,
261–262, 280–281, 425, 430, 441 417n11, 422, 425, 440, 442
Flinn, Juliana 276–277, 281, 286, 306 Hindley, Reg 253, 262
Flores, Amelia 417n5, 423 Hinojosa, Maria de la Paz 174–175, 177, 208,
Flores Farfán, José Antonio 14, 27, 434, 442 220–221
Florey, Margaret 437, 439, 442 Hinton, Leanne 241, 244–245, 430, 439, 442
Foley, William A. 23, 27, 420, 422 Hohenthal, W. D. 193, 201, 208, 221, 276, 282
Foot, Hugh C. 193, 200 Holmer, Nils 39n9, 65
Householder, Fred W. 92
Gaeliconline 242, 245 Huffines, Marion Lois 174, 177, 258, 261–262,
Gahng, Tae-Joong 273, 283 287, 306, 377, 389
Gal, Susan 19, 27, 174, 176, 208, 221, 377, Huss, Leena 302, 302n3, 306
379–381, 388 Hymes, Dell H. 64–66, 90, 92, 160, 162, 165,
Gambhir, Surendra 376, 388 193n1, 201, 208–209, 221, 311, 328
Gardner, Robert C. 212, 218, 221
Gardner-Chloros, Penelope 345–346 Ibrahim, Muhammad 247, 262
Gauchat, Louis 63, 65 Irvine, Judith T, 19, 27
Gellner, Ernest 266–267, 282
Georges, Robert A. 384, 388 Jackson, Jean 193, 201
Giles, Howard 193, 200–201 Jocks, Christopher 409, 423
Glazer, Nathan 201 Jones, Mari C. 133–134, 178, 191, 431, 442
Goffman, Erving 397, 408 Jones, Michael O. 384, 388
Goddard, Ives 24, 27 Joseph, John Earl 268–269, 282
Golla, Victor 417n9, 422
Greene, David 65, 174, 176, 229, 233 Karttunen, Frances 208, 221
Grillo, Ralph D. 266–268, 282, 288, 306 Keane, M. 233
Grinevald, Colette 17, 24, 27, 410, 416, Khleif, Bud B. 282
421–422, 426, 437, 439–440, 442 King, Kendall A. 28, 231, 245, 431, 434, 442
Gumperz, John J. 156, 158, 165, 345–346 Kiparsky, Paul 38n7
Gunn, A. 197, 199–201 Kloss, Heinz 169, 177
Krauss, Michael E. 20, 66, 92, 137, 143–145, Menn, Lise 176, 311, 328
149, 155 Mertz, Elizabeth 147, 155, 174, 177
Kroskrity, Paul V. 17, 19, 27, 178, 191, 275, Melchers, Gunnel 379, 389
270n3, 277, 280, 282, 296, 298–300, 303, Messing, Jacqueline H. E. 430, 443
306, 409, 423 Miller, Elmer 22, 28
Kulick, Don 287, 439, 442 Miller, Wick R. 66, 92, 149, 155, 376, 383, 389
Kuryɫowicz, Jerzy 82, 92 Milroy, James (Jim) 18, 28
Kushner, Gilbert 223, 233 Milroy, Lesley 163, 395, 408, 415–416, 423,
Kuter, Lois 20, 28, 226, 233, 235, 245, 257, 426–428, 443
262, 266, 282 Mithun, Marianne 349, 367, 372, 389
Miyashita, Mizuki 14, 24, 26
Labov, William 63, 65, 101, 114, 144–145, 158, Mohan, Peggy 385, 389
163, 166, 311, 328 Moore, Robert E. 20, 28
Ladefoged, Peter 22, 26, 28, 417, 423, 437, Mougeon, Raymond 115, 133–134
442 Muehlmann, Shaylih 20, 28
Lambert, Wallace E. 91–92, 218–219, 221, Mühlhäusler, Peter 67, 92
302, 307 Muntzel, Martha 16, 26
Landry, Rodrigue 302, 305 Musgrave, Simon 425, 436, 443
Langgaard, Per 274, 282
Lee, Jennifer 249–251, 262 Nagy, Naomi 299, 307, 410, 423, 439, 443
Lehmann, Winfred 35, 65 Nathan, David 425, 441
Leonard, Wesley Y. 16, 28 New York Times 266, 282
Lewis, (E.) Glyn 29, 146, 155, 210, 221, 291 Newman, Paul 4, 28, 417n10, 423, 437, 443
Leyew, Zelealam 349, 367 The Northern Times 421, 423
Long, Michael H. 14, 28
Ó Baoill, Dónall P. 253, 260, 263, 435, 443
mac a’ Ghobhainn, Seumas 210, 221, 223, 233 Obler, Loraine K. 176, 311, 328
Macaulay, R. K. S. 336n5, 346 O’Callahan, Joseph 256–257, 263
MacAulay, Donald 366–367 Ó Cinneide, M. S. 229, 233
MacCaluim, Alasdair 433, 435, 442 Ó Ciosáin, Séamus 223n1, 227, 233
MacEachern, Scott 292–293, 307 Ó Cuív, Brian 239, 245
MacLaren, James 41n14, 65 Ó Dochartaigh, Cathair 2, 28, 402, 408–409,
MacKay, J. 201 423
Mackey, William F. 146, 155 Oftedal, Magne 39n9, 41n14, 65, 353, 368,
Maclean, Sorley 366–367 401, 408
Macnamara, John 210, 221, 227–228, 233 Ó hIfearnáin, Tadhg 243, 245
Maguire, Gabrielle 244–245, 259–260, 262 Ó Riagáin, Pádraig 238–239, 243, 246, 263
Malinowski, Bronislaw 397, 408 Ó Tuathaigh, Gearóid 239, 246
Margolin, David 224–225, 233
Masayesva Jeanne, LaVerne 27, 440, 442 Paulston, Christina Bratt 264n, 290, 307
Matras, Yaron 240, 245, 417n10, 423, Peacock, John Hunt 14, 29
437–440, 442 Peal, Elisabeth 302n3, 307
McCorkle, Thomas 193, 201, 208, 221, 282 Pecos, Regis 434, 443
McEwan-Fujita, Emily 14, 28 Penfield, Susan D. 417n5, 423
McGill, Stuart 3, 25 Pensalfini, Rob 400n1, 408
Mc Laughlin, Fiona 294–295, 307 Perley, Bernard C. 13–14, 29
McLendon, Sally 17, 28, 209, 221 Petersen, William 193, 201
Meillet, Antoine 273n8, 282 Polinsky, Maria 11, 29
abandonment 5, 155, 195, 264, 266, 272, age group 36, 44, 46, 48, 63, 68n2, 94,
300–301, 329 96, 178, 189, 213, 217, 219, 230, 299, 373,
Abbé Gregoire 267 401, 431n1
Aborigine, Aboriginals (see also language) age-based proficiency
16, 26, 250, 329 groupings 186–190
Aboriginal Australia 15, 286 age-related shift pattern 300
access 17, 22, 153, 235, 286–87, 290, 305, 397, age-related variation 403
404, 413–414, 417, 419, 426–427, 429, 432 age and speaker skills 178–192
accuracy 22, 96, 143, 373, 409 age mate 119, 230, 299
accessibility 231, 371, 425 agriculture 211, 273, 407
acquisition (dialect/language) 1, 9–10, 13–15, agricultural population 33, 36n2
20, 107, 149n2, 153, 170, 212, 235, 239, 288, agriculturalist 33, 195, 351, 396
292, 296, 300–301, 303, 376, 386, 425, 432, Akkad, Akkadian 271
433–35 Alabama 227, 277n10, 280
adult 14 Alabama Koasati 277n10
child 14, 372 Albania, Albanian 168, 209, 258, 260,
foreign-language 206, 212 374–375
second-language 212, 218, 433 alienation 247, 434
acquisition(al) history 10, 107, 213, 213n4, allomorph 82, 82n5, 83, 90, 219, 319, 353,
214, 219 400
activist, activism 235, 237–238, 241–242, 417, Alsatian, Alsatians 267, 288, 345
439–440 alternation (linguistic; see also vowel) 34,
adjective 39, 70, 98, 115, 140, 259, 269n2, 143
319n12, 355, 363 Amazonia 180, 185
possessive 217 Amazon Basin 284, 293
adjective comparison 188 America (see also Indian) 208, 301, 439, 440
adverb (see also particle, phrase) 23, 39, 49, North America 205, 208, 264
98, 118, 118n3, 118n5, 132, 315n4, 347, 403 South America 16, 264, 439
directional 117, 118n3, 119–121, 132, the Americas 432, 440
188–189 Amharic 349
locational 117, 119–121, 123, 129, 132, Amish 148, 244, 258
188–189 analog 73, 255, 406
adverbial marker 346, 355 analogy (see also change, leveling,
Ælfric 261 regularization) 38, 43–44
affiliation 146, 193, 292 analogical plural formation, analogically
Africa 210, 269, 286, 291–292, 294, 440 formed plurals 85, 214
age (as a variable; see also variation) analysis (as language structure; see also
age-and-proficiency continuum, grammar) 219
age-based proficiency continuum 6, analytic(al) construction 140–141
9, 16, 116, 118n5, 119, 122–123, 124–131 analytic structure 143, 190
* Boldface entries represent chapter titles and chapter section or subsection headings that
feature the subject of the index entry.
311, 327, 329–346, 349–350, 380, 384–385, ceilidh (lexical item) 52, 55;
387, 402, 419 (institution) 163, 313
(fully) fluent 39, 64, 159, 239 Celtic 197, 217, 316n5, 327, 329, 342, 359,
(near-)passive 15, 90, 107n6, 154, 157–160, 361–362, 394–395, 402
162, 164, 170 census 117n2, 242, 299, 331n2, 379, 393
Biloxi 143, 372 Central Asia 235
birth order 10, 149n2 ceremony, ceremonials 168, 194, 208, 276
Boeotia 10 Chadic 291
Bolivia 271, 437 change (see also analogy, economics,
Bòrd na Gàidhlig (the Gaelic Board) 242 mutation, style, vowel)
borrowing 38, 109, 180, 250, 254–255, 261, analogical 82, 82n5
296, 318, 320, 348 attitudinal 421
negative borrowing 115–133 demographic 438
Brazil 194, 208, 298 grammatical 33–65, 67, 138, 187,
Brazilian Amazon 15, 400 399–401
Breton(s) 8–9, 67, 90, 147, 167, 226, 235, 257, internally motivated 115
266–267, 273n8, 288, 372, 377 morphological 73
Britain, Great Britain 4, 169, 175, 207, 215, morphophonemic 139–140
236, 269, 290, 384, 394 mutational 35–36, 40, 51, 61–62
British (language) 169, 238, 242 ongoing 7, 34, 39, 170, 399
broadcast, broadcasting, broadcaster 16, phonological 63, 399
168, 232, 239, 248, 255, 284, 313, 326, 365, quantity (phonology) 71–74, 77, 84, 88,
375 90
Brora 1, 2, 5–7, 9–10, 24, 33–34, 37n3, 37n5, social 438
38n8, 41n13, 42n15, 51n25, 52–53, 64, 68, 88, change in progress 5, 8, 35, 37, 39, 58, 63,
139, 142, 151, 187, 194–195, 197, 217, 316n5, 116–117, 123–124, 142, 186–187, 402
327, 329, 342, 359, 361–362, 394–395, 402 change process, process of change 13,
Bugotu 235 115, 133, 144, 190
Bulgar 265 direction of change 21, 63, 190, 374
Bureau of Indian Affairs 231 rate of change 63–64
Burma 293 Charmey 63
Cherokee 280, 373, 377
California 16, 137, 149, 168, 236, 410, 432, 434, Echota Cherokee 280
439 Chiapanec 16
Master-Apprentice Program 240, 244 Chicago, University of 392
calque 181–183, 226 childhood 10, 34, 107, 119, 131, 149–152, 178,
Cameroon 292 195, 217, 392, 402, 417–418, 420
Canada 167, 169, 208, 210, 237, 284, 412n2 child, children (see learner, speaker) 1–2,
Cape Breton 147, 167, 177 4–5, 9, 13, 38, 94, 112n8, 117, 138, 138n2, 139,
Carnegie, Andrew 227 149n2, 150, 152–153, 159, 170–171, 173–175,
the Carolines (Western Island group, 178, 180–182, 189n5, 195, 207, 209, 229–232,
Micronesia) 276 237, 240, 250–252, 257–258, 258n4,
case (grammar; see also dative, genitive, 259–260, 264, 266, 271, 273, 277n10, 287,
nominative, vocative) 291, 294, 297–299, 301–305, 323, 327, 329,
case distinction 112, 400 331, 337, 338n6, 376, 386, 302, 411, 414, 418,
case system 39, 49–61, 64 420, 428, 430–431, 433–435, 439
caste 175, 290 Chile 271
category: see grammar, morphology Chimbu 293
Cayuga 372 Chinese 210, 432
expert, expertise 13, 21–22, 182, 211, 244, 281, herring fishing/fishery 36n1, 317, 334
378, 388, 398, 437, 439 line fishing 402, 418, 428
expert knowledge 179, 297, 406, 417 fluency 10, 19, 96, 124, 137–138, 157, 162,
exposure (to language) 6, 197, 154, 170, 213, 171–172, 174, 179, 186, 186n1, 217, 236–237,
261, 298, 305, 383 234, 296, 409
expressivity, expressiveness 179, 272 focus marker, focus marking 352–367
expressive bleaching 350–352 formality (see also context) 168, 247, 325,
expressive capacity 270, 349–350 327
expressive power 362–367 formulaic expressions 162, 162n6
extinction (see also language) 33, 35–36, 82, fossil, fossilization (linguistic) 40n10, 93, 98
89, 91, 94, 139, 144, 144n7, 154n5, 167, 207, fossil(ized) expression, form, phrase 41,
209, 211, 213, 220, 270n4, 311–312, 326, 331, 51n24, 248
372, 377, 400, 402 Franco (Francisco) 273, 366
Eyak 66 Franco-Provençal 273
Franks 265
factions, factionalism 24, 256–257 Frederick the Great 210
factors (see grandmother) French 89n9, 169, 210, 237, 260–261,
demographic 302 267–268, 269n2, 270, 270n5, 279, 284, 288,
economic, socioeconomic 175, 299, 302 293–294, 303, 315n4, 345, 378, 432
motivating 277, 366 Cajun 167
Faetar, Faeto 299 Norman 261
family 167–177 Ontario 115
language 143, 298 Pidgin 89n9
nuclear 4, 149, 151, 154 Standard 89n9, 273
Faroese 224–225, 379 French Canadian 167
fate 20, 66–92, 104, 209, 223, 405 frequency (of occurrence, linguistic) 6,
Fatick 294 15–16, 39, 58–61, 63, 70, 73–74, 77–78,
feature see dialect, discourse, grammar, 85–86, 90, 97n1, 98, 101, 106–108, 111, 118,
morphology, salience 128–129, 132–133, 151, 157, 169, 171, 319, 352,
fieldwork, fieldworkers (see also 354, 366, 377
anthropology) 1–6, 8, 23–25, 33, 46n18, friendship 378, 383–384, 406, 415–416
137–138, 142, 144, 159n2, 175, 186–187, friendship networks 148, 159, 384
188n4, 240, 290, 294–295, 314–315, 371, Friulian 207, 377
373, 375–376, 383–384, 386–387, 391–393, Fulnio 194, 208
395–398, 404–407, 409–410, 415, 416–417, function (language) 6, 12, 91, 113, 146, 194,
418, 420–421, 425–429, 435–437 205, 209, 238, 312–314, 326, 353, 375, 433
late-stage 499, 410–414 functional restriction 167
long-term 25, 391, 406 functional segregation 169
team, team-based 21, 427 funding 22, 242, 392–393, 396, 410n1, 421,
telephone 23 415, 433, 435
Finnish, Finland 210, 238 future (tense) 140, 142–143, 172–173, 333, 336,
First Mesa 278, 301 346, 355, 358, 361
fishing (see also industry) 5, 12, 33, 116, 148,
150, 407, 411, 418, 420, 435 Gael 433
fishing community 3, 6, 12, 151, 197, 199, Gaelic (see also Irish)
313, 393–394, 402, 404, 428 Bible, Biblical 36, 95, 117, 313, 324, 335,
fishing village 1, 4–6, 11–12, 58n33, 116, 377
152, 156n1, 172, 241, 335 Common 2
standing (see also social standing) 168, style (see also factors, speech) 23, 93, 247,
237–238, 242, 264, 272, 274, 284, 386, 421, 251–252, 314–316, 324, 326–327, 329, 332,
436 347, 367, 380
Statistical Account (of Scotland, of careful 321
Sutherlandshire) 197, 199, 207 casual 311, 323–324, 326
status (see also speaker) 15, 61, 119, 149, formal 168, 251, 325
153n4, 170, 174, 195, 223, 229, 237, 240–241, narrative 329
249n1, 267, 269, 271, 278, 284, 290, 302, 332, personal 96
337, 376, 380, 385, 416 style-shifting 382
official (language) 271, 284, 292 stylistic consistency 275
social 269, 287, 327 stylistic option 67, 140, 143
socioeconomic 153n4 stylistic range 314, 316, 326–27
status distinctions 164 stylistic variation 24, 311–328, 380
status factors 302 subjunctive 347–350, 355
status sets 278 subordination (language structure; see also
stereotypes, stereotyping 12, 144, 149, 164, clause) 217
232, 364 subordination (social) 260, 284, 288, 406
stereotyped feature 364 subordinate group/people/population
stigma, stigmatization 116, 149, 236, 241, 169, 265–266, 268–269, 285, 288, 290,
403, 438 300–301
stigmatized groups 5, 164, 215, 218, 230, subordinate language 93, 266, 268–269,
240, 345 278n12
stimulus sentences (see also test sentence) subsistence, sub-subsistence (economic
69, 71, 104, 111–112 mode) 195, 272, 297, 304, 351, 396,
story (see also tradition) 6, 23, 50, 58n33, 438–439
180, 183, 190, 230, 236, 251, 289, 314n2, 316, substitution (of one linguistic element for
318–326, 329, 332–333, 336n5, 339, 343, 350, another) 39, 62, 71, 97, 106, 109–110, 112,
359, 362–364, 382, 399, 420, 427–428 140, 143, 399
story collection 22 success (see also economics) 16, 23, 107,
storybook 417, 421, 426, 436 111, 113, 162, 174, 210, 217–218, 223, 229, 231,
storyteller, storytelling 181–183, 185, 190, 237–238, 244, 247, 255–256, 272, 274–275,
329–230, 330n1, 367, 379, 381 303–305, 373, 383, 385, 425, 435, 439
stratification 18, 268 suffix (see also diminutive, morphology,
structure (see also analysis, deviation, plural) 271n5, 320n13, 349–351, 400, 408
grammar, integrity, interference, lexicon, conditional 355
morphology, noun, passive, phonology, deictic 351, 353
society, surface structure, syntax) emphatic 269n2, 352–367
conservative 6, 19, 96, 101, 132, 188–189 emphatic-deictic 353–354, 359, 363–365,
embedded 171, 173 367
high-frequency 67 idiosyncratic 78
language, linguistic 1, 23, 163, 183, 271 subjunctive 355
matched 116, 123–124, 129–131 3rd-person 348
non-traditional 127 vocative 118n4
socioeconomic 194, 209 suffixation 71–73, 75, 77–86, 86n7, 88–90,
synthetic 143, 219 98–99, 102, 246, 404, 418
traditional 123–124, 127, 129, 131 simple 75, 80, 82–83, 86, 88–90
troublesome 96–105 Sumer, Sumerian 271
unmatched 116, 117–123, 124, 129–132 suppletion 41, 72, 74–75, 78–79, 86, 86n7,
verb(al) 40, 249 87–88, 324
verb complement 100–101, 106 suppletive past-tense forms 124, 568