SOCIOLINGUISTIC

PATTERNS

William Labov
(Conduct and Communication No. 4)
 Digitized by the Internet Archive
in 2012

http://archive.org/details/sociolinguisticpOOwill
 fi

Sociolinguistic Patterns

*»*
Sociolinguistic
Patterns

WILLIAM LABOV

University of Pennsylvania Press

Philadelphia
Copyright © 1972 by the University of Pennsylvania Press, Inc.
All rights reserved
Second printing 1974
Third printing 1975
Fourth printing 1976
Fifth printing 1977
Sixth printing 1978
Seventh printing 1978
•The Social Motivation of a Sound Change" © 1963 by Word; "The Reflection
of Social Processes in Linguistic Structures" ©
1968 by Mouton, The Hague,
Netherlands; "Hyperconection by the Lower Middle Classes" ©
1968 by
Mouton, The Hague, Netherlands; "On the Mechanism of Linguistic Change"
© 1965 by Georgetown University School of Languages and Linguistics; "The
Study of Language in Its Social Context" ©
1970 by Springer-Verlag, i.A.,
Heidelberg, Germany

Library of Congress Catalog Card Number 72-80375

ISBN (clothbound edition): 0-8122-7657-4
ISBN (paperbound edition): 0-8122-1052-2
Printed in the United States of America
For Uriel Weinreich
CONTENTS

List of Figures ix

List of Tables xi

Introduction xiii

1. The Social Motivation of a Sound Change 1

2. The Social Stratification of (r) in New York City
Department Stores 43
3. The Isolation of Contextual Styles 70
4. The Reflection of Social Processes in Linguistic
Structures 110
5. Hypercorrection by the Lower Middle Class as a
Factor in Linguistic Change 122
6. Subjective Dimensions of a Linguistic Change in
Progress 143
7. On the Mechanism of Linguistic Change 160
8. The Study of Language in Its Social Context 183
9. The Social Setting of Linguistic Change 260
Bibliography 327
Index 337

Vll
LIST OF FIGURES

Fig. 1.1. Location of 69 Informants on Martha's Vineyard 5

Fig. 1.2. Measurement of Typical (ay) Diphthongs at First
Formant Maximum 15
Fig. 1.3. Correlation of Instrumental Measurement and
Impressionistic Ratings of Centralization 16
Fig. 1.4. Phonetic Determination of Centralization 19
Fig. 2.1. Overall Stratification of (r) by Store 51
Fig. 2.2. Percentage of All (r-1) by Store 52
Fig. 2.3. Stratification of (r) by Store for Native New York
White Sales Women 56
Fig. 2.4. Stratification of (r) by Occupational Groups in
Macy's 56
Fig. 2.5. Stratification of (r) by Store and Age Level 59
Fig. 2.6. Classification of (r) by Age and Class 60
Fig. 2.7. Development of Class Stratification of (r) for Casual
Speech 62
Fig. 2.8. Hypothetical Distribution of (r) as an Incoming
Prestige Feature 63
Fig. 3.1. Correction of (aeh) in Formal Styles 76
Fig. 3.2. Effect of White Noise and Orientation on Use of
(dh) 98
Fig. 4.1. Class Stratification of (th) 113
Fig. 4.2. Class Stratification of (r) 114
Fig. 4.3. Development of Class Stratification of (r) 116
Fig. 5.1. Class Stratification for (eh) 127
Fig. 5.2. Style Stratification of (oh) 128
Fig. 5.3. Class Stratification for (oh) 129
Fig. 5.4. Class Stratification of (r) for Styles B-D' 137
Fig. 5.5. Acquisition of Prestige Norms in Families with Two
or More Children 140
Fig. 6.1. Subjective Evaluation Forms 147
Fig. 6.2. Class Stratification and Subjective Evaluation of (r) 151
Fig. 6.3. Development of Social Evaluation of (r) 153
ix
 LIST OF FIGURES

Fig. 7.1. Stages in Centralization of (aw) 168

Fig. 7.2. Class Stratification for (oh) 172
Fig. 7.3. Distribution of (oh) Index Scores 173
Fig. 8.1. "All the circles don't have dots in them." 195
Fig. 8.2. Four Relations Governing -t,d Deletion 219
Fig. 8.3. Class and Style Stratification of (ing) in New York
City 239
Fig. 8.4. Class and Style Stratification of (ing) in Norwich 242
Fig. 8.5. Percent Rating Middle-Class Speaker Higher than
Working-Class Speaker on 3 Scales 250
Fig. 9.1. Four Stages in the New York City Chain Shift
ahr -» ohr —» uhr 280
Fig. 9.2. Vowel System of CM., 16 years, Chicago 288
LIST OF TABLES

Table 1.1. Population of Martha's Vineyard 5

Table 1.2. Centralization of (ay) and (aw) by Age Level 22
Table 1.3. Geographical Distribution of Centralization 25
Table 1.4. Centralization by Occupational Groups 26
Table 1.5. Centralization by Ethnic Groups 26
Table 1.6. Centralization and Orientation towards Martha's
Vineyard 39
Table 2.1. Detailed Distribution of (r) by Store and Word
Position 53
Table 2.2. Distribution of (r) for Complete Responses 53
Table 2.3. Distribution of (r) for Black Employees 54
Table 2.4. Distribution of (r) by Floor in Saks 57
Table 2.5. Distribution of (r) by Estimated Age 58
Table 5.1. Distribution of Index of Linguistic Insecurity
Scores 133
Table 6.1. (r)-Positive Response to the Two-Choice Test by
SEC and Age 149
Table 6.2. (r)-Positive Response to the Four-Choice Test by
SEC and Age 152
Table 6.3. (r)-Positive Response to the Four-Choice Test by
Age and (th)-Sensitivity 155
Table 6.4. (r)-PositiveResponse to the Four-Choice Test for
New York and Out-of-Town Respondents 156
Table 7.1. Centralization Indexes by Age Level 167
Table 7.2. Average (oh)-Indexes by Age Level and Ethnic
Group in Casual Speech 173
Table 7.3. (oh)-Negative Response by SEC and Age Level 176
Table 8.1. Effect of Variant Forms of Fig. 8.1 on NEG-V and
NEG-Q Responses 196
Table 8.2. Inconsistent NEG-Q and NEG-V Responses to
Diagrams and Sentences 197
Table 8.3. Constraints on -t,d Deletion

in the Black Speech Community 222

xi
 LIST OF TABLES

Table 8.4. The Effect of Following Grammatical Environment

on Deletion of is in BEV with pronoun subjects 230
Table 9.1. Four Sound Changes in Progress in the Swiss
Village of Charmey, 1899 278
Table 9.2. Centralization of (ay) and (aw) in Three
Generations of English Speakers 279
Table 9.3. Phones Used for (ar) and (or) in Casual Speech in
Minersville, Utah 281
Table 9.4. Average (ay) and (aw) Values for All White New
York City Informants 289
Table 9.5. (r) in Casual Speech by Age and Class 291
Table 9.6. Percent of (r)-00 Speakers in Word Lists by Age
and Class 291
Table 9.7. Frequency of Final and Preconsonantal (r) for
White Speakers in Hillsboro, N.C. 292
Table 9.8. Social Stratification of Five Spanish Variables in
Panama 293
Table 9.9. Development of Five Spanish Variables by Age
Groups 294
Table 9.10. Distribution of (eh) by Age and Ethnic Group in
New York City 297
Table 9.11. Comparison of Men and Women for (eh) and
(oh) in New York City 302
Table 9.12. Acquisition of Philadelphia Phonology by Four
Preadolescent Boys 306
Table 9.13. Percent Stigmatizing Raised (eh) in SR Tests by
Age and Ethnic Group 312
Table 9.14. Percent Stigmatizing Raised (eh) in SR Test by
Age and SEC 312
Table 9.15. Proportion Using Any [ai] in Interviews by Age
and SEC 316
 NTRODUCTION

I have resisted the term sociolinguistics for many years, since

it implies that there can be a successful linguistic theory or prac-
tice which is not social. When I first published the studies of Mar-
tha's Vineyard and New York City that form the basis of the first

part of this book, it seemed necessary make that point again

to
and again. In spite of a considerable amount of sociolinguistic
activity, a socially realistic linguistics seemed a remote prospect
in the 1960's. The great majority of linguists had resolutely turned
to the contemplation of their ownhave not yet
idiolects. We
emerged from the shadow of our intuitions, but it no longer seems
necessary to argue about what is or is not linguistics. There is a
growing realization that the basis of intersubjective knowledge in
linguistics must be found in speech —
language as it is used in
everyday life by members of the social order, that vehicle of com-
munication in which they argue with their wives, joke with their
friends, and deceive their enemies.
When I first entered linguistics as a student, in 1961, it was my
intention to gather data from the secular world. The early projects
that I constructed were "essays in experimental linguistics," carried

out in ordinary social settings. My aim was to avoid the inevitable

obscurity of texts, the self-consciousness of formal elicitations, and
the self-deception of introspection. A decade of work outside the
university as an industrial chemist had convinced me that the
everyday world was stubborn but consistently so, baffling at the
outset but rewarding in the long run for those who held to its
rational character. A simple review of the literature might have
convinced me that such empirical principles had no place in
linguistics: there were many ideological barriers to the study of
language in everyday life. First, Saussure had enunciated the prin-
ciple that structural systems of the present and historical changes
of the past had to be studied in isolation (1949:124). That principle
xm
xiv INTRODUCTION

had been consistently eroded by Martinet (1955) and others who

found structure in past changes, but little progress had been made
in locating change in present structures. The second ideological
barrier explicitly asserted that sound change could not in principle
be directly observed. Bloomfield defended the regularity of sound
change against the irregular evidence of the present by declaring
(1933:364) that any fluctuations we might observe would only be
cases of dialect borrowing. Next Hockett observed that while sound
change was too slow to be observed, structural change was too fast
(1958:457). The empirical study of linguistic change was thus re-
moved from the program of 20th-century linguistics.
A third restriction was perhaps the most important: free varia-
tion could not in principle be constrained. The basic postulate of
linguistics (Bloomfield 1933:76) declared that some utterances were
the same. Conversely, these were in free variation, and whether
or not one or the other occurred at a particular time was taken to
be linguistically insignificant. Relations of more or less were there-
fore ruled out of linguistic thinking; a form or a rule could only
occur always, optionally, or never. The internal structure of varia-
tion was therefore removed from linguistic studies and with it,
the study of change in progress.
It was also held that feelings about language were inaccessible

and outside of the linguist's scope (Bloch and Trager 1942). The
social evaluation of linguistic variants was therefore excluded from
consideration. This is merely one aspect of the more general claim
that the linguist should not use nonlinguistic data to explain lin-
guistic change (see the first section of Ch. 9). Throughout these
discussions, we see many references to what the linguist can or
cannot do as a linguist.
I might indeed have disregarded all these restrictions by the
force of my own inclination and resistance to authority. But I was
fortunate to encounter at Columbia University a teacher not much
older than myself, whose own insight, imagination, and creative
force had long since bypassed these restrictions. It is impossible
for me to estimate the contribution of Uriel Weinreich to the stud-
ies reported here. I learned from him in courses on syntax, seman-
tics, dialectology, and the history of linguistics; he supervised the

work on Martha's Vineyard (Ch. 1) which was my Master's essay,

and the study of New York City (Chs. 2-6) which was my disserta-
tion; yet in all this he did not put forward his own view or direct
suggestion about which way to turn. But with caution, restraint,
Introduction xv

and example, he helped to direct my own projects into the most

profitable channels. Weinreich had an extraordinary sense of direc-
tion in linguistics; he rarely made a misstep in his own research
projects, and we all profited by his insight. I have recently had
the opportunity to read some of Weinreich's unpublished sketches
and projects for the study of multilingualism and social variation
in the speech community; I found that his thinking had anticipated
my own by many years and undoubtedly played a larger part in
the results given here than might appear in the overt references.
More than anything else, I benefitted from Weinreich's calm con-
viction that we were moving in the direction that a rational and
realistic linguistics must inevitably follow.
In 1966, Weinreich proposed to Marvin Herzog and myself that
we write a joint paper on "Empirical Foundations for a Theory
of Language Change" for a conference at the University of Texas.
As we delivered it, this paper embodied the results of my own
work in New York and Martha's Vineyard, Herzog's findings on the
dialectology of Yiddish in Northern Poland, and Weinreich's overall
insight that created the Language and Culture Atlas of Ashkenazic
Jewry. This was set in a larger view of the history of linguistics that
was entirely the product of Weinreich's scholarship. In the spring
of 1967, when Weinreich realized that he had only a short time to
live, he turned paper with great energy.
to the final revision of this
In the last two weeks of his life, Weinreich recast the introduction
to this paper in a way that captures clearly his overall view of the
nature of language and its relation to society. It states the major
theme of this volume better than any passage of my own:

The have not so far jibed well with the structural

facts of heterogeneity
approach language ... for the more linguists became impressed with
to
the existence of structure of language, and the more they bolstered this
observation with deductive arguments about the functional advantages
of structure, the more mysterious became the transition of a language
from state to state. After all, if a language has to be structured in order
to function efficiently, how do people continue to talk while the language
changes, that is, while it passes through periods of lessened systematic-
ity? . . . The solution, we will argue, lies in the direction of breaking down
the identification of structuredness with homogeneity. The key to a ra-
tional conception of language change— indeed, of language itself— is the
possibility of describing orderly differentiation in a language serving a
community. We will argue that nativelike command of heterogeneous
 INTRODUCTION

structures is not a matter of multidialectalism or "mere" performance,

but is part of unilingual linguistic competence. One of the corollaries

of our approach is that in a language serving a complex (i.e., real) com-
munity, it is absence of structured heterogeneity that would be dysfunc-
tional. [Weinreich, Labov, and Herzog 1968:100-101]

The first six chapters of this book are reports of particular studies
which form part of the evidence for the view which Weinreich ex-
pressed and the last three continue the argument in a broader frame-
work. Some chapters have appeared previously, but are here con-
siderably revised. Ch. 3 and 6 have not appeared in print before,
although they are based on material reported in part in Labov 1966.
Chs. 2-6 cover a large part of the methods and findings of The
Social Stratification of English in New York City (Labov 1966a);
each is organized about a particular problem which was attacked
in this work. Chs. 7-9 are synthetic studies which combine these
findings with others to project a larger view of the nature of lan-
guage structure and language change. Chapter 7, "On the mechanism
of linguistic change," uses the Martha's Vineyard and New York
findings in an overall projection for the course of change which
preceded the statement of "Empirical Foundations." Ch. 8, "The
study of language in its social context" is a general survey of the
problems, findings, and prospects of a socially realistic linguistics.
It may be considered a short version of a more general text on

"Secular Linguistics," addressed to those who are interested in

entering the speech community to find a solid empirical base for
linguistic theory. This chapter draws heavily upon research on the
black English vernacular which is reported in greater detail in
Chs. 1-4 in Language in the Inner City (1972) and in the two-volume
research report on Cooperative Research Project3288 (Labov,
Cohen, Robins, and Lewis 1968). This work explores more deeply
than Chs. 2-6 the variation which is found within a rule schema,
and develops the formal treatment of variable rules. Readers who
would like to pursue the linguistic issues raised in Ch. 8 will find
a more detailed presentation in Chs. 3 and 4 of Language in the Inner
City, which are concerned, respectively, with contraction and de-
letion of the copula and negative attraction and concord. Ch. 3 of
that book presents the development of variable rules in much greater
detail than was possible in Ch. 8, incorporating the revisions of the
Cedergren-Sankoff model.
Introduction XVII

Ch. 9 forms a companion section to Ch. 8, examining the ana-

chronic aspects of the synchronic matters presented there. Both
Chs. 8 and 9 draw heavily upon the findings of other investigators,
and the further developments that proceeded from the work re-
ported in Chs. 1-6 are represented there. Together, Chs. 8 and 9
form a short text in sociolinguistic problems and methods, as well
as the sociolinguistic approach to linguistic theory, and provide a
framework on which sociolinguistic courses have been organized.
I am indebted to many colleagues for help and critical improve-

ments in the work reported here. Michael Kac carried out many of
the Lower East Side interviews which provide the data for Chs. 3-6.
This secondary survey was carried out with the help of the re-
search department of Mobilization for Youth; I am indebted to the
research director, Wyatt Jones, for his assistance at many points.
The survey itself benefited greatly from the instruction and advice
of Herbert Hyman of Columbia University. The work on the black
English vernacular referred to in Ch. 8 was the product of a joint
effort. The analysis and theoretical insights provided by Paul Cohen
made important contributions to the conclusions presented there,
and the field work of Clarence Robins and John Lewis was the
basis of everything that was accomplished.
The general analysis of variation in Ch. 8 has benefitted from
many exchanges with C.-J. Bailey. The formal interpretation of
variable rules included many
contributions of Joshua Waletzky; in
the present revised formulation, my indebtedness to Henrietta
Cedergren and David Sankoff is acknowledged at many points. Their
quantitative contribution to the treatment of variable constraints
seems to me a major advance which will strongly influence the
future direction of this field.
At many research and analysis, I received
critical points in this
important help from my wife, Teresa, who has deepened my own
understanding of the structure of the social orders that I have en-
countered.
During the first stages in the preparation of this book, I was the
recipient of a fellowship from the Guggenheim Foundation, whose
help I would like to acknowledge here with thanks.
During the year 1971-72, I served as Research Professor with the
Center for Urban Enthnography, and I am deeply indebted to the
Center for the support which made it possible to assemble this
volume, along with Language in the Inner City. The original im-
xviii INTRODUCTION

petus to put these studies together into a single volume and organize
them into a single coherent framework came from Erving Goffman,
whose help and encouragement is acknowledged with many thanks.

Throughout these pages, it will be obvious that my heaviest

indebtedness must be toward the many speakers of English who
have invited me into their homes, shared their porches, street cor-
ners and park benches with me, and turned aside from other busi-
ness to talk, transforming their own experience into language for
my Only a small part of what I have learned from them
benefit.
can be found here. But I hope that this work does reflect the in-
finite variety of every day life, and the great satisfaction of encoun-
tering and recording the users of language. Those who have tapped
the real resources of the speech community find that field work is

a rich pursuit that never exhausted. I have found that there is

is

no greater pleasure than to travel as a privileged stranger to all

parts of the world, to be received with kindness and courtesy by
men and women everywhere, and to share their knowledge and
experience with them as it reappears in their language. The linguist
who enters that world can only conclude that man is the right
inheritor of the incredibly complex structure which we are now
trying to analyze and understand.
 The Social Motivation
1 of a Sound Change

THE work which is reported in this chapter concerns the

direct observation of a sound change in the context of the community
life from which it stems. 1 The change is a shift in the phonetic
position of the first elements of the diphthongs /ay/ and /aw/, and
the community is the island of Martha's Vineyard, Massachusetts.
By studying the frequency and distribution of phonetic variants of
/ay/ and /aw/ in the several regions, age levels, occupational and
ethnic groups within the island, it will be possible to reconstruct
the recent history of this sound change; by correlating the complex
linguistic pattern with parallel differences in social structure, it will
be possible to isolate the social factors which bear directly upon the
linguistic process. It is hoped that the results of this procedure will
contribute to our general understanding of the mechanism of lin-
guistic change.
The problem of explaining language change seems to resolve itself
into three separate problems: the origin of linguistic variations; the
spread and propagation of linguistic changes; and the regularity of
linguistic change. The model which underlies this three-way division
requires as a starting point a variation in one or several words in
the speech of one or two individuals. 2 These variations may be
induced by the processes of assimilation or differentiation, by anal-
ogy, borrowing, fusion, contamination, random variation, or any
number of processes in which the language system interacts with

1. First published in Word, 19: 273-309 (1963). An abbreviated version was given at
Annual Meeting
the 37th of the Linguistic Society of America in New York City on
December 29, 1962.
2. See Sturtevant 1947: Ch. 8: "Why are Phonetic Laws Regular?"
1
2 SOCIOLINGUISTIC PATTERNS

the physiological or psychological characteristics of the individual.

Most such variations occur only once, and are extinguished as
quickly as they arise. However, a few recur, and, in a second stage,
they may be imitated more or less widely, and may spread to the
point where the new forms are in contrast with the older forms along
a wide front. Finally, at some later stage, one or the other of the
two forms usually triumphs, and regularity is achieved.
Whereas for the first stage we are often overwhelmed with an
excess of possible explanations, we have quite the reverse situation
in attempting to account for the propagation and regularity of lin-
guistic changes. A number of earlier theories which proposed general
psychological, physiological or even climatic determinants have been
discarded for some time. 3 The contribution of internal, structural
forces to the effective spread of linguistic changes, as outlined by
Martinet (1955), 4 must naturally be of primary concern to any linguist
who is investigating these processes of propagation and regulariza-
tion. However, an account of structural pressures can hardly tell the
whole story. Not all changes are highly structured, and no change
takes place in a social vacuum. Even the most systematic chain shift
occurs with a specificity of time and place that demands an explana-
tion.
Widely divergent ideas appear to exist as to what comprises an
explanation of the mechanism of change. The usual diachronic
procedure, as followed in palaeontology or geology, is to explore the
mechanism of change between states by searching for data on inter-
mediate states. It follows that we come closer and closer to an
accurate depiction of the mechanism of change as the interval be-
tween the two states we are studying becomes smaller and smaller.
This is certainly the method followed by such historical linguists
as Jespersen, Kokeritz, and Wyld, and it is the motivation behind
their extensive searches for historical detail. On the other hand, a
viewpoint which favors the abstract manipulation of data from
widely separated states has been propounded by M. Halle (1962);
explicit defense of a similar attitude may be found in H. Pilch's (1955)

3. A number of these theories are reviewed in Sommerfelt 1930.

4. The empirical confirmation of many of Martinet's ideas to be found in Moulton's
investigation of Swiss German dialects has provided strong motivation for some of
the interpretations in the present essay. In particular, see Moulton 1962.
The Social Motivation of a Sound Change 3

study of the vowel systems of Shakespeare, Noah Webster, and

present-day America. Neither Halle nor Pilch distinguish the three
aspects of change outlined above.
It would seem that the historical approach is more appropriate

to an empirical science concerned with change, even over a narrow

time span, as this approach leads to statements which are increas-
ingly subject to confirmation or disconfirmation. At the same time,
such a close view of historical change makes us increasingly sceptical
of the value of limitations on the kinds of data which may be con-
sidered: as, for instance, that the linguist explain linguistic events
only by other linguistic events. One would expect that the appli-
cation of structural linguistics to diachronic problems would lead
to the enrichment of the data, rather than the impoverishment of
5
it.

The point of view of the present study is that one cannot under-
stand the development of a language change apart from the social
life of the community in which it occurs. Or to put it another way,

social pressures are continually operating upon language, not from

some remote point in the past, but as an immanent social force acting
in the living present.
Sturtevant (1947:74-84) has outlined a concise theory of the spread
and consolidation of language changes which consistently views this
process in its social dimension. One sentence in particular will serve
as an excellent theme for this investigation:

Before a phoneme can spread from word to word ... it is necessary that
one of the two rivals shall acquire some sort of prestige. 6

It is hoped that the study of the particular case under discussion

will lend support to this general view of the role of social interaction
in linguistic change.

5. For a parallel criticism of restrictions on the data imposed by Bloomfieldian

linguistics, see W. Diver's review of W. P. Lehmann's Historical Linguistics, in Word,
19:100-105 (1963).
See also H. Hoenigswald's remarks in "Are There Universals of Linguistic
6.

Change?" in Greenberg, ed., 1963, fn. 8: "Sound changes can apparently not be entirely
predicted from internal systemic stresses and strains, nor can they be explained as
the effect of scatter around a target or norm; they have direction and are in that sense
specific, much like other happenings in history."
 SOCIOLINGUISTIC PATTERNS

The Island of Martha's Vineyard

The island of Martha's Vineyard, Dukes County, Massachusetts,

was chosen as a laboratory for an initial investigation of social
patterns in linguistic change. 7 Martha's Vineyard has the advantage
from the mainland by a good
of being a self-contained unit, separated
three miles of the Atlantic Ocean. At the same time, the Vineyard
has enough social and geographic complexity to provide ample room
for differentiation of linguistic behavior. We are also fortunate in
having the records of the Linguistic Atlas of New England (hence-
forth abbreviated LANE) as a background for the investigation. 8 It
is over thirty years since Guy Lowman visited Martha's Vineyard;

his interviews with four members of the old families of the island
give us a firm base from which to proceed, and a time depth of one
full generation which adds considerably to the solidity of the con-
clusions which can be drawn.
Fig. 1.1 shows the general outlines of Martha's Vineyard, and Table
1.1 gives the population figures from the 1960 Census. The island
is divided into two parts by an informal, but universally used dis-

tinction between up-islcmd and down-isicmd. Down-island is the

region of the three small towns where almost three-fourths of the
permanent population live. Up-isiand is strictly rural, with a few
villages, farms, isolated summer homes, salt ponds and marshes, and
a large central area of uninhabited pine barrens.
As we travel up-island from Vineyard Haven, we come first to the
town of West Tisbury, which contains some of the most beautiful
farms and fields of the island, now largely untilled and ungrazed.
At Chilmark, the ground rises to a series of rolling hills which look
out to the Atlantic on one side, and to Vineyard Sound on the other.
Chilmark's salt pond is permanently open to the Sound through a
narrow channel, and so serves as a permanent harbor for the dozen
fishermen who still operate from the docks of the village of
Menemsha in Chilmark. Finally, at the southwest corner of the island,

7. For further details on the social and economic background of Martha's Vineyard,
see my
1962 Columbia University Master's Essay, "The Social History of a Sound
Change on the Island of Martha's Vineyard, Massachusetts," written under the
direction of Professor Uriel Weinreich.
8. Kurath et al. 1941. Background information on the informants is to be found in
Kurath 1939.
The Social Motivation of a Sound Change

Down— islo nd

Fig.1.1. Location of the 69 informants on Martha's Vineyard.

Ethnic origin is indicated as follows: English, Portuguese,
V Indian. Symbols placed side by side indicate members of the
same family.

TABLE 1.1.

POPULATION OF MARTHA'S VINEYARD

Down-island [towns] 3,846

Edgartown 1,118
Oak Bluffs 1,027
Vineyard Haven 1,701
Up-island [rural] 7,777
Edgartown 256
Oak Bluffs 292
Tisbury 468
West Tisbury 360
Chilmark 238
Gay Head 103

Total 5,563

Source: From U.S. Bureau of the Census, U.S.

Census of Population: 1960. Number of Inhab-
itants. Massachusetts. Final Report PC(1)— 23a
(Washington, D.C.: GPO, 1962), Table 7, p. 23-11
6 SOCIOLINGUISTIC PATTERNS

there is the promontory of Gay Head, and the houses of the 103
Indians who represent the original inhabitants of Martha's Vineyard.
The 6,000 native Vineyarders fall into four ethnic groups which
are essentially endogamous. First, there are the descendants of the
old families of English stock, who first settled the island in the 17th
and 18th centuries: the Mayhews, Nortons, Hancocks, Aliens, Tiltons,
Vincents, Wests, Pooles— all closely related after ten generations of
intermarriage. Secondly, there is a large group of Portuguese descent,
immigrants from the Azores, Madeira, and the Cape Verde Islands.
There are Portuguese all along the southeastern New England coast,
but the Vineyard has the largest percentage of any Massachusetts
county. In 1960, 11 percent of the population was of first- or second-
generation Portuguese origin; with the third- and fourth-generation
Portuguese, the total would probably come close to 20 percent. 9
The third ethnic group is the Indian remnant at Gay Head. The
fourth is the miscellaneous group of various origins: English, French
Canadian, Irish, German, Polish. Though the sum total of this resid-
ual group is almost 15 percent, it is not a coherent social force, and
we will not consider it further in this paper. 10
Another group which will not be considered directly is the very
large number of summer residents, some who flood the island
42,000,
in June and July of every year. This tide of summer peopie has had
relatively little direct influence on the speech of the Vineyard, al-

though the constant pressure from this direction, and the growing
dependence of the island upon a vacation economy, has had power-
ful indirect effects upon the language changes which we will con-
sider.

The Vineyard is best known to linguists as an important relic area

of American English: an island of r-pronouncers in a sea of r-lessness.
With a 320-year history of continuous settlement, and a long record
of resistance to Boston ways and manners, the island has preserved
many archaic traits which were probably typical of southeastern

9. From U.S. Bureau of the Census, U.S. Census of Population: 1960. General Social

—
and Economic Characteristics. Massachusetts. Final Report PC(1) 23c (Washington,
D.C.: Government Printing Office, 1962), Table 89, p. 23-260.
10. There is a sizeable number of retired mainlanders living on the Vineyard as

year-round residents. While they are included in the population total, they do not
form a part of the social fabric we are considering, and none of the informants is
drawn from this group.
The Social Motivation of a Sound Change 7

New England before 1800. The most striking feature, still strongly
entrenched, is the retention of final and preconsonantal /r/.
11
New
England short /o/ is still well represented among the older speakers.
Exploratory studies of the Vineyard in 1961 showed that most of the
special traits of the island speech shown on the LANE maps may
still be found among traditional speakers from 50 to 95 years old.

Lexical survivals of 17th-century English are even clearer indica-

tions of the archaic nature of the Vineyard tradition. We find ban-
nock, for a fried cake of corn meal, studdled for 'dirty, roiled' water,
in addition to such items as tempest and buttry listed in the LANE.
Perhaps the most dramatic evidence of the fact that the Vineyard
represents an underlying stratum is the presence of beJIy-gut, for a
face-down sled ride. In LANE records, this form is shown on the
Vineyard and in western New England; in the intervening area, it
has been overlaid by three successive layers beliy-bump, belly-flop, —
and currently, no word at all. 12
As interesting as the structure of Martha's Vineyard English may
be, it is not the purpose here to contrast one static system with
another. We would like to understand the internal structure of
Vineyard English, including the systematic differences which now
exist and the changes now taking place within the island. For this
purpose, we will select for study a linguistic feature with the widest
possible range of variation and the most complex pattern of distri-
bution characteristic of Martha's Vineyard.

Selection of the Linguistic Variable

It would be appropriate to ask at this point what are the most

useful properties of a linguistic variable to serve as the focus for

11. On the LANE maps, we find that Guy Lowman regularly recorded the up-island
/r/ as (V| in [weia\ haa^d, baam], and down-island /r/ as [?] in the same positions.
Essentially the same pattern is to be found among the older speakers today, though
not with the regularity that Lowman noted. It is possible that this treatment of /r/
was in fact LANE was much more concerned
intended as a broad transcription, for the
with vowels than consonants.
12. See Kurath 1949, Fig. 162. Belly-flop (and the corresponding lexical item in other

regions) has generally shifted for the younger generation to denote a flat dive into
the water. Coasting is now a less important sport, and its terminology is appropriately
impoverished. The lexical data derived from my own study of Martha's Vineyard is
analyzed in detail in W. Labov, "The Recent History of Some Dialect Markers in
Martha's Vineyard, Massachusetts," in Davis 1972.
8 SOCIOLINGUISTIC PATTERNS

the study of a speech community. First, we want an item that is

frequent, which occurs so often in the course of undirected natural

conversation that its behavior can be charted from unstructured
contexts and brief interviews. Secondly, it should be structural: the
more the item is integrated into a larger system of functioning units,
the greater will be the intrinsic linguistic interest of our study. Third,
the distribution of the feature should be highly stratified: that is, our
preliminary explorations should suggest an asymmetric distribution
over a wide range of age levels or other ordered strata of society.
There are a few contradictory criteria, which pull us in different
directions. On the one hand, we would like the feature to be salient,
for us as well as for the speaker, in order to study the direct relations
of social attitudes and language behavior. But on the other hand,
we value immunity from conscious distortion, which greatly sim-
plifies the problem of reliability of the data. 13
conducted on the Vineyard in 1961,
In the exploratory interviews
many were noted that were plainly parallel to
structural changes
changes taking place on the mainland under the influence of the
standard Southeast New England pattern. Changes in phonemic
inventory were found: New England short /o/ is rapidly disappear-
ing; the two low back vowels, /a/ and /o/ are merging. Important
changes in phonemic distribution are occurring: the /or~or/ dis-
tinction is disappearing: initial /hw/ is giving way to /h/.
14
Shifts
in structured lexical systems, all in the direction of regional stand-
ards, can be traced. Archaic syntactic features are disappearing. Yet
as interesting as these changes may be, there is no reason to think
that their distribution will follow a pattern peculiar to the Vineyard.
In the case of postvocalic /r/, however, we do have a linguistic
variable defined by the geographical limits of the island, which
follows a social pattern idiosyncratic to Martha's Vineyard. In some
island areas, retroflexion is increasing, and in others, decreasing; as

13. Many ingenious devices are needed to detect and eliminate deceit on the part
of metropolitan informants, whether intended or not. On Martha's Vineyard, this is
much problem, but the effects of the interview situation are evident in the
less of a
careful style of some informants.
14. The disappearance of New England short /o/ follows the pattern described in
Avis (1961). Exploratory interviews at other points in southeastern New England

(Woods Hole, Falmouth, New Bedford, Fall River, Providence, Stonington) indicate
that the loss of the /or~or/ and /hw —
w-/ distinctions is parallel to that on
Martha's Vineyard.
The Social Motivation of a Sound Change 9

we will note later, the social implications of this fact can not be
missed. The variations in /r/ are frequent, salient, and involve
far-reaching structural consequences for the entire vowel system.
However, the preliminary exploration of the Vineyard indicated
that another variable might be even more interesting: differences in
the height of the first element of the diphthongs /ay/ and /aw/.
Instead of the common southeast New England standard [ai] and [au],
one frequently hears on Martha's Vineyard [m] and [eu], or even [ai]
and [au]. This feature of centralized diphthongs 15 is salient for the
linguist, but not for most speakers; it is apparently quite immune
to conscious distortion, as the native Vineyarders are not aware of
it,nor are they able to control it consciously. As far as structure is
concerned, we cannot neglect the structural parallelism of /ay/ and
/aw/; on the other hand these diphthongs are marked by great
structural freedom in the range of allophones permitted by the
system. These are strictly subphonemic differences. Since there are
no other up-gliding diphthongs with either low or central first ele-
ments in this system, it is not likely that continued raising, or even
fronting or backing, would result in confusion with any other pho-
neme.
The property of this feature of centralization which makes it
appear exceptionally attractive, even on first glance, is the indication
of a complex and subtle pattern of stratification. This very complex-
ity proves to be rewarding: for when the centralizing tendency is
charted in the habits of many speakers, and the influence of the
phonetic, prosodic, and stylistic environment is accounted for, there
remains a large area of variation. Instead of calling this "free" or
"sporadic" variation, and abandoning the field, we will pursue the
matter further, using every available clue to discover the pattern
which governs the distribution of centralized diphthongs.
The problem becomes all the more significant when it becomes
apparent that the present trend on Martha's Vineyard runs counter
to the long-range movement of these diphthongs over the past two
hundred years. And while this sound change is not likely to become

15. The terms centralized diphthongs, centralization, and degree of centralization

will beused throughout this study to refer to the various forms of the diphthongs
/ay/ and /aw/ with first elements higher than [a]. It is not intended that the terms
themselves should imply any process or direction of change, except when used with
explicit statements to that effect.
10 SOCIOLINGUISTIC PATTERNS

a phonemic change in the foreseeable future, it operates in an area

where far-reaching phonemic shifts have taken place in the past. It
is, in effect, the unstable residue of the Great Vowel Shift.

The History of Centralized Diphthongs

It seems generally agreed that the first element of the diphthong

/ay/ was a mid-central vowel in 16th- and 17th-century English

(Jespersen 1927:234; Kokeritz 1953:216). 16 We may assume that when
Thomas Mayhew took possession of his newly purchased prop-
first

erty of Martha's Vineyard in 1642, he brought with him the pronun-

ciation [ai] in right, pride, wine and wife. The later history of this
vowel in America indicates that [ai] continued to be the favored form
well into the 19th century. 17
When we examine the records of the LANE, we find that central-
ized /ay/ was a healthy survivor in the speech of the Atlas inform-
ants. 18 We find it scattered throughout the rural areas of New Eng-
land, and strongly entrenched in the Genesee Valley of western New
York. It had disappeared completely from the Midland, but was quite
—
regular before voiceless consonants— in both the Upper and Lower
South. This differential effect of voiceless and voiced following
consonants was only a directing influence in the North, but stood
as a regular phonetic rule in the South. On Martha's Vineyard, as
on neighboring Nantucket and Cape Cod, centralized /ay/ was
frequently recorded.
The history of /aw/ differs from that of /ay/ more than our general
expectations of symmetry would lead us to predict. There is reason
to believe that in England the lowering of /aw/ was considerably
in advance of /ay/, and it is not likely that the same Thomas Mayhew
used [au] in house and out (Jespersen 1927:235-36; Kokeritz 1953:

16. Among recent historical linguists, H. C. Wyld (1920:223-25) is a notable exception

in positing a front first element in the transition of M.E.i: to Mod. E. /ay/, relying
on occasional spellings with ey and ei, but without considering the many other
indications of central position.
17. Abundant evidence is given in Krapp 1925(2):186-91.
18. The best view of the distribution of /ay/ may be had from Maps 26-27 in Kurath
and McDavid 1951. Centralized diphthongs are well known as a feature of Canadian
English, where the effect of the voiceless-voiced consonant environment is quite
regular.
The Social Motivation of a Sound Change 11

144_49; Wyld 1920:230-31). The American evidence of the late 18th

and 19th centuries, as summed up by Krapp (1925 2:192-96), points
to [ou] as the conservative, cultured form, giving way to [au] or [au],
with the rural New England form as [aeu] or [eu]. The Linguistic
Atlas records show only a hint of parallelism of /ay/ and /aw/.
(Kurath and McDavid 1951: Maps 28-29). We find [au] mainly in
eastern Virginia, before voiceless consonants, with some small rep-
resentation in upstate New York, but the principal New England form
of [au] stood out against a background of rural and recessive [aeu].
Martha's Vineyard shows very little centralization of /aw/ in the
LANE maps.
This brief review indicates that the isolated position of /aw/ has
facilitated phonetic variation on a truly impressive scale. The first
element has ranged from [i] to [a], from [e] to [o] all within the same
general structural system. Perhaps one reason why /ay/ has not
shown a similar range of variation is the existence of another up-
gliding diphthong, /dv/. 19 In any case, as the stage is set for our
present view of Martha's Vineyard diphthongs, /ay/ is well central-
ized, but /aw/ is not. It may be too strong a statement to say that
this represents the phonetic heritage of the seventeenth-century
Yankee settlers of the island, but we may venture to say that we have
no evidence of any intervening events which disturbed the original
pattern.
As we begin the systematic study of this centralization pattern,
we will refer to the linguistic variables (ay) and (aw) instead of the
phonemes /ay/ and /aw/. Where thesubphonemic differences in
the position of the nucleus of /ay/ and /aw/ are considered to be
in free variation, and linguistically insignificant, the variants of (ay)
and (aw) show significant differences in their distribution and carry
sociolinguistic information. In this case (but not always), the varia-
bles (ay) and (aw) represent the same phonetic substance as the
invariant categories /ay/ and /aw/; the parentheses indicate a
different approach to the analysis of variation. Whereas // means
that internal variation is to be disregarded as insignificant, ( ) indi-
cates that this variation is the prime focus of study.

19. The possibility of phonemic confusion with /oy/ apparently became a reality
in the 17th and 18th centuries, in bothEngland and America, when both diphthongs
had central first elements.
12 SOCIOLINGUISTIC PATTERNS

The Investigation of (ay) and (aw)

The summer visitor to Martha's Vineyard gets only a fleeting

impression of the native speech pattern. Seven out of every eight
human beings on the island are visitors like himself. But for the
Vineyarder, there is no effect of dilution. For him, summer visitors
have very little status on the island and their ephemeral nature is
convincingly demonstrated in the first week in September of every
year, when they disappear even more quickly than the insect popu-
lation of the summer months. The normal native speech of Martha's
Vineyard can then be heard as the dominant sound in public places.
A knock on any up-island door will no longer produce a Back Bay
stockbroker, but the rightful owner in possession once again. As a
rural up-islander he is very likely to use a high degree of central-
ization of (ay) and (aw); but in the small town areas of down-island
one may also hear this feature, particularly in words such as right,
white, twice, life, wife, like, but not so much in while, time, line,
I, my, try. Similarly, one may hear in the streets of Vineyard Haven

centralized forms in out, house, doubt, but not so much in now, how,
or around.
In order to study this feature systematically, it was necessary to
devise an interview schedule which would provide many examples
of (ay) and (aw) in casual speech, emotionally colored speech, careful
speech, and reading style. The first of these diphthongs is more than
twice as frequent as the second, but even so, several devices were
required to increase the concentration of occurrences of both.

1. A lexical questionnaire, using the regional markers shown as most

significant in the maps of the LANE, supplemented with recent
observations, and concentrating on the following words contain-
ing (ay) and (aw):

spider rareripe iodine dying out

sliding swipe quinine flattening out
scrimy dowdy
white bread nigh outhouse
white of egg pie frying pan backhouse
nightcrawler sty fry pan crouch
lightning bug firefly mow
Italian shiretown rowen
The Social Motivation of a Sound Change 13

2. Questions concerning value judgments, exploring the social ori-

entation of the respondent, were so phrased as to elicit answers
containing (ay) and (aw) forms. 20 Answers to such questions often
gave a rich harvest of diphthongal forms, with contrasting uses
of emotionally stressed and unstressed variants.
3. A special reading, used mainly in the high school, was offered
ostensibly as a test of the ability to read a story naturally. 21 Since
these readings gave the most exact comparisons between speakers,
they were utilized for the spectrographic measurements discussed
below.

In addition to the formal interview, observations were made in

a great many casual situations: on the streets of Vineyard Haven and
Edgartown, in diners, restaurants, bars, stores, docks, and many
places where the general sound
of public conversation could be
noted, not effectively recorded. But these notations only served
if

as a supplementary check on the tape-recorded interviews. The basic

information was gathered in the course of 69 interviews with native
island speakers made in three periods: August 1961, late Septem-
ber-October 1961, January 1962. These 69 interviews provide the basis
for the discussion to follow.
The 69 speakers, somewhat more than 1 percent of the population,
represent a judgment sample of the community of native residents,
and the groups which are important in the social life and value
systems of the island. The sampling is proportional to area rather
than population: 40 are up-islanders, and only 29 are from down-
island, though over 70 percent of the people live down-island. The
most important occupational groups are represented: 14 in fishing,
8 in farming, 6 in construction, 19 in service trades, 3 professionals,

20. "When we speak of the right to life, liberty and the pursuit of happiness, what

does right mean? ... Is it in writing? ... If a man is successful at a job he doesn't
like, would you still say he was a successful man?" These questions were generally

successful in eliciting the informant's versions of the italicized words.

21. This 200-word reading is constructed as a story told by a teenage Vineyard boy,
of the day he found out his father wasn't always right. An excerpt will show the
technique involved: "After the high winds last Thursday, we went down to the mooring
to see how the boat was making out. . . . My father started to pump out the bottom,
and he told me to find out if the outboard would start. I found out all right. I gave
her a couple of real hard pulls but it was no dice. 'Let me try her,' my father said.
'Not on your life,' I told him. 'I've got my pride.'"
14 SOCIOLINGUISTIC PATTERNS

5 housewives, 14 students. The three main ethnic groups are repre-

sented: 42 of English descent, 16 Portuguese, and 9 Indian.
The locations of the 69 informants are shown on Fig. 1.1, coded
by ethnic group. It may be understood that a large proportion of
those engaged in fishing are to be found in Chilmark; the farmers
are well inland, mainly in West Tisbury; the service trades are
heavily concentrated in Edgartown and Vineyard Haven. Of Guy
Lowman's four LANE informants, one was in Chilmark, one in West
Tisbury, and two in Edgartown.
As a result of these 69 interviews, we have about 3,500 instances
of (ay) and 1,500 instances of (aw) as the basic data for this study.

Scales of Measurement

An important step was to construct a reliable, inter-subjective

index to the degree of centralization. In the original transcriptions
of the tape-recorded interviews a six-point scale of height of the first
element was used, ranging from the standard New England form [ai]
to the fully centralized [ai]. 22 Such a transcription was intended to
push the distinctions noted to the limits of auditory discrimination.
This corresponded to the practice of the LANE, in which the same
number of degrees of height can be symbolized. However, it was
recognized that such fine distinctions could probably only be repro-
duced consistently by individuals who had attained a high degree
of convergence, and then over a very short time span.
Independent instrumental measurements were used to reduce the
scale by objective criteria, and to give a certain degree of objective
validity to the entire system of transcription.
Acoustic spectrograms were made of 80 instances of (ay) as spoken
and recorded by seven different Vineyarders. 23 A study of the as-

22. The interviews were recorded at 3% inches per second on a Butoba MT-5, using

a Butoba MD-21 dynamic microphone. A tape recording of the standard reading, "After
the high winds ." read by five of the speakers whose formant measurements appear
. .

on Fig. 1.3, and other examples of centralized diphthongs used by Vineyard speakers
in natural conversation, may be obtained from the writer, Department of Linguistics,
University of Pennsylvania, Philadelphia, Pa. 19104.
23. Spectrograms were made on the Kay Sonograph, using both wide and narrow
bands. Seven of these, showing 15 instances of /ai/ and /au/, are reproduced in the
master's essay cited above.
The Social Motivation of a Sound Change 15

sembled formant patterns indicated that one particular point in time

might be best suited for measuring the degree of height of the first
element of the diphthong. This is shown in Fig. 1.2, as the point where
the first formant reaches a maximum. Measurements of the first and
second formant positions at this point seemed to correspond well
to the formant measurements for steady state vowels [a] to [9] in
Peterson and Barney's (1952) vowel studies. 24

[raid r b f d]

Fig. 1.2. Measurement of typical (ay) diphthongs at first formant

maximum.

The 80 measurements were then plotted on a bi-logarithmic scale,

with abscissa and ordinate corresponding to first and second for-
mants. The original impressionistic transcriptions were then entered
for each measurement, and the result examined for clear separation
of impressionistic levels. On the whole, the stratification was good:
the impressionistic ratings with more open first elements showed
higher first formant and lower second formant readings. However,
the separation of grades 2 from 3, and 4 from 5, was not as clear
as the others. A reduced four-step scale was then established, and
the resulting correlation shown in Fig. 1.3, and the table below. 25

24. The degree of overlap shown in Fig. 1.3 seems roughly comparable to Peterson

and Barney's results.

25. A parallel problem of condensing a finely graded impressionistic scale is dis-

cussed in Gauchat et al. 1925: ix. A seven-level transcription of the mid vowels was
reduced to five levels, but without the instrumental justification presented here.
16 SOCIOLINGUISTIC PATTERNS

2000 \
\

1900

1800

1700

1600

1500

1400

1300

1200

1100

900 800 700 600 500

__l

First formant frequency

Fig. 1.3. Correlation of instrumental measurement and impres-

sionistic ratings of centralization. Nos. 0-3: Scale
II equivalents

of impressionistic ratings of height of first elements of 86 (ay)

diphthongs, assigned before spectrographic measurement.
Seven different Martha's Vineyard speakers, males aged 14-60,
are represented. ° identifies speaker EP, age 31; + identifies
speaker GW, age 15.
The Social Motivation of a Sound Change 17

Fig. 1.3 shows the values for Scale II mapped on the bi-logarithmic
scale. This is a satisfactory result, with good separation of the four
grades of centralization. We have also obtained some justification
for the use of the first formant maximum in measuring spectrograms,
rather than the second formant minimum. Since the lines separating
the four grades parallel the second-formant axis more than the
first-formant axis, we have a graphic demonstration that our phonetic
impressions are more sensitive to shifts in the first formant than the
second.
When this display was originally planned, there was some question
as to whether it would be possible to map many different speakers
on the same graph. We know that there are significant differences
in individual frames of formant reference. Small children, for in-
stance, appear to have vowel triangles organized at considerably
higher frequencies than adults. The seven speakers whose readings
are displayed in Fig. 1.3 are all male; four are high school students,
aged 14 to 15. But the other three are adults, from 30 to 60 years
old, with widely different voice qualities.
Ideally, if we were studying the acoustic nature of the (ay) and
(aw) diphthongs, we would want a more uniform group of speakers.
Secondly, we would ask for better and more uniform recording
conditions: one recording was outdoors, two were in living rooms,
four in an empty conference room. However, since the object of the
testing was to lend objective confirmation to an impressionistic scale
of discrimination, it is only realistic to use a range of recordings as
varied as the body of material on which the entire study is based.
Absence of separation of the four grades in Fig. 1.3 might then have
indicated only defects in instrumental technique, but a positive result
can hardly be derived from such a bias.
It is interesting to note that measurements from no one speaker
18 SOCIOLINGUISTIC PATTERNS

are distributed over more than half of Fig. 1.3, and some speakers
are sharply limited to a narrow sector — still occupying portions of
all the grades of centralization. For instance, the highly centralized

speaker EP, aged 31, accounts for all of the readings in the lower
right portion marked with a ° sign: 0°, 2°, etc. He shows no readings
higher than 650 or 1500 cps. On the other hand, speaker D W, aged
15, also highly centralized, accounts for the upper left portion; his
readings, marked with a + sign, are all higher than 625 or 1550 cps.
Again, speaker GM, aged 15, is limited to a belt from lower left to
upper right, filling the space between the two just mentioned. Despite
the differences in vowel placement, these seven speakers utilize the
same dimension to produce the effect of centralized or open vowels:
widely separated formants for centralized vowels, adjacent formants
for open vowels. The opposition, though not distinctive, is clearly
seen as ranging from compact to (relatively) noncompact.
This display then indicates for us that the reduced impressionistic
scale shows good stratification in terms of physical parameters, and
we may proceed to employ such ratings with some confidence in
their validity.

The Linguistic Environment

We can now plot the distribution of centralized forms for each

speaker. This is done for each of the 69 interviews on a chart such
as is shown in Fig. 1.4. We find that these charts fall into three basic
types:

a. uncentralized norms: all words, or almost all, fall into Grade 0,

with at most only a few Grade l's in favored words such as right
and out.
b. centralized norms: most words with Grade 2, and only a few
Grade unfavored forms, such as time and cow.
l's for

c. phonetic conditioning: the influence of the phonetic environment

is reflected in a range of values from Grades to 2. Fig. 1.4 is
an example of this type.

Such phonetic conditioning is reminiscent of the phonetic regularity

found in the southern United States. (Shewmake 1927). But on
Martha's Vineyard, the distribution is more complex, and nowhere
codified with the precision to be found in the South. Before pro-
ceeding to chart the various social factors which influence this
 Grade 1 2 1

right out

night about
white trout

like house
sight
quite south
striped mouth
swiped couch
wife
life now
knife how

spider sound
side down

tide
applied round
characterized hound
Ivory ground
live

five

I've
(aw)-39
by
fly in

high
fryin
why
my
try

I'll

piles
while
mile
violence

shiners
kind
iodine
quinine
time
line

fired
tire

(ay)-75

Fig. 1.4.Phonetic determination of centralization. Centralization

chart for North Tisbury fisherman GB.
20 SOCIOLINGUISTIC PATTERNS

feature,we should consider the influence of the linguistic environ-

ment, and primarily phonetic conditioning.

Segmental Environment
The influence of the following consonant may be indicated by
tabulating five general articulatory dimensions:

Not favoring Favoring

centralization centralization

(a) sonorants zero final obstruents

(b) nasals orals
(c) voiced voiceless
(d) velars labials apicals
(e) fricatives stops

If we apply these oppositions in the order given, from (a) to (e), we

arrive at a consonant series from most favoring to least favorable
to centralization, which seems to conform quite well to the facts:

/t, s; p, f; d, v, z; k, 0, 5: 0: 1, r; n; m/ 26
The preceding consonant follows a rather different pattern, almost
the reverse, and has considerably less effect.The most favoring initial
consonants in centralized syllables are /h, 1, r, w, m, n/, with the
glottal stop allophone of zero heading the list. Thus the most favored
words are right, wife, night, light, nice, life, house, out.

Prosodic Factors
Stress regularly increases the degree of centralization for speakers
with type (b) and type (c) charts. This is not at all an obvious rule,
for the speech of many metropolitan areas shows the opposite tend-
ency: one may note an occasional centralized diphthong in rapid
reduced forms, but the same word under full stress is completely
uncentralized. This corresponds to the difference between a cen-
tralized occurrence and a centralized norm.
A typical case of centralization under stress occurs in this excerpt
from a story told by a North Tisbury fisherman:
Why I could do anything with this dog. I used to drop a [naif] or my
handkerchief or something, and I'd walk pretty near a quarter of a mile,

26. (ay) and (aw) are rare before /b, g, n, c, j/; /t/ includes [?]. The non-distinctive
[?] variant of zero onset also favors centralization heavily, as in the 1 forms of Fig.
1.3.
The Social Motivation of a Sound Change 21

and I'd stop and I'd turn to the dog: "You go get that! Where'd I lose that
[neif]!"

Stylistic Influence

While we find that most urban speakers have a variety of shifting

styles of speech, and that interviews under varying conditions will
produce varying counts of phonological features, this is not the case
with most Vineyarders. The majority are essentially single-style
speakers. Sometimes the conversation will take a livelier tone, or
a more formal aspect, but the percentage of centralized forms is not
significantly affected. Changes in centralization are apparently as-
pects of a pattern which develops over longer periods of time. 27

Lexical Considerations

A few special words are given greater centralization than their

phonetic form or prosodic position would usually account for. An
example is sliding, meaning coasting with a small sled. It may be
that confusion with an alternant form sledding is responsible, or that
words which originate in childhood, and are seldom spelled, are
more prone to centralization.

Distribution by Age and Time

The overall degree of centralization for each speaker is expressed
by the mean of the numerical values of the variants multiplied by
100. Thus for Fig. 1.4, the values are (ay)-75 and (aw)-39. We can then
find the values of the variable for any group of persons by averaging
the values for the members of the group.
We may first wish to see if centralization varies with the age level
of the speaker. Table 1.2 indicates that it does. Centralization of (ay)
and (aw) appear to show a regular increase in successive age levels,
reaching a peak in the 31 to 45 group. We must now consider the
reasons for assessing this pattern as evidence for an historical change
in the linguistic development of Martha's Vineyard. Is this an exam-
ple of sound change, or is it merely evidence for a regular change
in speaking patterns which is correlated with age?
At this point it is necessary to consider the general question as
to whether sound change can be directly observed. The well-known
27. One small stylistic influence which appeared was in the standard reading. Those
with centralized norms, whose charts were of type b and c, had slightly higher indexes
of centralization for reading than for conversation. The opposite effect was noted for
those with uncentralized norms.
22 SOCIOLINGUISTIC PATTERNS

TABLE 1.2.

CENTRALIZATION OF (ay)
AND (aw) BY AGE LEVEL
Age (ay) (aw)

75- 25 22
61-75 35 37
46-60 62 44
31-45 81 88
14-30 37 46

statement of Bloomfield (1933:347) seems to contradict this possi-

bility:

The process of linguistic change has never been directly observed; we

shall see that such observation, with our present facilities, is inconceivable.

When this opinion is viewed in the light of Bloomfteld's entire

discussion of phonetic change, appears to be strongly motivated
it

by arguments sound change. Bloomfield

for the absolute regularity of
wishes to show that such change is quite autonomous, "a gradual
favoring of some non-distinctive variants and a disfavoring of others,"
and quite distinct from the normal fluctuation of non-distinctive
forms, "at all times highly variable." Yet since direct observations
will always pick up this normal fluctuation, "even the most accurate
phonetic record of a language at any one time could not tell us which
phonemes were changing" (p. 365). The changes we do observe are
likely to be the effects of borrowing and analogic change.
Hockett (1958:439), while recognizing the possibility of divergent
views, has further refined the doctrine of imperceptible changes as
a basic mechanism of linguistic change. Movements of the center
of the normal distribution of random variations are, for all practical
purposes, not subject to direct observation, while the cruder forms
of change which are observed must be due to minor mechanisms.
Weinreich (1959) has pointed out the theoretical limitations of this
position; 28 here we may profitably examine the result of applying
such neo-grammarian thinking to empirical observations.
The prototype of close studies of sound change in a single com-

28. hard to feel comfortable with a theory which holds that the great changes
"It is

of the past were of one kind, theoretically mysterious and interesting, whereas
everything that is observable today is of another kind, transparent and (by implication)
of scant theoretical interest."
The Social Motivation of a Sound Change 23

munity is Gauchat's (1905) investigation of the patois of Charmey,

in French-speaking Switzerland. Gauchat observed and tabulated

differences in six phonological features in the speech of three gener-
ations: speakers over 60 years old, those between 30 and 60, and those
under 30 (see Chs. 7 and 9). Hermann returned to the scene in 1929,
one generation later, to investigate four of these features: his results
confirmed the interpretation of Gauchat's data as evidence for his-
torical change, since three of the four had advanced considerably
in the same direction. Yet Hermann (1929) also showed that real time
depth is essential for an accurate view, since the fourth feature had
not changed since 1903, and was apparently subject to a number of
conflicting influences.
The neo-grammarian viewpoint is that such observable shifts are
the results of a series of borrowings, imitations, and random varia-
tions. 29These complicated explanations could be applied without
contradiction to the present observations on Martha's Vineyard. But
we need not make the gratuitous assumption that sound change is
something else again, an ineluctable process of drift which is beyond
the scope of empirical studies. Here I would like to suggest that the
mixed pattern of uneven phonetic conditioning, shifting frequencies
of usage in various age levels, areas, and social groups, as we have
observed it on Martha's Vineyard, is the process of linguistic change
in the simplest form which deserves the name. Below this level, at
the point of individual variation, we have events which are sub-
linguistic in significance. At the first stage of change, where linguistic
changes originate, we may observe many sporadic side-effects of
articulatory processes which have no linguistic meaning: no socially
determined significance is attached to them, either in the differen-
tiation of morphemes, or in expressive function. Only when social
meaning is assigned to such variations will they be imitated and
begin to play a role in the language. Regularity is then to be found
in the end result of the process, as Sturtevant (1947:78-81) has argued,
and not in the beginning. 30

29. Such arguments were indeed advanced in some detail to explain Gauchat's
results,by P. G. Goidanich, "Saggio critico sullo studio de L. Gauchat," Archivio
Glottologico Italiano 20:60-71 (1926) [cited by Sommerfelt 1930]. As implausible as
Goidanich's arguments seem, they are quite consistent with Bloomfield's position cited
above.
30. See Hoenigswald 1963 for further considerations which support this view.
24 SOCIOUNGUISTIC PATTERNS

If we now accept the evidence we have on hand as adequate in

quantity, reliable, and valid,we must still decide if this particular
case is an example of a change in community habits of speech. Two
aspects of the question seem to make a good case for a positive
answer.
First, the records of the LANE show
only moderate centralization
of (ay) for the four informants of 1933, aged 56 to 82. It is impossible

to calibrate the Lowman transcription against our present scale,

especially since his data put more stress on short utterances with
stressed, elicited forms. But if we take the LANE symbol [e] as
equivalent to our present [e] of Grade 2, it appears that these speakers
had centralized norms for (ay) averaging about 86, as high as the
highest point reached in our sample for age level 60 to 90, but only
half as high as the highest point for age level 30 to 60. If we weigh
their performance against a matched group of present-day speakers,
we may conclude that there has been an intervening drop of central-
ization before the present rise.
Secondly, the question of (aw) is conclusive. The LANE informants
had an average rating of 06 for (aw): that is, for all practical purposes,
zero. The record shows a steady rise in centralization of (aw) which —
we have seen to be a completely new phenomenon in Martha's
—
Vineyard English reaching values of well over 100 for most old
family, up-island speakers, and going as high as 211 in one case. No
postulated change in speaking habits with age could account for this
rise.
The amount of centralization for the very old, and
fact that the
the very young speakers, is at a minimum, shows that the effect of
age cannot be discounted entirely, and it may indeed be a secondary
factor in this distribution over age levels.

Possible Explanations for a Rise in Centralization

So far, our discussion of centralization, the dependent variable

under study, has been merely descriptive. As we turn to the problem
of explanation, we are faced with the question of what independent
variables to examine. Certainly the structural parallelism of (ay) and
(aw) is significant here. 31 Let us assume for the moment that central-

31. We might wish to construct a rule here which would, in essence, convert
[+ compact] to [ — compact], simpler by one feature than a rule which would merely
The Social Motivation of a Sound Change 25

ization declined to a low point in the late 1930's, and then, after the
war, began to rise. At this point we find that a rising first element
of (ay) carries the first element of (aw) with it. Such a change in
direction would seem to give us a plausible explanation for the
parallelism being called into play at this time, rather than the as-
sumption that it suddenly began to operate after a three-hundred-
year hiatus.
There remains the prior question, that of explaining (or giving a
larger context for) the general rise of centralization on the island.
Why should Martha's Vineyard turn its back on the history of the
English language? I believe that we can find a specific explanation
if we study the detailed configuration of this sound change against

the social forces which affect the life of the island most deeply.
If we choose a purely psychological explanation, or one based only

on phonological paradigms, we have as much as said that social

variables such as occupation, income, education, social aspirations,
attitudes, are beside the point. We could only prove such a claim
by cross-tabulating the independent social variables, one at a time,
with the degree of centralization, and showing that any greater-
than-chance correlations are spurious.
However, our first attempts reveal some striking social correlations
which are not easily explained away. Table 1.3 shows us the geo-

TABLE 1.3.

GEOGRAPHICAL DISTRIBUTION
OF CENTRALIZATION

(ay) (aw)

Down-island 35 33
Edgartown 48 55
Oak Bluffs 33 10
Vineyard Haven 24 33
Up-island 67 66
Oak Bluffs 71 99
N. Tisbury 35 13
West Tisbury 51 51

Chilmark 100 81
Gay Head 51 81

convert [ai] to a centralized form. While such a statement is satisfying in its simplicity
and neatness, it should be clear from the following discussion that it would explain
only a small part of the mechanism of linguistic change.
26 SOCIOLINGUISTIC PATTERNS

graphical bias of centralization, favoring rural up-island against

small-town down-island areas. Table 1.4 shows the occupational
biases, with fishermen at the top and farmers at the bottom. If we
add to this the data of Table 1.5, showing the distribution by ethnic

TABLE 1.4.

CENTRALIZATION BY
OCCUPATIONAL GROUPS

(ay) (aw)

Fishermen 100 79
Farmers 32 22
Others 41 57

TABLE 1.5.

CENTRALIZATION BY ETHNIC GROUPS

English Portuguese Indian

Age level (ay)(aw) (ay)(aw) (ay)(aw)

Over 60 36 34 26 26 32 40
46 to 60 85 63 37 59 71 100
31 to 45 108 109 73 83 80 133
Under 30 35 31 34 52 47 88
All ages 67 60 42 54 56 90

groups, we embarrassed with too many explanations.

find ourselves
Are these social variables connected inany demonstrable way with
the linguistic change? Are they truly independent from one another,
or are some of the correlations spurious, the result of some depend-
ence on a larger factor which is logically prior to these? If such a
larger pattern exists, we must ask how it originated, and in what
way it is connected with the linguistic events. A simple-minded
bookkeeping approach will not answer such questions. We will have
to gain some insight into the social structure of the island, and the
pressures which motivate the social changes of present-day Martha's
Vineyard.
The Social Motivation of a Sound Change 27

The Interaction of Linguistic and Social Patterns 32

To understand Martha's Vineyard, we must first realize that this

is a very beautiful place, and a very desirable place to live. But it

is not an easy place to earn the kind of living which agrees well

with the achievement orientation of modern American society. The

1960 Census shows that it is the poorest of all Massachusetts counties:
it has the lowest average income, the highest number of poor people,

and the smallest number of rich people. 33 The Vineyard has the
highest rate of unemployment: 8.3 percent as against 4.2 percent for
the state, and it also has the highest rate of seasonal employment.
One might think that life on the island is nevertheless easier: perhaps
the cost of living is lower. Nothing could be further from the truth:
the high cost of ferrying is carried over to a higher price for most
consumer goods. As a result, there are more married women with
young children working than in any other county: 27.4 percent as
against 17.3 percent for the state as a whole.
The reason for this economic pressure, and the resulting depend-
ence on the tourist trade, is not hard to find. There is no industry
on Martha's Vineyard. The island reached its peak in the great days
of the whaling industry; for a time, commercial fishing in the local
waters buoyed up the economy, but the run of fish is no longer what
it used to be. Large-scale fishing is now out of New Bedford on the

Grand Banks. Farming and dairying have declined sharply because

32. The information given on Martha's

in the following discussion of social patterns
Vineyard was derived from conversations with the 69 informants. Even more
in part
significant, perhaps, was information gained from discussions with community leaders
who were in a position to view these patterns as a whole. I am particularly indebted
to Mr. Benjamin Morton, head of the Chamber of Commerce, Mr. Henry Beetle Hough,
editor of the Vineyard Gazette, and Mr. Charles Davis, superintendent of the Martha's
Vineyard Regional High School. Among my informants, I am especially grateful to
Mr. Donald Poole of Chilmark, Mr. Benjamin Mayhew, selectman of Chilmark, and
Mr. Albert Prada, town clerk of Edgartown.
33. Table 36 of the 1960 census report PC(1)— 23c, cited above in fn. 9, shows some

striking contrasts among Massachusetts counties. The median family income for the
Vineyard is $4,745, as against $6,272 for the state as a whole. Barnstable County (Cape
Cod) and Nantucket are also dependent on a vacation economy, yet they show median
incomes of $5,386 and $5,373. The most agricultural county in Massachusetts, Franklin,
shows a median of $5,455. The state as a whole has only 12.4 percent of families with
incomes under $3,000; the Vineyard has 23 percent. The state has 17.0 percent with
incomes over $10,000; the Vineyard has only 6.6 percent.
28 SOCIOLINGUISTIC PATTERNS

of the ferry rate, which raises the cost of fertilizer but lowers the
profit on milk.
The 1960 Census shows us that the island's labor force of 2,000
souls is heavily occupied with service trades. Only 4 percent are in
manufacturing, one-seventh of the state average. Five percent are
in agriculture, 2.5 percent in fishing, and 17 percent in construction;
these percentages are five, ten, and three times as high as those for
the state as a whole. 34
These economic pressures must be clearly delineated in order to
assess the heavy psychological pressures operating on the Vine-
yarders of old family stock. Increasing dependence on the summer
trade acts as a threat to their personal independence. The more
far-seeing Vineyarders can envisage the day when they and their
kind will be expropriated as surely as the Indians before them. They
understand that the vacation business cannot help but unbalance
the economy, which produces far too little for the summer trade,
but far too much for the winter. Yet it is very hard for the Vineyarder
not to reach for the dollar that is lying on the table, as much as he
may disapprove of it. We have already noted that many Vineyarders
move out of their own homes to make room for summer people.
Those who feel that they truly own this island, the descendants
of the old families, have a hard time holding on. Summer people,
who have earned big money in big cities, are buying up the island.
As one Chilmarker said, "You can cross the island from one end to
the other without stepping on anything but No Trespassing signs."
The entire northwest shore has fallen to the outsiders. In Edgartown,
the entire row of spacious white houses on the waterfront has capit-
ulated to high prices, with only one exception, and the descendants
of the whaling captains who built them have retreated to the hills
and hollows of the interior.
This gradual transition to dependence on, and outright ownership
by the summer people has produced reactions varying from a fiercely
defensive contempt for outsiders to enthusiastic plans for furthering
the tourist economy. A study of the data shows that high central-
ization of (ay) and (aw) is closely correlated with expressions of
strong resistance to the incursions of the summer people.
The greatest resistance to these outsiders is felt in the rural up-
island areas, and especially in Chilmark, the only place where fishing

34. See Table 82 of the 1960 census report, as in fn. 33.

The Social Motivation of a Sound Change 29

is still economy. 35 Chilmarkers are the most

a major part of the
different, independent, the most stubborn defenders of their own way
of living. In order to assess the changing orientation of island groups
towards the old family tradition, I included in my interview a battery
of questions dealing with the semantics of the word Yankee. One
question read: "Where on the island would a typical old Yankee be
most apt to live?" By far the most common answer was "Chilmark."
Chilmarkers were named most often as examples of "typical old
Yankees."
Chilmarkers pride themselves on their differences from main-
landers:

You people who come down here Martha's Vineyard don't understand
to
the background of the old families of the island strictly a maritime . . .

background and tradition and what we're interested in, the rest of
. . .

America, this part over here across the water that belongs to you and we
don't have anything to do with, has forgotten all about. . .

I think perhaps we use entirely different type of English language . . .

. think differently here on the island

. . it's almost a separate language
. . .

within the English language.

To a large extent, this last statement is wishful thinking. Much of

the language difference depended upon whaling terms which are
now obsolete. It is not unnatural, then, to find phonetic differences
becoming stronger and stronger as the group fights to maintain its
identity. We have mentioned earlier that the degrees of retroflexion
in final and preconsonantal /r/ have social significance: at Chilmark,
retroflexion is at its strongest, and is steadily increasing among the
younger boys.
In Table 1.3, we note that centralization is higher up-island than
down-island, and highest of all in Chilmark. In Table 1.4, we note
that of all occupational groups, fishermen show the highest central-
ization. Our total number of cases is too small to allow extensive
cross-tabulations, but if we take the group of Chilmark fishermen

35. Despite the low number of Vineyarders listed as fishermen by occupation in

the Census, a much larger number of islanders relyupon part-time fishing to supple-
ment their income. In particular, harvesting bay scallops in the salt ponds is a prized
source of revenue in the summer months. A great deal of local legislation is designed
to protect the professional fishermen from the great number of part-time scallopers
taking in too large a share. Much discussion and considerable bitterness develops as
a result of this conflict of interest, in which the truly professional Chilmarkers are,
psychologically at least, on top.
30 SOCIOLINGUISTIC PATTERNS

in the middle age level, from 30 to 60, we find that these five in-
formants have average indexes of 148 for (ay) and 118 for (aw), higher
than any other social group which we might select on the island.
Conversely, let us list the six speakers with the highest degree of
—
centralization in order of (ay) that is, the upper 10 percent:

{ay) (aw)

Chilmark fisherman, age 60 170 111

Chilmark fisherman, age 31 165 211
Chilmark fisherman, age 55 150 124
Edgartown fisherman, age 61 143 107
Chilmark fisherman, age 33 133 79
Edgartown fisherman, age 52 131 131

It should be noted here that the two Edgartown fishermen listed are
brothers, the last descendants of the old families to maintain their
position on the Edgartown waterfront in the face of the incroachment
of summer people.
We have now established within reason that the strong upturn in
centralization began up-island, among Chilmark fishermen, under the
same influence which produced parallel results among the few
Edgartown residents who shared their social orientation.
Table 1.5 shows the developments by age level for each of the three
main ethnic groups. All of the examples we have used so far deal
with the English group of old family descent; in Chilmark, this is
the only group of any size. Let us continue to follow the development
of this group through the succeeding age levels, and examine the
interaction of social and linguistic patterns.
We see that centralization reaches a peak in the age level from
30 to 45, and that centralization of (aw) has reached or surpassed
This age group has been under heavier stress than
(ay) at this point.
any other; the men have grown up in a declining economy, after
making a more or less deliberate choice to remain on the island
rather than leave it. Most of them have been in the armed forces
during World War II or in the Korean conflict. Many have been to
college, for the English-descent group has a strong bent towards
higher education. At some point, each of these men elected to make
a smaller living on Martha's Vineyard, while many of their contem-
poraries left to gain more money or more recognition elsewhere.
Severe strains are created in those who are pulled in both direc-
tions; the traditional orientation of Martha's Vineyard has long been
The Social Motivation of a Sound Change 31

inward and possessive, yet the pull of modern achievement-oriented

America is even greater for some.

I think actually a very hard thing to make that decision ... It comes
it's

to you later, that you should have made it before. I have another son
—
Richard is an aeronautical engineer. He really loves the island. And when
he decided to be an aeronautical engineer we discussed it at length and — —
I told him at that time: you just can't live on Martha's Vineyard He . . .

works at Grumman, but he comes home every chance he gets and stays
just as long as he can.

The speaker is a woman of 55, a descendant of the Mayhew family,

who business school in Boston, and returned to the island to
left
become a real estate agent. Her son made the opposite choice; but
another family, of long standing in Chilmark, had this to report about
their son:

. . . we had an away to school, but he really didn't want

idea that he'd go
to go away . . . When
he was at Chauncey Hall, they tried to get him to
go to M.I.T.; but he said no, he didn't want to go anywhere where he had
to learn to do something that he couldn't come back to this island.

We can learn a great deal about centralization by studying such

histories of particular families. The two speakers who head the list
of centralized speakers on the previous page are father and son. The
father, a Chilmark lobsterman, is a thoughtful, well-read man with
a passionate concern with the history of the whaling industry; he
is perhaps the most eloquent spokesman for the older Vineyard

tradition, and the author of the quotation on p. 29. His son is a college
graduate who tried city life, didn't care for it, came back to the island
and built up several successful commercial enterprises on the
Chilmark docks. He shows a high (ay) at 211, considerably more
centralized than anyone else I have heard at Chilmark. One evening,
as I was having dinner at his parents' house, the conversation turned
to speech in general, without any specific reference to (ay) or (aw).
His mother remarked, "You know, E. didn't always speak that way
. .it's only since he came back from college. I guess he wanted to
.

be more like the men on the docks ..."

Here we see a clear case of hypercorrection at work, and from
other evidence as well, it is reasonable to assume that this is a very
regular force in implementing the phonetic trend we are studying.
When we come to high-school students, we must realize that many
32 SOCIOLINGUISTIC PATTERNS

of the young people from the old-family group do not intend to

remain on the island, and this is reflected in the lower average index
of Table 1.5. Comparatively few of the sons of the English-descent
group will be earning their living on the Vineyard in the next 20
years. In a series of interviews in Martha's Vineyard Regional High
School, it was possible to compare speaking habits very closely by
means of the standard reading, "After the high winds ..." A marked
contrast was observed between those who plan to leave the island
and those who do not. The latter show strong centralization, while
the former show little, if any. To highlight this point, we may take
four 15-year-old students: the two down-islanders who intend to
leave for careers in business and finance show little or no central-
ization; the two up-islanders who hope to go to college and return
to make their living on the island show considerable centralization. 36
The indexes speak for themselves:
Down-island, leaving Up-island, staying

(ay)(aw) (ay)(aw)
00-40 90-100
00-00 113-119

One of the down-islanders, from Edgartown, has fallen very much

under the influence of the upper-class Bostonian summer visitors.
He has lost all constriction in postvocalic /r/, and has a fronted low
center vowel as well in such words as [ka:], 'car'.

Centralization among Other Ethnic Groups

We can now turn to the special position of the Portuguese and

Indian ethnic groups, and see if the same approach can account for
the distribution of centralized forms among them.
The most common view of the early Portuguese immigration is
that the settlers came from an island with a very similar economy,
shared the Yankee virtues of thrift and industry, and fitted into the
island life almost perfectly. The Azoreans who came first seemed
to have a strong inclination for farming and fishing, rather than
factory work; in the Vineyard's rather diffuse economy, there was

36. On the question of leaving the island, one of these boys said: "... I can't see
myself off island somewhere ... I like it a lot here, like my father goes lobstering.
That's quite a bit of fun ... as long as I get enough money to live and enjoy myself.
I was figuring on . . . going into oceanography because you'd be outdoors: it wouldn't
be office work."
The Social Motivation of a Sound Change 33

concentration of the Portuguese into the kinds of industrial

little

pockets we find on the mainland. 37 Even among the tough-minded

Chilmarkers, we find a certain grudging acknowledgement of the
Yankee-like orientation of the Portuguese:

. . . they worked, that's why they were respected. Nobody ever particularly
interfered with 'em. You hear somebody make a remark about the dumb
Portagee or something, but actually I think they've been pretty well re-
spected because they mind their own business pretty well. They didn't ask
for anything.

It took some time, however, for the Portuguese-descent group to

make its way into the main stream of island life. Intermarriage of

Portuguese and Yankee stock occurs, but it is rare. Second-generation

Portuguese certainly do not feel at home in every situation: as some
Vineyarders put it, these Portuguese have "a defensive attitude." A
member of the English group will as a rule speak his mind freely,
condemning the summer people and his neighbors with equal frank-
ness. But the second-generation Portuguese never criticizes the sum-
mer people in the interview situation, and he is extremely wary of
criticizing anyone. When the word Yankee is introduced, he shifts
uneasily in his chair, and refuses to make any comment at all.
While the speech of the Portuguese second generation is free of
any detectable Portuguese influence, 38 it is also lacking the special
Vineyard flavor. If we examine the Portuguese age groups over 45
in Table 1.5, which contain a large proportion of second-generation
speakers, we find little or no centralization.
This is not the case with third- and fourth-generation Portuguese
speakers. In this group, we find centralization very much on the
increase, particularly with (aw). In Table 1.5, we see that the age
group from 31 to 45 has a very high degree of centralization. This
age level contains a great many third-generation Portuguese. It is the
first Portuguese group which has entered the main stream of island
life, occupying positions as merchants, municipal officers, and many

other places of secondary leadership. These speakers consider

37. In many ways, the Vineyard seems to be more democratic than the mainland.

I have heard on the mainland strong expressions of hostility between Portuguese

groups from the Azores and those from the Cape Verde Islands, but never on Martha's
Vineyard.
38. On the other hand, I have heard a strong Portuguese accent from a second-
generation Portuguese man, about 40 years old, who was raised on a farm near
Taunton, Mass.
34 SOCIOLINGUISTIC PATTERNS

themselves natives of the island, and in response to the term Yankee,

they either include themselves, or make fun of the whole idea.
In the youngest age level, the Portuguese-descent group shows a
very regular use of centralization, whether second or third or fourth
generation, and their average centralization index in the table is, at
this point, higher than the English group.
One might think that centralization might be on the way to be-
coming a marker of the ethnic Portuguese on the island, if such a
trend continues. But this possibility runs counter to the strongly
democratic nature of present-day Vineyard society. Among high-
school students, for example, there appear to be no social barriers
between the ethnic groups, in clubs, at dances, and between friends.
This situation is some former mainlanders,
especially shocking to
who would like to draw a color line against some of the children
with Cape Verde backgrounds. But despite a few such counter-
currents, the unifying, protective nature of Vineyard society shields
the island native from the kind of reality which is practised on the
outside. 39
The reason that the youngest Portuguese group shows higher
centralization is that a larger percentage identify themselves with
the island and the island way of life, than is the case among the
English-descent group. Whereas almost all of the English group leave
the island to go to college, and few return, almost all of the Portu-
guese group remain. As a result, they are gradually supplanting the
English group in the economic life of the island.
It is fair enough to say that the main problem of the Portuguese

group has not been to resist the incursions of the summer people
but rather to assert their status as native Vineyarders. Their chief
obstacle has not been the outsiders, but rather the resistance to full
recognition from the English-descent group. With full participation
in native status has come full use of the special characteristics of
Martha's Vineyard English, including centralized diphthongs.

The Indian descent group is relatively small and homogeneous. The

hundred citizens of Gay Head are united in a few closely related
families. One would think that these survivors of the aboriginal

39. Vineyard youngsters have received rather severe shocks on

In several cases,
work in an area where caste restrictions
leaving the island for the armed services or for
were in force. One boy was put into a black regiment on entering the service, though
action from Vineyard leaders had him transferred soon afterwards.
The Social Motivation of a Sound Change 35

Wampanoag Indians would have had little trouble in asserting their

native status. On the contrary, a long tradition of denigration of the
Indian has served, for over a hundred years, to rob him of the dignity
which should accompany this feat of survival. The issue revolves
around the fact that the declining Indian community has necessarily
intermarried with outsiders over the past ten generations. The logic
of American society dictated that these outsiders should be black.
Thus as early as 1764, the Yankee officials of the Vineyard claimed
that only one quarter of the Indians were "of pure blood." 40 In 1870,
the Governor of Massachusetts took away the reservation status of
Gay Head, on the ground that they really weren't Indians at all, and
handed them over to the political ministrations of Chilmark.
For many decades, the Indians were literally second-class citizens,
and the resentment dating from this period is not entirely gone. On
the other hand, we find that a number of Vineyarders, of both English
and Portuguese descent, regard the Indians with a mixture of sarcasm
and scepticism:
. . . show me a Gay Head Indian and I'll like to see one.

The Indian people are aware of this situation, as shown in this

quotation from one of the Indian informants, a woman of 69:

These island folks, they don't want to mix at all, up this end. . . They
don't like to give the Indian his name, here on the island. I'll tell you that.
They like to be dirty with some of their talk.

Despite the great shift in Vineyard ideology over the past three
generations, the Indians still feel blocked, geographically and so-

cially, by the Chilmarkers, "up this end." Their attitude toward the
Chilmarkers is ambiguous: on the one hand, they resent the Chil-
markers' possessive attitude toward the island, and the traditional
hard-fisted, stiff-necked Yankee line. Their reaction to the word
Yankee is sarcastic and hostile. 41 But their main complaint is that
they deserve equal status, and whether they will admit it or not, they
would like to be just like the Chilmarkers in many ways.

40. A very rich vein of information on this score may be tapped from Richard L.

Pease's Report of the commissioner appointed to complete the examination of . . .

all boundary lines ... at Gay Head (Boston, 1871). Pease was acting essentially as

the hatchet man for the Governor of Massachusetts, to whom he was reporting.
41. "Where they come from — down south somewhere? . . . come from
Lot of 'em
Jerusalem, you know . .
."
36 SOCIOLINGUISTIC PATTERNS

As far as centralization is concerned, Table 1.5 indicates that the

Indians follow close behind the Chilmarkers. At the same time, they
show a greater relative increase of centralization of (aw), similar to
the Portuguese development, especially among the young people.
Here there are signs of an additional phonetic feature, shared by both
Portuguese and Indians: a backed form of (aw), which may be written
[au]. It is characteristic of five speakers in the sample, all under 30,
all fairly low in socioeconomic status. Whether it represents a gen-

eral trend cannot be determined at this point.

We may note that there has been a revival of Indian culture in
the form of pageants staged for the tourist trade, beadwork, and other
Indian crafts, and with these a revived emphasis on tribal orga-
nization. The younger Indians acknowledge that this revival was
commercially motivated in its beginnings, but they claim that it is
now more than that, and that Indian culture would survive if the
vacationers disappeared entirely. The Indian language has been dead
for several generations, however, and the ritual formulas must be
learned from a book. The Indians are truly traditional speakers of
English, and their claim to native status must be expressed in that
language.

The Social Meaning of Centralization

From the information we now have at hand, there readily emerges

the outline of a unifying pattern which expresses the social signifi-
cance of the centralized diphthongs.
It is apparent that the immediate meaning of this phonetic feature
is 'Vineyarder.' When a man says [reit] or [heus], he is uncon-
sciously establishing the fact that he belongs to the island: that he
is one of the natives to whom the island really belongs. In this
respect, centralization is not different from any of the other sub-
phonemic features of other regions which are noted for their local
dialect. The problem is, why did this feature develop in such a
complicated pattern on the Vineyard, and why is it becoming
stronger in the younger age levels?
The answer appears to be that different groups have had to respond
to different challenges to their native status. And in the past two
generations, the challenges have become much sharper through
severe economic and social pressures.
The old-family group of English descent has been subjected to
The Social Motivation of a Sound Change 37

pressure from the outside: its members are struggling to maintain

their independent position in the face of a long-range decline in the
economy and the steady encroachment of the summer people. The
member of the tradition-oriented community naturally looks to past
generations for his values: these past generations form a reference
group for him. 42 The great figures of the past are continually referred
to, and those who have died only a few years ago have already

assumed heroic stature. "If you could only have been here a few
years ago and talked to N. He could have told you so many things!"
The sudden increase in centralization began among the Chilmark
fishermen, the most close-knit group on the island, the most inde-
pendent, the group which is most stubbornly opposed to the incur-
sions of the summer people. There is an inherently dramatic
character to the fisherman's situation, and a great capacity
for self-dramatization in the fisherman himself, which makes him
an ideal candidate to initiate new styles in speech. In the early
morning, the curtain rises: a solitary figure appears upon the scene.
For the course of an entire day, this single actor holds the stage. Then
at last, the boat docks; the curtain descends. The play is over, yet
the reviews will be read and reread for generations to come.
I can remember as a boy, when I first started going to sea with my father,

he said to me: remember two things. Always treat the ocean with respect,
and remember you only have to make one mistake, never to come back.

Centralized speech forms are then a part of the dramatized island

character which the Chilmarker assumes, in which he imitates a
similar but weaker tendency in the older generation.
For younger members of the English-descent group, we can view
the mechanism in greater detail. For them, the old-timers and the
up-islanders in particular serve as a reference group. They recognize
that the Chilmark fishermen are independent, skillful with many
kinds of tools and equipment, quick-spoken, courageous, and physi-
cally strong. Most importantly, they carry with them the ever-present
conviction that the island belongs to them. If someone intends to
stay on the island, this model will be ever present to his mind. If
he intends to leave, he will adopt a mainland reference group, and
the influence of the old-timers will be considerably less. The differ-

42. In the technical sense developed by R. Merton, Social Theory and Social
Structure (Glencoe, III, 1957).
38 SOCIOLINGUISTIC PATTERNS

ential effect in the degree of centralization used is a direct result

of this opposition of values.
The Portuguese group is not faced with a dilemma of going or
staying.The main challenge to which this group has responded is
from the English group, which has certainly served as a reference
group for the Portuguese until very recent times. As the number of
Portuguese in prominent positions grows, it is no longer urgent to
minimize the effects of being Portuguese, but rather to assert one's
identity as an islander.
The Gay Head developments are dictated by the antinomy of
values which reigns there. On the one hand, the Indian group resents
any bar to full participation in the island life, and the Indians have
plainly adopted many of the same values as the Chilmarkers. But
on the other hand, they would like to insist as well on their Indian
identity. Unfortunately, they no longer have linguistic resources for
this purpose, and whether they like it or not, they will follow the
Chilmark lead.
The role of the Chilmarker, or "old-time typical Yankee" has
declined as the reference group which governs the meaning of
"islander" has shifted away from that which governs "Yankee." Even
among the Chilmarkers, the more far-sighted members of the com-
munity recognize that the term Yankee no longer fits the island.
Whereas word may still be a rallying cry in some parts of New
this
England, has outlived its usefulness on Martha's Vineyard. In
it

emphasizing descent status rather than native status, Yankee sum-

mons up invidious distinctions which are no longer good currency
on the island.

People don't make so much about it as they used to when I was young.
People would make that statement: "I'm a Yankee! I'm a Yankee!" But now
—
you very seldom mostly, read it in print. 43

In summary, we can then say that the meaning of centralization,

judging from the context in which it occurs, is positive orientation
towards Martha's Vineyard. If we now overlook age level, occupa-
tion, ethnic group, geography, and study the relationship of central-
ization to this one independent variable, we can confirm or reject
this conclusion. An examination of the total interview for each

43. The speaker is one of the Mayhews, a retired Chilmark fisherman, who has as

much claim to be a "typical old Yankee" as any person on Martha's Vineyard.

The Social Motivation of a Sound Change 39

informant allows us to place him in one of three categories: posi-

tive —
expresses definitely positive feelings towards Martha's Vine-
—
yard; neutral expresses neither positive nor negative feelings to-
—
wards Martha's Vineyard; negative indicates desire to live
elsewhere. When these three groups are rated for mean centralization
indexes, we obtain the striking result of Table 1.6.

TABLE 1.6.

CENTRALIZATION AND ORIENTATION

TOWARDS MARTHA'S VINEYARD
Persons (ay) (aw)

40 Positive 63 62
19 Neutral 32 42
6 Negative 09 08

The fact that this table shows us the sharpest example of stratifica-
tion we have yet seen, indicates that we have come reasonably close
to a valid explanation of the social distribution of centralized diph-
thongs.

The Intersection of Social and Linguistic Structures

The following abstract scheme may serve to summarize the argu-

ment which has been advanced so far to explain the spread and
propagation of this particular linguistic change.

1. A language feature used by a group A is marked by contrast with

another standard dialect.
2. Group A is adopted as a reference group by group B, and the
feature is adopted and exaggerated as a sign of social identity in
response to pressure from outside forces.
3. Hypercorrection under increased pressure, in combination with
the force of structural symmetry, leads to a generalization of the
feature in other linguistic units of group B.
4. A new norm is established as the process of generalization levels
off.

5. The new norm is adopted by neighboring and succeeding groups

for whom group B serves as a reference group.

There remains a gap in the logic of the explanation: in what way

do social pressures and social attitudes come to bear upon linguistic
40 SOCIOLINGUISTIC PATTERNS

structures? So far we have assembled

a convincing series of corre-
lations: yet we
propose a rational mechanism by which
still need to
the deep-seated elements of structure enter such correlations.
It has been noted that centralized diphthongs are not salient in

the consciousness of Vineyard speakers. They can hardly therefore

be the direct objects of social affect. The key to the problem may
lie in the fact that centralization is only one of many phonological

features which show the same general distribution, though none may
be as striking or as well stratified as (ay) and (aw). There are no less
than 14 phonological variables which follow the general rule that
the higher, or more constricted variants are characteristic of the
up-island, "native" speakers, while the lower, more open variants
are characteristic of down-island speakers under mainland influ-
ence. 44 We can reasonably assume that
"close-mouthed" articu- this
latory style is may well be that social
the object of social affect. It

evaluation interacts with linguistic structures at this point, through

the constriction of several dimensions of phonological space. Partic-
ular linguistic variables would then be variously affected by the
overall tendency towards a favored articulatory posture, under the
influence of the social forces which we have been studying. Evidence
for such an hypothesis must come from the study of many compara-
ble developments, in a variety of English dialects and other lan-
guages. It is enough to note here that it is a plausible mechanism
for sociolinguistic interaction which is compatible with the evidence
which has been gathered in this investigation.

Limitations of This Study

We noted earlier that one limitation of this study stems from the
fact that the variable selected not salient. This limitation, coupled
is

with the small size of the Vineyard population, made it impractical

to explore thoroughly the subjective response of native speakers to
centralized diphthongs. Other shortcomings of the technique used

44. In the following list of the variables in question, the up-island form is given
first. Phonemic inventory: /o/~/ou/ whole
in road, Phonemic toad, boat, . . .

distribution: /e/ only before intersyllabic /r/ instead of both /e/ and /ae/; /r/~/a/
in postvocalic position. Phonemic incidence: /i~e/ in get, forget, when, anyway, can
. . /e~ae/ in have, had, that; /a— a/ in got. Phonetic realization: [ei~ai] and
. ;

[eu~au]: [r~^]; [ir~ar] in work, person . . .

; [9~a] in furrow, hurry . . .
; [oeu~~ou]
a 9
in go, no ... ; [ii~ii] and [uu~uu]; [i ~i] and [e ~e].
The Social Motivation of a Sound Change 41

on Martha's Vineyard may be seen in the sampling method, which

was far from rigorous. 45 The statements made about developments
through various age levels among the Portuguese and Indians are
based on an inadequate number of cases. The sample is particularly
weak in the down-island area, especially in Oak Bluffs, and the
picture of down-island trends is correspondingly weaker than up-
island developments. Finally, it may be noted that the interviewing
technique was not as firmly controlled as it might have been: a
number of changes in the interview structure were made as the study
progressed.
With these reservations, we can say that the findings give good
confirmation of the main theme of the study: the correlation of social
patterns with the distributional pattern of one linguistic variable. 46
The reliability of the index used was tested in several cases where
the same informant was interviewed twice, with good results. 47
Indexes for reading style did not diverge sharply from other portions
of the interview.The validity of the scale of measurement was well
established by instrumental methods, and the validity of the whole
seems to be reinforced by the unitary nature of the final inter-
pretation.

The techniques developed on Martha's Vineyard were later refined

and applied to a much more complex situation in the urban core
of New Here multiple-style speakers are the rule, not the
York City.
exception; instead of three ethnic groups we have a great many;
mobility and change are far more rapid; and the population is huge.
Here the sampling requirements must be far more rigid; and the
techniques used to assess the social meaning of linguistic cues must

The problem of sampling technique for linguistic variables is a difficult one at

45.
the moment. While we are sure that linguistic behavior is more general than the
behavior usually traced by survey methods, we do not know how much more general
it is, nor can we estimate easily how far we may relax the sampling requirements,
if at all.
46. In addition to the positive correlations discussed above, the explanation given
is reinforced by certain negative results of alternate explanations. The educational
level of the informants is not correlated significantly with degree of centralization.
The distribution of substandard or archaic grammar does not correspond to the
distribution of centralized forms.
47. For example, two interviews with Ernest Mayhew, Chilmark fisherman, age 83,
showed these results: first interview, (ay)-67, (aw)-58; second interview, (ay)-59, (aw)-40.
The count for (aw) is based on about one-third as many items as for (ay).
42 SOCIOLINGUISTIC PATTERNS

be more subtle and complex. Yet the basic approach, of isolating

the socially significant variables, and correlating them with the
patterns of general social forces, was the same as that which was
used on Martha's Vineyard. We can expect that these methods will
give us further insight into the mechanism of linguistic change.
 The Social Stratification
2 of (r) in New York City
Department Stores
"As this letter is but a jar of the tongue, ... it is

the most imperfect of all the consonants."

John Walker,
Principles of English Pronunciation. 1791

ANYONE who begins to study language in its social context

immediately encounters the classic methodological problem: the
means used to gather the data interfere with the data to be gathered.
The primary means of obtaining a large body of reliable data on
the speech of one person is the individual tape-recorded interview.
—
Interview speech is formal speech not by any absolute measure,
but by comparison with the vernacular of everyday life. On the
—
whole, the interview is public speech monitored and controlled in
response to the presence of an outside observer. But even within
that definition, the investigator may wonder if the responses in a
tape-recorded interview are not a special product of the interaction
between the interviewer and the subject. One way of controlling for
this is to study the subject in his own natural social context
interacting with his family or peer group (Labov, Cohen, Robins, and
Lewis 1968). Another way is to observe the public use of language
in everyday life apart from any interview situation to see how —
people use language in context when there is no explicit observation.
This chapter is an account of the systematic use of rapid and anony-
mous observations in a study of the sociolinguistic structure of the
speech community. 1

1. This chapter is based upon Chs. 3 and 9 of The Social Stratification of English
in New York City (1966), revised in the light of further work with rapid and anonymous
observations. I am indebted to Frank Anshen and Marvin Maverick Harris for refer-
ence to illuminating replications of this study (Allen 1968, Harris 1968).

43
44 SOCIOLINGUISTIC PATTERNS

This chapter is the first of a series of six which deal primarily

with the sociolinguistic study of New York City. The main base for
that study (Labov 1966a) was a secondary random sample of the
Lower East Side, and this data will be considered in the following
chapters. But before the systematic study was carried out, there was
an extensive series of preliminary investigations. These included 70
individual interviews and a great many anonymous observations in
public places. These preliminary studies led to the definition of the
major phonological variables which were to be studied, including
(r): the presence or absence of consonantal [r] in postvocalic position

in car, card, four, fourth, etc. This particular variableappeared to

be extraordinarily sensitive toany measure of social or stylistic
stratification. On the basis of the exploratory interviews, it seemed
possible to carry out an empirical test of two general notions: first,
that the linguistic variable (r) is a social differentiator in all levels
of New York City speech, and second, that rapid and anonymous
speech events could be used as the basis for a systematic study of
language. The study of (r) in New York City department stores which
I will report here was conducted in November 1962 as a test of these

ideas.
We can hardly consider the social distribution of language in New
York City without encountering the pattern of social stratification
which pervades the life of the city. This concept is analyzed in some
detail in the major study of the Lower East Side; here we may briefly
consider the definition given by Bernard Barber: social stratification
is the product of social differentiation and social evaluation
(1957:1-3). The use of this term does not imply any specific type of
class or caste, but simply that the normal workings of society have
produced systematic differences between certain institutions or
people, and that these differentiated forms have been ranked in status
or prestige by general agreement.
We begin with the general hypothesis suggested by exploratory
interviews: if any two subgroups of New York City speakers are
ranked in a scale of social stratification, then they will he ranked
in the same order by their differential use of (r).
It would be easy to test this hypothesis by comparing occupational

groups, which are among the most important indexes of social

stratification. We could, for example, take a group of lawyers, a group
of file clerks, and a group of janitors. But this would hardly go
beyond the indications of the exploratory interviews, and such an
The Social Stratification of (r) 45

extreme example of differentiation would not provide a very exact-

ing test of the hypothesis. It should be possible to show that the
hypothesis is so general, and the differential use of (r) pervades New
York City so thoroughly, that fine social differences will be reflected
in the index as well as gross ones.
It therefore seemed best to construct a very severe test by finding

a subtle case of stratification within a single occupational group: in

people of large department stores in Manhattan.
this case, the sales
If we department stores, from the top, middle, and
select three large
bottom of the price and fashion scale, we can expect that the cus-
tomers will be socially stratified. Would we expect the sales people
to show a comparable stratification? Such a position would depend
upon two correlations: between the status ranking of the stores and
the ranking of parallel jobs in the three stores; and between the jobs
and the behavior of the persons who hold those jobs. These are not
unreasonable assumptions. C. Wright Mills points out that salesgirls
in large department stores tend to borrow prestige from their cus-
tomers, or at least make an effort in that direction. 2 It appears that
a person's own occupation is more closely correlated with his lin-
guistic behavior —
for those working actively than any other single —
social characteristic. The evidence presented here indicates that the
stores are objectively differentiated in a fixed order, and that jobs
in these stores are evaluated by employees in that order. Since the
product of social differentiation and evaluation, no matter how
minor, is social stratification of the employees in the three stores,
the hypothesis will predict the following result: salespeople in the
highest-ranked store will have the highest values of (r); those in the
middle-ranked store will have intermediate values of (r); and those
in the lowest-ranked store will show the lowest values. If this result
holds true, the hypothesis will have received confirmation in pro-
portion to the severity of the test.

The three stores which were selected are Saks Fifth Avenue,

2. C. Wright Mills, White Collar (New York: Oxford University Press, 1956), p. 173.

See also p. 243: "The tendency of white-collar people to borrow status from higher
elements is so strong that it has carried over to all social contacts and features of
the work-place. Salespeople in department stores . . . frequently attempt, although
often unsuccessfully, to borrow prestige from their contact with customers, and to
cash it in among work colleagues as well as friends off the job. In the big city the
who works on 34th Street cannot successfully
girl claim as much prestige as the one
who works on Fifth Avenue or 57th Street."
46 SOCIOLINGUISTIC PATTERNS

Macy's, and S. Klein. The differential ranking of these stores may

be illustrated in many ways. Their locations are one important point:
Highest-ranking: Saks Fifth Avenue
and 5th Ave., near the center of the high fashion shopping
at 50th St.
district, along with other high-prestige stores such as Bonwit Teller, Henri
Bendel, Lord and Taylor
Middle-ranking: Macy's
Herald Square, 34th St. and Sixth Ave., near the garment district, along
with Gimbels and Saks-34th St., other middle-range stores in price and
prestige.
Lowest-ranking: S. Klein
Union Square, 14th St. and Broadway, not far from the Lower East Side.

The advertising and price policies of the stores are very clearly
stratified.Perhaps no other element of class behavior is so sharply
differentiated in New York City as that of the newspaper which
people read; many surveys have shown that the Daily News is the
paper read first and foremost by working-class people, while the New
York Times draws its readership from the middle class. 3 These two
newspapers were examined for the advertising copy in October
24-27, 1962: Saks and Macy's advertised in the New York Times,
where Kleins was represented only by a very small item; in the News,
however, Saks does not appear at all, while both Macy's and Kleins
are heavy advertisers.

No. of pages of advertising

October 24-27, 1962

NY Times Daily News

Saks 2
Macy's 6 15
S. Klein y4 10

We may also consider the prices of the goods advertised during

those four days. Since Saks usually does not list prices, we can only

compare prices for all three stores on one item: women's coats. Saks:
$90.00, Macy's: $79.95, Kleins: $23.00. On four items, we can compare
Kleins and Macy's:

3. This statement is fully confirmed by answers to a question on newspaper reader-

ship in the Mobilization for Youth Survey of the Lower East Side. The readership
of the Daily Daily Mirror (now defunct) on the one hand, and the New
News and
York Times and Herald Tribune (now defunct) on the other hand, is almost comple-
mentary in distribution by social class.
The Social Stratification of (r) 47

Macy's S. Kiein
dresses $14.95 $ 5.00
girls' coats 16.99 12.00
stockings 0.89 0.45
men's suits 49.95-64.95 26.00-66.00

The emphasis on prices is also different. Saks either does not mention
prices, or buries the figure in small type at the foot of the page.
Macy's features the prices in large type, but often adds the slogan,
"You get more than low prices." Kleins, on the other hand, is often
content to let the prices speak for themselves. The form of the prices
is also different: Saks gives prices in round figures, such as $120;

Macy's always shows a few cents off the dollar: $49.95; Kleins usually
prices its goods in round numbers, and adds the retail price which
is always much higher, and shown in Macy's style: "$23.00, marked

down from $49.95."

The physical plant of the stores also serves to differentiate them.
Saks is the most spacious, especially on the upper floors, with the
leastamount of goods displayed. Many of the floors are carpeted,
and on some of them, a receptionist is stationed to greet the cus-
tomers. Kleins, at the other extreme, is a maze of annexes, sloping

concrete floors, low ceilings; has the maximum amount of goods

it

displayed at the least possible expense.

The upon the employees is the prestige
principal stratifying effect
of the store,and the working conditions. Wages do not stratify the
employees in the same order. On the contrary, there is every indica-
tion that high-prestige stores such as Saks pay lower wages than
Macy's.
Saks is a and the general wage structure is not
nonunion store,
a matter of public record. However, conversations with a number
of men and women who have worked in New York department
stores, including Saks and Macy's, show general agreement on the
direction of the wage differential. 4 Some of the incidents reflect a

4. Macy's sales employees are represented by a strong labor union, while Saks is
not unionized. One former Macy's employee considered it a matter of common
knowledge that Saks wages were lower than Macy's, and that the prestige of the store
helped to maintain its nonunion position. Bonuses and other increments are said to
enter into the picture. It appears that it is more difficult for a young girl to get a job
at Saks than at Macy's. Thus Saks has more leeway in hiring policies, and the tendency
of the store officials to select girls who speak in a certain way will play a part in
the stratification of language, as well as the adjustment made by the employees to
their situation. Both influences converge to produce stratification.
48 SOCIOLINGUISTIC PATTERNS

willingness of sales people to accept much lower wages from the

store with greater prestige. The executives of the prestige stores pay
a great deal of attention to employee relations, and take many
unusual measures ensure that the sales people feel that they share
to
5
in the general prestige of the store. One of the Lower East Side
informants who worked at Saks was chiefly impressed with the fact
that she could buy Saks clothes at a 25 percent discount. A similar
concession from a lower-prestige store would have been of little
interest to her.
From the point of view of Macy's employees, a job in Kleins is
well below the horizon. Working conditions and wages are generally
considered to be worse, and the prestige of Kleins is very low indeed.
As we will see, the ethnic composition of the store employees reflects
these differences quite accurately.
A socioeconomic index which ranked New Yorkers on occupation
would show the employees of the three stores at the same level; an
income scale would probably find Macy's employees somewhat
higher than the others; education is the only objective scale which
might differentiate the groups in the same order as the prestige of
the stores, though there is no evidence on this point. However, the
working conditions of sales jobs in the three stores stratify them in
the order: Saks, Macy's, Kleins; the prestige of the stores leads to
a social evaluation of these jobs in the same Thus the two
order.
aspects of social stratification — differentiation
and evaluation are —
to be seen in the relations of the three stores and their employees.
The normal approach to a survey of department store employees
requires that one enumerate the sales people of each store, draw
random samples in each store, make appointments to speak with
each employee at home, interview the respondents, then segregate
the native New Yorkers, analyze and resample the nonrespondents,
and so on. This is an expensive and time-consuming procedure, but
for most purposes there is no short cut which will give accurate and
reliable results. In this case, a simpler method which relies upon the

5. A former Macy's employee told me of an incident that occurred shortly before

Christmas several years ago. As she was shopping in Lord and Taylor's, she saw the
president of the company making the rounds of every aisle and shaking hands with
every employee. When she told her fellow employees at Macy's about this scene, the
most common remark was, "How else do you get someone to work for that kind of
money?" One can say that not only do the employees of higher-status stores borrow
prestige from their employer— it is also deliberately loaned to them.
The Social Stratification of (r) 49

extreme generality of the linguistic behavior of the subjects was used

to gather a very limited type of data. This method is dependent upon
the systematic sampling of casual and anonymous speech events.
Applied in a poorly defined environment, such a method is open
to many biases and it would be difficult to say what population had
been studied. In this case, our population is well defined as the sales
people (or more generally, any employee whose speech might be
heard by a customer) in three specific stores at a specific time. The
result will be a view of the role that speech would play in the overall
social imprint of the employees upon the customer. It is surprising
that this simple and economical approach achieves results with a
high degree of consistency and regularity, and allows us to test the
original hypothesis in a number of subtle ways.

The Method

The application of the study of casual and anonymous speech

events to the department-store situation was relatively simple. The
interviewer approached the informant in the role of a customer
asking for directions to a particular department. The department was
one which was located on the fourth floor. When the interviewer
asked, "Excuse me, where are the women's shoes?" the answer would
normally be, "Fourth floor."
The interviewer then leaned forward and said, "Excuse me?" He
would usually then obtain another utterance, "Fourth floor," spoken
in careful style under emphatic stress. 6
The interviewer would then move along the aisle of the store to
a point immediately beyond the informant's view, and make a writ-
ten note of the data. The following independent variables were
included:

the store
floor within the store 7
sex
age (estimated in units of five years)

6. The interviewer in all cases was myself. I was dressed in middle-class style, with
jacket, white shirt and tie, and used my normal pronunciation as a college-educated
native of New Jersey (r-pronouncing).
7. Notes were also made on the department in which the employee was located,
but the numbers for individual departments are not large enough to allow comparison.
50 SOCIOLINGUISTIC PATTERNS

occupation (floorwalker, sales, cashier, stockboy)

race
foreign or regional accent, if any

The dependent variable is the use of (r) in four occurrences:

casual: fourth floor

emphatic: fourth floor

Thus we have preconsonantal and final position, in both casual and

emphatic styles of speech. In addition, all other uses of (r) by the
informant were noted, from remarks overheard or contained in the
interview. For each plainly constricted value of the variable, (r-1)
was entered; for unconstricted schwa, lengthened vowel, or no
representation, (r-0) was entered. Doubtful cases or partial con-
striction were symbolized d and were not used in the final tabulation.
Also noted were instances of affricates or stops used in the word
fourth for the final consonant, and any other examples of nonstand-
ard (th) variants used by the speaker.
This method of interviewing was applied in each aisle on the floor
as many times as possible before the spacing of the informants
became it was noticed that the same question had been
so close that
asked before. Each floor of the store was investigated in the same
way. On the fourth floor, the form of the question was necessarily
different:
"Excuse me, what floor is this?"
Following this method, 68 interviews were obtained in Saks, 125
in Macy's, and 71 in Kleins. Total interviewing time for the 264
subjects was approximately 6.5 hours.
At this point, we might consider the nature of these 264 interviews
in more general terms. They were speech events which had entirely
different social significance for the two participants. As far as the
informant was concerned, the exchange was a normal salesman-
customer interaction, almost below the level of conscious attention,
in which relations of the speakers were so casual and anonymous
that they may hardly have been said to have met. This tenuous
relationship was the minimum intrusion upon the behavior of the
subject; language and the use of language never appeared at all.
From the point of view of the interviewer, the exchange was a
systematic elicitation of the exact forms required, in the desired
context, the desired order, and with the desired contrast of style.
The Social Stratification of (r] 51

Overall Stratification of (r)

The results of the study showed clear and consistent stratification

of (r) in the three stores. In Fig. 2.1, the use of (r) by employees of

Saks, Macy's and Kleins is compared by means of a bar graph. Since

Saks
Macy's

32
31
b. Klein

Fig. 2.1.
N
if
68
O
W4
125
4

Overall stratification of (r) by store. Shaded area = %

17

71

all (r-1);unshaded area = % some (r-1); % no (r-1) not shown.

N= total number of cases.

the data for most informants consist of only four items, we will not
use a continuous numerical index for (r), but rather divide all in-
formants into three categories.

all (r-1): those whose records show only (r-1) and no (r-0)
some (r-1): those whose records show at least one (r-1) and one
(r-0)
no (r-1): those whose records show only (r-0)

From Fig. 2.1 we see that a total of 62 percent of Saks employees,

51 percent of Macy's, and 20 percent of Kleins used all or some (r-1).
The even sharper for the percentages of all (r-1). As
stratification is

the hypothesis predicted, the groups are ranked by their differential

use of (r-1) in the same order as their stratification by extralinguistic
factors.
Next, we may wish examine the distribution of (r) in each of
to
shows this type of display, where
the four standard positions. Fig. 2.2
once again, the stores are differentiated in the same order, and for
each position. There is a considerable difference between Macy's and
Kleins at each position, but the difference between Macy's and Saks
varies. In emphatic pronunciation of the final (r), Macy's employees
come very close to the mark set by Saks. It would seem that ^pro-
nunciation is the norm at which a majority of Macy employees aim,
52 SOCIOLINGUISTIC PATTERNS

4th floor 4th floor

80
63 64
60
^^61
40 H ^^44\
M 27
V 22 /
20.
8
"— 18
5
13

Fig. 2.2 Percentage of all (r-1) by store for four positions. (S = Saks,
M = Macy's, K = Kleins.)

yet not the one they use most often. In Saks, we see a shift between
casual and emphatic pronunciation, but it is much less marked. In
other words, Saks employees have more security in a linguistic
sense. 8
The low should not
fact that the figures for (r-1) at Kleins are
obscure the fact that Kleins employees also participate in the same
pattern of stylistic variation of (r) as the other stores. The percentage
of r-pronunciation rises at Kleins from 5 to 18 percent as the context
becomes more emphatic: a much greater rise in percentage than in
the other stores, and a more regular increase as well. It will be
important to bear in mind that this attitude that (r-1) is the most —
appropriate pronunciation for emphatic speech is shared by at least —
some speakers in all three stores.
Table 2.1 shows the data in detail, with the number of instances
obtained for each of the four positions of (r), foreach store. It may
be noted that the number of occurrences in the second pronunciation
of four is considerably reduced, primarily as a result of some
speakers' tendency to answer a second time, "Fourth."

8. The extremestyle shifting of the second-highest status group appears throughout

the New
York City pattern, and is associated with an extreme sensitivity to the norms
of an exterior reference group (see Ch. 5 in this volume). In Table 5.1 on p. 133
the data on the Index of Linguistic Insecurity is given, which is the number of items
in which a speaker distinguishes between his own pronunciation of a word and the
correct pronunciation. The second-highest group has the highest scores on this index.
We find parallel phenomena in Shuy, Wolfram, and Riley 1967, Wolfram 1969, and
Levine and Crockett 1966, who found in their study of Hillsboro. North Carolina that
the second-highest group on the basis of education showed the most extreme stylistic
shift of (r).
The Social Stratification of (r) 53

TABLE 2.1.
DETAILED DISTRIBUTION OF (r) BY STORE AND WORD POSITION

Saks Macy's S. Klein

Casual Emphatic Casual Emphatic Casual Emphatic

(0 4th floor 4th floor 4th floor 4th floor 4th floor 4th floor

(r-1) 17 31 16 21 33 48 13 31 3 5 6 7

(r-0) 39 18 24 12 81 62 48 20 63 59 40 33
d 4 5 4 4 3 1 1 1 3 3

No
data* 8 14 24 31 11 12 63 74 4 6 22 28

Total no. 68 68 68 68 125 125 125 125 71 71 71 71

*The "no data" category for Macy's shows relatively high values under the emphatic category.
This discrepancy is due to the fact that the procedure for requesting repetition was not stand-
ardized in the investigation of the ground floor at Macy's, and values for emphatic response
were not regularly obtained. The effects of this loss are checked in Table 2.2, where only com-
plete responses are compared.

Since the numbers in the fourth position are somewhat smaller

than the second, it might be suspected that those who use [r] in Saks
and Macy's tend to give fuller responses, thus giving rise to a spurious
impression of increase in (r) values in those positions. We can check
this point by comparing only those who gave a complete response.
Their responses can be symbolized by a four-digit number, repre-
senting the pronunciation in each of the four positions respectively
(see Table 2.2).
Thus we see that the pattern of differential ranking in the use of

TABLE 2.2.

DISTRIBUTION OF (r) FOR COMPLETE RESPONSES

% of total respoi Tses in

(0 Saks Macy's 5. Klein

All (r- 1) 1111 24 22 6

Some (r- 1)0111 46 37 12
11
10 1 etc.

No (r 1) 30 41 82

ioo 100 100

N = 33 48 34
54 SOCIOLINGUISTIC PATTERNS

preserved in this subgroup of complete responses, and omission

(r) is

of the final "floor" by some respondents was not a factor in this

pattern.

The Effect of Other Independent Variables

stores, may explain

Other factors, besides the stratification of the
the regular pattern of r-pronunciation seen above, or this effect may
be the contribution of a particular group in the population, rather
than the behavior of the sales people as a whole. The other inde-
pendent variables recorded in the interviews enable us to check such
possibilities.

Race
There are many more black employees in the Kleins sample than
in Macy's, and more in Macy's than in Saks. Table 2.3 shows the
percentages of black informants and their responses. When we
compare these figures with those of Fig. 2.1, for the entire population,
it is evident that the presence of many
black informants will con-
tribute to a lower use of (r-1). The black subjects at Macy's used
less (r-1) than the white informants, though only to a slight extent;
the black subjects at Kleins were considerably more biased in the
r-less direction.
The higher percentage of black sales people in the lower-ranking
stores is consistent with the general pattern of social stratification,
since in general, black workers have been assigned less desirable
jobs. Therefore the contribution of black speakers to the overall
pattern is consistent with the hypothesis.

TABLE 2.3.

DISTRIBUTION OF (r) FOR BLACK EMPLOYEES

0,
/o of responses in

(0 Saks Macy's S. Klein

All (r-1) 50 12
Some (r-1) 35 6
No (r-1) 50 53 94

100 100 100

N = 2 17 18
% of black
informants: 03 14 25
The Social Stratification of (r) 55

Occupation
There are other differences in the populations of the stores. The
types of occupations among the employees who are accessible to
customers are quite different. In Macy's, the employees who were
interviewed could be identified as floorwalkers (by red and white
carnations), sales people, cashiers, stockboys,and elevator operators.
In Saks, the cashiers are not accessible to the customer, working
behind the sales counters, and stockboys are not seen. The working
operation of the store goes on behind the scenes, and does not intrude
upon the customer's notice.On the other hand, at Kleins, all of the
employees seem be operating on the same level: it is difficult to
to
tell the difference between sales people, managers, and stockboys.
Here again, the extralinguistic stratification of the stores is rein-
forced by objective observations in the course of the interview. We
can question if these differences are not responsible for at least a
part of the stratification of (r). For the strongest possible result, it
would be desirable to show that the stratification of (r) is a property
of the most homogeneous subgroup in the three stores: native New
York, white sales women. Setting aside the male employees, all

occupations besides selling itself, the black and Puerto Rican em-
ployees, and all those with a foreign accent, 9 there are still a total
of 141 informants to study.
Fig. 2.3 shows the percentages of (r-1) used by the native white
sales women of the three stores, with the same type of graph as in
Fig. 2.1. The stratification is essentially the same in direction and
outline, though somewhat smaller in magnitude. The greatly reduced
Kleins sample still shows by far the lowest use of (r-1), and Saks
is ahead of Macy's in this respect. We can therefore conclude that

the stratification of (r) is a process which affects every section of

the sample.

9. In the sample as a whole, 17 informants with distinct foreign accents were found,

and one with regional characteristics which were clearly not of New York City origin.
The foreign language speakers in Saks had French, or other western European accents,
while those in Kleins had Jewish and other eastern European accents. There were
three Puerto Rican employees in the Kleins sample, one in Macy's, none in Saks. There
were 70 men and 194 women. Men showed the following small differences from women
in percentages of (r-1) usage:

men women
all (r-1) 22 30
some (r-1) 22 17
no (r-1) 57 54
56 SOCIOLINGUISTIC PATTERNS

Saks
Macy's

S. Klein

26

4 ezzzzzz
N 27

Fig. 2.3. Stratification of by store for native New York white

(r)

sales women. Shaded area = % all (r-1); unshaded area = %

some (r-1); % no (r-1) not shown. N = total number of cases.

Wecan now turn the heterogeneous nature of the Macy's sample

to advantage. Fig. 2.4 shows the stratification of (r) according to
occupational groups in Macy's: in line with our initial hypothesis,
this is much sharper than the stratification of the employees in
general.The total percentage of those who use all or some (r-1) is

almost the same for the floorwalkers and the sales people but a
much higher percentage of floorwalkers consistently use (r-1).

Another interesting comparison may be made at Saks, where there

is a great discrepancy between the ground floor and the upper floors.

The ground floor of Saks looks very much like Macy's: many
crowded counters, salesgirls leaning over the counters, almost elbow
to elbow, and a great deal of merchandise displayed. But the upper
floors of Saks are far more spacious; there are long vistas of empty
carpeting, and on the floors devoted to high fashion, there are models
who display the individual garments to the customers. Receptionists

Floorwalkers Sales

35

Stockboys

14
vss/r/
N= 13 105

Fig. 2.4. Stratification of (r) by occupational groups in Macy's.

Shaded area = % all (r-1); unshaded area = % some (r-1); % no
(r-1) not shown. N = total number of cases.
The Social Stratification of (r) 57

are stationed at strategic points to screen out the casual spectators

from the serious buyers.
It would seem logical then, to compare the ground floor of Saks

with the upper floors. By the hypothesis, we should find a differential

use of (r-1). Table 2.4 shows that this is the case.
TABLE 2.4.

DISTRIBUTION OF (r) BY FLOOR IN SAKS

(0 Ci round floor Upper floors

% all (r-1) 23 34
% some (r-1) 23 40
% no (r-1) 54 26

100 100
N = 30 38

In the course of the interview, information was also collected on

the (th) variable, particularly as occurred in the word fourth. This
it

is one of the major variables used in the study of social stratification

in New York (Labov 1966a) and elsewhere (Wolfram 1969; Anshen

1969). The most strongly stigmatized variant is the use of the stop
[t] in fourth, through, think, etc. The percentage of speakers who used

stops in this position was fully in accord with the other measures
of social stratification which we have seen:

Saks 00%
Macy's 04
S. Klein 15

Thus the hypothesis has received a number of semi-independent

confirmations. Considering the economy with which the information
was obtained, the survey appears to yield rich results. It is true that
we do not know a great deal about the informants that we would
like to know: their birthplace, language history, education, partici-
pation in New York culture, and so on. Nevertheless, the regularities
of the underlying pattern are strong enough to overcome this lack
of precision in the selection and identification of informants.

Differentiation by Age of the Informants

The age of the informants was estimated within five-year intervals,

and these figures cannot be considered reliable for any but the
simplest kind of comparison. However, it should be possible to break
58 SOCIOLINGUISTIC PATTERNS

down the age groups into three units, and detect any overall direction
of change.
as we have indicated, (r-1) is one of the chief characteristics
If,

of a new prestige pattern which is being superimposed upon the

native New York City pattern, we would expect to see a rise in
r-pronunciation among the younger sales people. The overall distri-
bution by age shows no evidence of change, however in Table 2.5:

TABLE 2.5.

DISTRIBUTION OF (r) BY ESTIMATED AGE

Age groups

(r) 15-30 35-50 55-70

% all (r-1) 24 20 20
% some (r-1) 21 28 22
% no (r-1) 55 52 58

This lack of direction is surprising, in the light of other evidence

that the use of (r-1) as a prestige variant is increasing among younger
people in New York City. There is clearcut evidence for the absence
of (r-1) in New York City in the 1930's (Kurath and McDavid 1951)
and a subsequent increase in the records of Hubbell (1950) and
Bronstein (1962). When we examine the distributions for the indi-
vidual stores, we find that the even distribution through age levels
disappears. Fig. 2.5 shows that the expected inverse correlation with
age appears in Saks, but not in Macy's or Kleins. Instead, Macy's
shows the reverse direction at a lower level, with older subjects using
more (r-1), and Kleins no particular correlation with age. This com-
plex pattern is even more puzzling, and one is tempted to dismiss
it as the absence of any pattern. But although the numbers of the

subgroups may appear to be small, they are larger than many of the
subgroups used in the discussions of previous pages, and as we will
see, it is not possible to discount the results.
The conundrum represented by Fig. 2.5 is one of the most sig-
nificant results of the procedures that have been followed to this
point. Where all other findings confirm the original hypothesis, a
single resultwhich does not fit the expected pattern may turn our
attention in new and profitable directions. From the data in the
department store survey alone, it was not possible to account for
The Social Stratification of (r) 59

Saks

Saks
Macy's Macy's
Saks Macy's
39

26

?e = 15-30
m35-50
?S

55-70
35

>>J/A V 71 /v

15-30 35-50 55-70

N= 9 31 23 38 54 26

S. Klein 5. Klein
S. Klein
15
19

Age=
?M 15-30
15

35-50
4
55-70
N= 20 26 22

Fig. 2.5. Stratification of (r) by store and age level. Shaded

area= % all (r-1); unshaded area = % some (r-1); % no (r-1) not
shown. N = total number of cases.

except in speculative terms. In the original report on the

Fig. 2.5
department store survey, written shortly after the work was com-
pleted, we commented:
How can we account for the differences between Saks and Macy's? think I

we cansay this: the shift from the influence of the New England prestige
pattern the Midwestern prestige pattern (r-ful) is felt most com-
(r-less) to
pletely at Saks. The younger people at Saks are under the influence of the
r-pronouncing pattern, and the older ones are not. At Macy's, there is less
sensitivity to the effect among a large number of younger speakers who
are completely immersed in the New York City linguistic tradition. The
stockboys, the young salesgirls, are not as yet fully aware of the prestige
attached to r-pronunciation. On the other hand, the older people at Macy's
tend to adopt this pronunciation: very few of them rely upon the older
pattern of prestige pronunciation which supports the r-less tendency of
older Saks sales people. This is a rather complicated argument, which would
certainly have to be tested very thoroughly by longer interviews in both
stores before it could be accepted.
 60 SOCIOLINGUISTIC PATTERNS

The complex pattern of Fig. 2.5 offered a considerable challenge

for interpretation and explanation, but one possibility that always
had be considered was that it was the product of the many sources
to
of error inherent in rapid and anonymous surveys. To confirm and
explain the results of the department store survey it will be necessary
to look ahead to the results of the systematic interviewing program
discussed in Chs. 3-7. When the results of the major study of the
Lower East Side were analyzed, it became clear that Fig. 2.5 was
not an artifact of the method but reflected real social patterns (Labov
1966a:342 ff). The Lower East Side data most comparable to the
department store study are the distribution of (r) by age and class
in Style B —
the relatively careful speech which is the main bulk of
the individual interview (see Ch. 3 for the definition of styles). To
Saks, Macy's, Kleins, we can compare upper middle class, lower
middle class, and working class as a whole. The age ranges which
are most comparable to the department store ranges are 20-29, 30-39,
and 40-. (Since the department store estimates are quite rough, there
would be no gain in trying to match the figures exactly.) Fig. 2.6 is

60 L'MC

Style B
40
(r)

UMC LMC
20
LMC

age 20-29 30-39 40-

IMP

20-29 30-39 40-

18
1
1
WC

20-29
1

1 1
wu
30-39
1
m
WC

40-

N: 5 2 9 6 5 7 4 55

Fig. 2.6. Classification of (r) by age and class on the Lower East
Side: in style B, careful speech.

then the age and class display for the Lower East Side use of (r) most
comparable to Fig. 2.5. Again, we see that the highest status group
shows the inverse correlation of (r-1) with age: younger speakers use
more (r-1); the second-highest status group shows (r) at a lower level
and the reverse correlation with age; and the working-class groups
at a still lower level with no particular correlation with age.
This is a very striking confirmation, since the two studies have
quite complementary sources of error. The Lower East Side survey
The Social Stratification of (r) 61

was a secondary random sample, based on a Mobilization for Youth

survey, with complete demographic information on each informant.
The interviews were tape-recorded, and a great deal of data on (r)

was obtained from each speaker in a wide variety of styles. On the

other hand, the department-store study involved a much greater
number of counts: the small amount of data
likelihood of error on a
per informant, the method of notation, the absence of tape recording
and reliance on short-term memory, the method of sampling, the
estimation of age of the informant, and the lack of background data
on the informants. Most of these sources of error are inherent in the
method. To compensate for them, we had the uniformity of the
interview procedure, the location of the informants in their primary
role as employees, the larger number of cases within a single cell,
the simplicity of the data, and above all the absence of the biasing
effect of the formal linguistic interview. The Lower East Side survey
was weak in just those areas where the department-store study was
strong, and strong where it was weak. The methodological differ-
ences are summed up in the table below.

Lower East Side Department-store

survey study

LES > DS
sampling random informants available at specific

locations
recording of data tape-recorded short term memory and notes
demographic data complete minimal: by inspection and
inference
amount of data large small
stylistic range wide narrow

DS > LES
size of sample moderate large
location home, alone at work, with others
social context interview request for information
effect of observation maximal minimal
total time per subject
(location and interview) 4-8 hours 5 minutes

The convergence of the Lower East Side survey and the depart-
ment-store survey therefore represents the ideal solution to the
Observer's Paradox (Ch. 8): that our goal is to observe the way people
use language when they are not being observed. All of our methods
involve an approximation to this goal: when we approach from two
 62 SOCIOLINGUISTIC PATTERNS

and get the same result, we can feel confident

different directions,
that we have reached past the Observer's Paradox to the structure
that exists independently of the analyst.
Given the pattern of Fig. 2.5 as a social fact, how can we explain
it? The suggestions advanced in our preliminary note seem to be

moving in the right direction, but at that time we had not isolated
the hypercorrect pattern of the lower middle class nor identified the
crossover pattern characteristic of change in progress. We must draw
more material from the later research to solve this problem.
Figs. 2.5 and 2.6 are truncated views of the three-dimensional
distribution of the new r-pronouncing norm by age, style, and social
class. Fig. 2.7 shows two of the stylistic cross sections from the more

100 r
Style A

80

(r) 60

20-29 yrs
40 30-39 yrs
Class 9

20 - 40-49 yrs
50-75 yrs

0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8

SEC
(r)

80
Style B

60 h 20-29 yrs

r Class 9
30-39 yrs
40 40-49 yrs

r
20 h 50-75 yrs

0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8

Fig. 2.7. Development of class stratification of (r) for casual

speech (Style A) and careful speech (Style B) in apparent time.
SEC = socioeconomic class scale.
The Social Stratification of (r) 63

detailed study of the Lower East Side population, with four sub-
divisions by The dotted line shows us how the highest status
age.
group (Class 9) introduces the new r-pronouncing norm in casual
speech. In Style A only upper-middle-class speakers under 40 show
any sizeable amount of (r-1). None of the younger speakers in the
other social groups show any response to this norm in Style A,
though some effect can be seen in the middle-aged subjects, espe-
group (Class 6-8, lower middle
cially in the second-highest status
exaggerated, with the mid-
class). In Style B, this imitative effect is
dle-aged lower-middle-class group coming very close to the upper-
middle-class norm. In more formal styles, not shown here, this
subgroup shows an even sharper increase in r-pronunciation, going
beyond the upper-middle-class norm in the "hypercorrect" pattern
that hasappeared for this group in other studies (see Ch. 5 in this
volume; Levine and Crockett 1966; Shuy, Wolfram, and Riley 1967).
Fig. 2.7 is not a case of the reversal of the age distribution of (r-1);

rather it is a one-generation lag in thepeak of response to the new

norm. The second-highest status group responds to the new norm
with a weaker form of imitation in connected speech, with middle-
aged speakers adopting the new norm of the younger high-status
speakers; Fig. 2.8 shows this schematically. Our studies do not give
the exact profile of the use of (r) among younger upper-middle-class
speakers, since we did not focus on that age range. In later observa-
tions, I have met some upper-middle-class youth who use 100 per-
cent (r-1), but in most families, (r-1) is still a superposed pronuncia-
tion in adolescence and Fig. 2.8 reflects this. If we wish to express
the (r-1) distribution in a single function, we can say that it is in-

UMC

Fig. 2.8. Hypothetical distribution of (r) as an incoming prestige

feature.
64 SOCIOLINGUISTIC PATTERNS

versely correlated with distance from the highest-status group (taking

Class 9 as 1, Classes 6-8 as 2, Classes 2-5 as 3, and Classes 0-1 as 4).
It is also directly correlated with the formality of style and the

amount of attention paid to speech (taking casual speech, style A, as

0, careful speech, style B, as 1, etc.). The slope of style shifting is

modified by a function which may be called the "Index of Linguistic

Insecurity" (ILI), which is maximized for the second-highest status
group (see Table 5.1, p. 133 for the quantitative index). The age dis-
tribution must be shown as greatest for the upper middle class at age
20 and at age 40 for the lower middle class. We can formalize these
observations by writing

(r-1) = -a (Class) + b (Style)(ILI) - c|(Class) • 20 - (Age)| + d

The third term is minimized upper middle class at age 20,

for the
for the lower middle class age 40, the working class at age 60,
at
etc. Fig. 2.7 supports this semiquantitative expression of a wave
effect, which has a number of unspecified constants.
still

There is between the behavior of the

a considerable difference
highest-status group and the others. The upper middle class develops
the use of (r-1) early in life— as a variable expression of relative
formality to be found at all stylistic levels. For the other groups in
New York City, there is no solid basis for (r-1) in the vernacular style
of casual speech; for them, (r-1) is a form which requires some
attention paid to speech if it is realized at all. As in so many other
formal marks of style-shifting, the lower middle class overdoes the
process of correction. This is a process learned late in life. When
speakers who are now 40-50 were growing up, the prestige norm
was not (r-1) but (r-0). Before World War II, the New York City
schools were dominated by an Anglophile tradition which taught
that (r-1) was a provincial feature, an incorrect inversion of the
consonant, and that the correct pronunciation of orthographic r in
car was (r-0), [ka ], in accordance with "international English". 10
•

10. See for example Voice and Speech Problems, a text written for New York City
schools in 1940 by Raubicheck, Davis, and Carll (1940:336):
There are many people who feel that an effort should be made to make the
pronunciation conform to the spelling, and for some strange reason, they are
particularly concerned with r. We all pronounce calm, psaim, aimond, know,
eight, night, and there without worrying Yet people who would not dream
. . .

of saying kni: or psai'kuladsi insist on attempting to sound the r in words like

pa-k or faSa just because an r marks the spot where our ancestors used a trill
The Social Stratification of (r) 65

No adjustment in the pronunciation of this consonant was then

necessary for New Yorkers who were trying to use the prestige
—
norm it was only vowel quality which had to be corrected. This
r-less norm can be seen in the formal speech of upper-middle-class
speakers, over 40, and lower-middle-class speakers over 50. It also
appears in subjective-reaction tests (Ch. 6) for older speakers. The
lower-middle-class speakers who now shift to (r-1) in formal styles
have abandoned their prestige norm and are responding to the
form used by the younger high-status speakers that they come
into contact with. On the other hand, many upper-middle-class
speakers adhere to their original norm, in defiance of the prevailing
trend. The pattern which we have observed in the department-store
survey is therefore a reflection of the linguistic insecurity of the
lower middle class, which has led the older generation to adopt the
most recent norm of (r-1) in preference to the older norm. The
process of linguistic socialization is slower for lower-middle-class
groups who do not go to college than for upper-middle-class
speakers, who begin adjusting to the new norm in the upper class
tracks of the academic high schools. For those who do not follow
this path, it takes 10 or 20 years to reach maximum sensitivity to
the hierarchical organization of formal language in their community.

Some Methodological Directions

The most important conclusion of the department-store study is

that rapidand anonymous studies can be a valuable source of
information on the sociolinguistic structure of a speech community.
There are a number of directions in which we can extend and
improve such methods. While some sources of error are inherent
in the method, others can be eliminated with sufficient attention.

. . . More often than not, people do not really say a third sound in a word like
pak with the tongue tip curled back toward the
but merely say the vowel a:

throat. This type of vowel production is known as "Inversion."

LetitiaRaubicheck was the head of the speech program in the New York City schools
for many years and exerted a powerful influence on the teaching of English there.
The norm of "international English" was maintained by William Tilly of Columbia
and followed by Raubicheck and many others in the 1930's and 1940's. As far as I
know, this norm has lost entirely its dominant position in the school system: a detailed
study of its disappearance from the radio networks and the school system in the 1940's
would tell us a great deal about the mechanism of such shifts in the prestige form.
66 SOCIOLINGUISTIC PATTERNS

In the department-store survey, the approach to sampling might

have been more systematic. would have been preferable to select
It

every nth sales person, or to use some other method that would avoid
the bias of selecting the most available subject in a given area. As
long as such a method does not interfere with the unobtrusive
character of the speech event, it would reduce sampling bias without
decreasing efficiency. Another limitation is that the data were not
tape-recorded. The transcriber, myself, knew what the object of the
test was, and it is always possible that an unconscious bias in tran-
scription would lead to some doubtful cases being recorded as (r-1)
in Saks, and as (r-0) in Kleins.
11
A third limitation is in the method
used to elicit emphatic speech. Fig. 2.2 indicates that the effect of
stylistic variation may
be slight as compared to the internal pho-
nological constraint of preconsonantal vs. final position. The total
percentages for all three stores bear this out.

% of all (r-1) for each position

Casual Emphatic
fourth floor fourth floor

23 39 24 48

A simple request for repetition has only a limited effect in inducing

more formal speech. The use of reading passages, word lists, and
minimal pairs in the Lower East Side study gave a wider range of
styles. might be possible to enlarge the stylistic range in rapid and
It

anonymous studies by emphasizing the difficulty in hearing by one

technique or another.
The sources of error in the department-store study are offset by
the comparability of the three subsections, the size of the sample,
and the availability of the population for rechecking. Though the
individual speakers cannot be relocated, the representative popula-
tion can easily be reexamined for longitudinal studies of change in
progress. There are limitations of such a "pseudopanel" as compared
to a true panel study of the same individuals; but the advantages
in cost and efficiency are overwhelming.

11. When the phonetic transcriptions were first made, doubtful cases were marked
as d and were not included in the tabulations made later. There is however room
for interviewer bias in the decision between (r-0) and d and between d and (r-1).
The Social Stratification of (r) 67

With such promising results in hand, it should be possible to refine

and improve the methods used, and apply them in a wider range
of contexts. In large cities reasonable to select single large
it is

institutions like department but there is no reason to limit

stores,
rapid and anonymous surveys to sales people or to institutions of
this character. We can turn to any large body of individuals located
at fixed "social addresses" and accessible to interaction with the
public: policemen, postal clerks, secretaries, ushers, guides, bus
drivers, taxi drivers, street peddlers and demonstrators, beggars,
construction workers, etc. The public groups which are most clearly
identified tend to be concentrated towards the lower end of the social
scale, with sales people at the upper end. But we can reach a more
general public by considering shoppers, spectators at sports events,
parades or construction sites, amateur gardeners, park strollers, and
passersby in general; here the general character of the residential
area can serve the same differentiating function as the three depart-
ment stores mentioned above. Many professionals of relatively high
social standing are available for public interaction: particularly
teachers, doctors, and lawyers. Such public events as courtroom
trials and public hearings allow us to monitor the speech of a wide
range of socially located and highly differentiated individuals. 12
There is in all such methods a bias towards those populations that
are available to public interaction, and against those which are so
located as to insure privacy: business and social leaders, or those
engaged in aesthetic, scholarly, scientific, or illegal activities. Any
of these groups can be studied with sufficient ingenuity: sociolin-
guistic research should certainly rise to the challenge to develop
rapid and anonymous studies that will escape the limitations of
convenience. But it should be emphasized that since those who are

12. Hearings of the New York City Board of Education were recorded during the
study of New York and preliminary analysis of the data shows that the pattern
City,
of social and stylistic stratification of (r) can easily be recovered from the wide variety
of speakers who appear in these hearings. Courtroom proceedings at the New York
Court of General Sessions are a natural focus for such studies, but speakers often
lower their voices to the point that spectators cannot hear them clearly. Only a small
beginning has been made on the systematic study of passersby. Plakins (1969) ap-
proached a wide variety of pedestrians in a Connecticut town with requests for
directions to an incomprehensible place, phrased at three levels of politeness. She
found systematic differences in mode of response according to dress (as an index of
socioeconomic position) and mode of inquiry; there were no "rude" responses [huh?]
to polite inquiries.
68 SOCIOLINGUISTIC PATTERNS

most available may have the most direct effect

to public interaction
upon change and the sociolinguistic system, the bias
linguistic
through missing the more extreme and obscure ends of the social
spectrum is not as great as it may first appear.
Since the department-store survey was carried out in Manhattan,
several parallel studies have been made. In Suffolk County, Long
Island, rapid and anonymous observations of the use of (r) were
made by Patricia Allen (1968). In three stratified stores, 156 em-
ployees were observed. In the highest-status store (Macy's), only 27
percent of the subjects used no (r-1); in the intermediate store (Grant
City), 40 percent; and in the low-status store (Floyd's), 60 percent.
We see that the general New York City pattern has moved outward
from the city, producing a comparable stratification of (r) in three
stores of a somewhat narrower range than those studied in Man-
hattan. Our own analysis of the New York City situation shows that
rapid and anonymous surveys of this kind cannot be interpreted fully
without detailed knowledge of the dialect history of the area, and
a more systematic study of the distribution of linguistic variables
and subjective norms. 13 In this case, rapid and anonymous surveys
should be considered a supplement or preliminary to other methods,
not substitutions for them. Yet there are cases where rapid methods
can give solutions to problems that have never been circumnavigated

13. Allen's tables resemble the New York City patterns but with one major differ-

ence; the number of speakers who use all (r-1) is roughly constant in all three stores:
27 percent in Floyd's, 27 percent in Grant City, 32 percent in Macy's. Examination
of the distribution in apparent time showed that this phenomenon was due to the
presence of a bimodal split in the lower-store adults (over 30 years old). Eighty percent
used no (r-1) and 20 percent used a consistent all (r-1): there were none who varied.
On the other hand, 50 percent of the adults were showing variable two other
(r) in the
stores. This points to the presence of an older r-pronouncing vernacular which is now
dominated by the r-less New York City pattern (Kurath and McDavid 1961), but
survives among working-class speakers. The disengagement of such bimodal patterns
is a challenging problem (Levine and Crockett 1966), and certainly requires a more

systematic survey. Similar complexity is suggested in the results of rapid and anony-
mous survey of stores in Austin, Texas by M. M. Harris (1969). In this basically
r-pronouncing area, the prestige norms among whites appear to be a weak constricted
[r], with a strongly retroflex consonant gaining ground among younger speakers. But

for the few blacks and Mexican-Americans encountered, this strong [r] seems to be
the norm aimed at in careful articulation. Although these results are only suggestive,
they are the kind of preliminary work which is required to orient a more systematic
investigation towards the crucial variables of the sociolinguistic structure of that
community.
The Social Stratification of (r) 69

by conventional techniques. We have used observations of the

speech of telephone operators to construct a national map of the
merger of the low back vowels in hock and hawk, and the merger
of i and e before nasals in pin and pen. In our recent study of the
Puerto Rican speech community in New York City, we utilized such
natural experimentation to find out what percentage of those heard
speaking Spanish on the street were raised in the United States,
and what percentage were born in Puerto Rico (Labov and Pedraza
1971).
Future studies of language in its social context should rely more
heavily on rapid and anonymous studies, as part of a general program
of utilizing unobtrusive measures to control the interactive effect of
the observer (Webb et al. 1966). But our rapid and anonymous studies
are not passive indices of social use, like observations of wear and
tear in public places. They represent a form of nonreactive experi-
mentation in which we avoid the bias of the experimental context
and the irregular interference of prestige norms but still control the
behavior of subjects. We are just beginning to study speech events
like asking for directions, isolating the invariant rules which govern
them, and on this basis develop the ability to control a large body
of socially located public speech in a natural setting. We see rapid
and anonymous observations as the most important experimental
method in a linguistic program which takes as its primary object
the language used by ordinary people in their everyday affairs.
 The Isolation of
3 Contextual Styles

THE investigation of sound change on Martha's Vineyard can

be seen as the step in a program for the study of language in
first

its The second was an attack on a much larger prob-

social context.
lem: to find some system or order in the extensive variation of
English in New York City. 1 Previous reports had registered a chaotic
proliferation of free variation in almost every part of the vowel
system (Labov 1966a:2). Those who identify structure with homo-
geneity will find very little structure in New York City. In addition
to a great range of social variation, there was also reported wide-
spread stylistic variation, giving the general impression that anyone
could say anything. Typical was Hubbell's report on (r):

The speaker heard both types of pronunciation about him all the time, both
seem almost equally natural to him, and it is a matter of pure chance which
one comes to his lips. (1950:48)

Linguists have never been unconscious of the problems of stylistic

variation.The normal practice is to set such variants aside not —
because they are considered unimportant, but because the tech-
niques of linguistics are thought to be unsuitable or inadequate to
handle them. Structural analysis is normally the abstraction of those
unvarying, functional units of language whose occurrence can be

1. This paper is adapted from Ch. 4 of The Social Stratification of English in New

York City (1966), and represents the techniques for isolating casual speech and other
styles which were developed in the 1963-64 study of the Lower East Side of New
York City. These methods are still basic to any series of individual interviews, and
are now utilized regularly in studies of sound change in progress in a wide variety
of English, Spanish, and French dialects. For later techniques utilizing group inter-
action, see Labov et al. 1968:1.

70
The Isolation of Contextual Styles 71

predicted by rule. Since the influence of stylistic conditioning on

linguistic behavior is said to be merely statistical, it leads to state-
ments of probability rather than rule and is therefore uninteresting
to many linguists.
For the present purposes, I would rather say that stylistic variation
has not been treated by techniques accurate enough to measure the
extent of regularity which does prevail. The combination of many
stylistic factors imposed upon other influences may lead to seemingly
erratic behavior; but this apparent irregularity is comparable to the
inconsistencies which seemed to govern the historical development
of vowels and consonants until some of the more subtle conditioning
factors were perceived.
In the last chapter, we
considered one approach to discovering
the system within this The department-store survey
variation.
showed some stylistic variation as well as vertical stratification. But
the major attack on the New York City system requires much richer
data: long interviews with individuals whose social position and
geographic history is known; here the problem of stylistic variation
becomes paramount.
The New York City study began with 70 exploratory interviews
which examined in detail the phonological variation of a wide range
of speakers. They were concentrated in the Lower East Side, where
the population had been enumerated and a sociological survey
carried out by the research branch of Mobilization for Youth, a job
training agency. It seemed possible to do a secondary survey of the
Lower East Side, using the sample already constructed by MFY.
These exploratory interviews showed five phonological variables
that seemed to exhibit regular variation in different styles and con-
texts. The five variables will form the main substance for Chs. 3-6,
and will enter into the more general discussions of Chs. 7-9. We may
therefore consider them in some detail at the outset. To define a
linguistic variable, we must (a) state the total range of linguistic
contexts in which it occurs, (b) define as many phonetic variants
as we can reasonably distinguish, (c) set up a quantitative index for
measuring values of the variables. These steps have already been
illustrated in the discussion of centralization in Martha's Vineyard;
we will follow them now for the five New York City variables, (r),

(eh), (oh), (th), and (dh).

The notational conventions used in the discussion of the variables
and throughout this volume are given below, pp. 72-78. As pointed
72 SOCIOLINGUISTIC PATTERNS

out in Ch. 1, the variable indicates a focus on significant distributions

within the unit, constraining what would otherwise be considered
free and unconstrained variation. The variables (ay) and (aw) were
isomorphic with the phonemes /ay/ and /aw/; but the variable (r)
corresponds to the presence or absence of /r/; and the variable (eh)
includes in its range the phonemes /aeh/, /eh/, and /ih/.
Particular variants or values of the variables will be indicated by
a number within the parentheses, as (r-1) or (eh-4). Index scores
derived from mean values of the variants will be indicated by num-
bers outside the parentheses, as (r)-21 or (eh)-28. Brackets will con-
tinue to indicate phonetic notation, showing impressionistic repre-
sentations of the speech sounds heard; slashes will indicate here
autonomous phonemes as /eh/ or /r/: the system of contrastive units
independent of grammatical alternations. The more abstract mor-
phophonemic units or systematic phonemes will be indicated by
italicized forms as r or short a which are often close to the ortho-
graphic representation. Since we are dealing with low-level phono-
logical rules, we will not normally be concerned with this higher-
level representation, but it will be helpful to note at various stages
the occurrence of merger of the autonomous level. This is particu-
larly useful in establishing discrete index values in a continuous
range of phonetic forms.
The correct analysis of the linguistic variable is the most important
step in sociolinguistic investigation.We want to isolate the largest
homogeneous which all subclasses vary in the same way.
class in
If we fail to do this, and throw together invariant subclasses, high-

frequency, and low-frequency subclasses, our view of the socio-

linguistic structures will be blurred. The regular pattern of the
variable may be submerged by a large number of irregular cases or —
even elements varying in a reverse direction. Once we have estab-
lished this linguistic definition of the variable, we are in a position
to follow the important principle of accountability: we will report
values for every case where the variable element occurs in the
relevant environments as we have defined them.

The Five Phonological Variables

(r): the presence or absence of consonantal constriction for post-

vocalic, word-final and preconsonantal /r/. This includes beer,
beard, bare, bared, moor, moored, bore, board, fire, fired, flower,
The Isolation of Contextual Styles 73

flowered, where the /r/ is usually represented by a vocalic inglide

[?]; unstressed syllables in Saturday, November, where we have only

a schwa [a]; and bar, barred, where /r/ is usually represented by

a lengthened vowel. However, we can get long monophthongs with
high vowels, as in beer [b*:], and there is sometimes an inglide heard
with bar.
Specifically excluded from the variable are the cases where /r/
follows a mid-central vowel, as in her and bird. These two subclasses
have different histories and behavior in New York City as in most
r-less dialects (Labov 1966a:10). In stressed her we have an alterna-
tion of [hA~hAr~~ h3~h3:] and with bird, the stigmatized palatal
upglide [b3 J d] which is replaced with a constricted /r/ [b^d] more
often than the main subclasses. We can account for this by repre-
senting bird and her in the dictionary as /hr/ and /brd/ (see Bloom-
field 1933); the term "postvocalic" in our definition thus eliminates
this class, and the phonetic vowel is inserted by a later rule (see Ch.
5 in Labov 1972a).
We also exclude word-final /r/ where the next word begins with
a vowel, as in four o'clock. This forms a separate subcase in New
York City, with a much
higher percentage of constricted /r/.
The two basic variants of (r) are thus
(r-1) [r, &, ?] i.e., presence of weak or strong consonantal con-
striction
(r-0) [?, a, :] i.e., absence of constriction

Borderline cases are recorded in parentheses and excluded from the

count. There are relatively few of these. The (r) index is then the
mean value of the variants recorded multiplied by 100: i.e., the
percentage of constricted forms.
(eh): the height of the nucleus of the vowel in tensed short a or
/aeh/. This phoneme is established in New York City by a complex
tensing rule which selects certain phonological subclasses; the
lengthened, fronted [ae* :] is then affected by a raising rule which
carried the vowel to [e* :?], [e* :§] and [V :?].

The tensing rule selects short a before front nasal consonants /m/
and /n/, voiceless fricatives, and
voiced stops /b, d, 3,
/f, 0, s, J/,

g/. The rule is variable (by types and tokens) for voiced fricatives
/v,z/ so that razz, jazz, rasberry are unpredictable. The consonants
mentioned must be followed by a word boundary # # or inflectional
boundary # or an obstruent; if a vowel or a liquid /r,l/ follows
74 SOCIOLINGUISTIC PATTERNS

directly, the tensing rule does not apply. Thus NYC opposes tense
waggin', draggin', stabbiri, to lax wagon, dragon, cabin. In general,
the rule does not recognize a derivational boundary +, giving lax
passage, Lassie, etc., though there is some variation after sibilants
The rule does not apply to weak words
as in fashion, fascinate, etc.
that function words which can have schwa as their only vowel:
is,

am, an, can(Aux), has, had, as, etc. There are lexical exceptions like
tense avenue, and variably tense wagon, magic, etc. The most regular
aspect of the rule takes the form:

nas
+ lowl back #
-backj [ + tense] [-Wk] tense + obstr]
|_ acont
«(

For further details on the New York City tensing rule see Trager 1942
and Cohen 1970. It is plain here that there is a great deal of variation
in polysyllables and derivational forms. Learned words like lass and
mastodon are also quite variable. Since we are interested primarily
in the raising of tense (eh), we can best focus on the invariant core
of the tense class: monosyllables before front nasals, voiced stops,
and voiceless fricatives. Among monosyllables, this invariant tense
classcan be opposed to a class of invariant lax and variably tense
forms:

(a) always lax cap, bat, batch, bat, pal, can (Aux),
had, has
(b) variable jazz, salve
bang
(c) tense cab, bad, badge, bag
half, pass, cash, bath,
ham, dance
The third word class is uniformly affected in the New York City
vernacular by a lower-level raising rule. This can best be shown as
a variable rule which variably decreases the openness of the vowel:

[+ tense]
<X - 6 ° Pen>
Lback_T
In this form, the rule progressively affects all front vowels as the
scope of x is increased to include the most open (low) vowels, and
The Isolation of Contextual Styles 75

lessopen (mid) vowels. The quantity 8 is a function of age, sex, style,

and social class and ethnic group as we will see.
For the purposes of our study, it is necessary to establish discrete
phonetic variants for the (eh) variable. Though the height of the
vowel is a continuous variable, we can establish such discrete coding
points with the help of other word classes that are relatively fixed.

Scale for (eh) Index

Approximate phonetic
No. quality Level with the vowel of
(eh-1) [V :*] NYC beer, beard
(eh-2) [e* :
s
]
NYC bear, bared
[e -ol
To' .
J

(eh-3) [ae*:]

(eh-4) [ae:] NYC bat, batch

(eh-5) [a:] E. New England pass, aunt

The last point on the scale occurs only in hypercorrection or imita-

tion of the older prestige norm
England broad a.
of New
The index score for (eh) is determined by coding each occurrence
of a member of word class (c) above as one of the six variants, taking
an average of the numerical values and multiplying by 10. Thus
(eh)-25 would be the index value for a person who pronounced half
of the (eh) words with (eh-3) and half with (eh-2). A person who
always used a tense vowel, level with the nucleus of bat, will be
assigned (eh)-40.
We have examined the raising of (eh) in much greater detail by
spectrographic means in recent work on sound change in progress.
Our instrumental studies confirm most of the impressionistic ratings
assigned by the above scale, and show the emergence of a sharp
differentiation between lax and tense vowels as well. Vowels affected
by lowered to first formant positions equal
social correction are often
to bat, but with higher second formant positions that is, with more —
extreme fronting. For further details, see Labov 1970b, and Labov,
Yaeger, and Steiner (1972). Fig. 3.1 shows spectrographic meas-
urements of the vowel system of one informant from the New York
City study: the subject is Jacob S., an older man, who shows a
moderate degree of raising of (eh) but a clear differentiation of the
tense and lax class. Within the tense class, there are further differ-
entiations of the three subclasses with short a before front nasals
showing the most advanced positions.
76 SOCIOLINGUISTIC PATTERNS

Fig. 3.1.Vowel system of Jacob S., 57, New York City (from
Labov, Yaeger and Steiner 1972).
The Isolation of Contextual Styles 71

(oh): The corresponding back vowel (oh) is also raised variably

in New York City: vowels more, talk, caught,
in the class of off, lost,
wash, etc., are raised progressively to mid and high position. There
is no need for a tensing rule: common short o words before voiceless

fricatives /f, 9, s/ and back nasals/n/ were raised at an earlier period

and are now included in the New York City class of long open o
words: off, lost, cloth, long, song, wrong. These are added to the
miscellaneous class of words which have coalesced into long open
o and form the basis for the variable (oh), along with o before /r/.
The raising rule for (oh) is a generalization of the raising rule for
(eh). only necessary to remove the feature [-back] from the left
It is

hand side of the rule. The variable constraint <-back> will then
appear in the environment for different social classes and ethnic
groups, as we will see in Ch. 5 below.
A six-point linear scale parallel to that for (eh) is used to measure
the height of this vowel: the great number of diacritics needed in
the phonetic quality is matched by the miscellaneous collection of
reference points. The difficulty of the phonetic description of this
vowel is so great that none of these methods are satisfactory, and
the following discussion may be of some help.

Scale for (oh) Index

Approximate
No. Phonetic quality Level with the vowel of
(oh-1) [u.-] NYC sure
(oh-2) [o" :*]

(oh-3) [o":«] General American for, nor

(oh-4) [K] IPA cardinal /o/
(oh-5) M E.New England hot, dog
(oh-6) [a] NYC dock, doll
(oh-4) is the vowel height
with the fixed position for
level
cardinal [d]. It is speech of upstate New York
heard frequently in the
residents, and in many other parts of the country, but never with
enough consistency for the speech of a particular region to serve
as a firm reference point, (oh-3) is somewhat higher, and may be
identified fairly accurately as the sound preceding [r] in for, or, nor,
in almost any region of the United States where [r] is pronounced
in those words.
(oh-2) is higher than (oh-3), more forward, and more rounded. The
78 SOCIOLINGUISTIC PATTERNS

centering glide which follows is often more marked than with (oh-3),
but a glide does not necessarily follow, (oh-1) is raised and centered
beyond (oh-2), level with most pronunciations of sure, and is rounded
with what appears to be considerable tension. The rounding is quite
different from that observed in British tense [o:]: it is actually a
pursing of the lips, in women; in men, a similar but distinct phonetic
quality is imparted by what seems to be a hollowing of the tongue.
The impressionistic transcription of (oh) has been confirmed and
checked by spectrograph^ measurement in our studies of sound
change in progress. On Fig. 3.1 we can see the raising of (oh) for
Jacob S., with a fairly advanced state of the variable.
(th) and (dh). These two variables are the initial consonants of
thing and then; they are well known throughout most of the United
States as the stereotype dese, dem, and dose. These consonants do
not of course show any close relation to the vowel system; they are
incorporated in this study as a pair of correlated variables which
are not involved in any of the processes of structural change which
affect the first three variables.

(th) (dh)
1 an interdental fricative [e] [5]
2 an affricate [te] [dS]
3 a lenis dental stop [t] [d]

The prestige form is the fricative, and the stop with

in this scale
its [t]-like or [d]-like effecteverywhere considered to have less
is

prestige. This stop consonant may be formed in a number of different

ways, but its essential quality is that no turbulent, fricative, or
scraping sound is heard as it is articulated. The affricate is a rapid
—
succession of the two forms or more precisely, it is heard as the
fricative with a sudden onset, instead of a gradual beginning.
The stop that is formed is usually dental. The [t] is usually not
aspirated as fully as the phoneme /t/ and the [d] is usually not voiced
as fully as /d/. Under stress, these phones can merge with /t/ and
/d/, yielding an intersection of the phonemes /6,t/, and /5,d/. Never-
theless, native speakers keep the two word classes quite separate;
we hear no hypercorrection in formal style such as /3a°nSe:r/ for
down there.
The zero variant in 'at, 'ere, etc. is rated as (dh-2), with the same
value as the affricate.
The Isolation of Contextual Styles 79

Contextual Styles

The initial New York City

exploration of the use of English in
suggested regular variation in different styles and contexts for these
five phonological variables. The problem is to control the context,
and define the styles of speech which occur within each context,
so that this hypothesis of regular variation can be tested.
For accurate information on speech behavior, we will eventually
need compare the performance of large numbers of speakers.
to
Furthermore, we will want to study a sample which is representative
of a much larger group, and possibly of the New York speech com-
munity as a whole. This cannot be done without random sampling.
Yet to complete random sampling, and to make the data for many
speakers comparable, we need structured, formal interviews. But the
formal interview itself defines a speech context in which only one
speaking style normally occurs, that which we may call careful
speech. The bulk of the informant's speech production at other times
may be quite different. He may use careful speech in many other
contexts, but on most occasions he will be paying much less attention
to his own speech, and employ a more relaxed style which we may
call casual speech. We can hear this casual speech on the streets
of New York, in bars, on the subway, at the beach, or whenever we
visit friends in the city. Yet anonymous observations in these con-
texts will also be biased. Our friends are a very special group, and
so too are those New Yorkers who frequent bars, play stickball in
the streets, visit public beaches, or talk loud enough in restaurants
to be overheard. Only through a painstaking random sampling of
the entire population can we avoid serious bias. The problem is now
to see what can be accomplished within the bounds of the interview.
We will begin with the dominant situation of the face-to-face in-
terview, which we will designate Context B, reserving Context A
for those situations which escape the social constraints of the in-
terview situation.

Context B. The Interview Situation

The simplest style to define is the one we have called careful
speech. In our investigation, this is the type of speech that normally
occurs when the subject is answering questions which are formally
recognized as "part of the interview." Generally speaking, an in-
terview which has as its professed object the language of the speaker
80 SOCIOLINGUISTIC PATTERNS

will rate higher on the scale of formality than most conversation. 2

It is not as formal a situation as a public address, and less formal

than the speech which would be used in a first interview for a job,
but it is certainly more formal than casual conversation among
friends or family members. The term "consultative," introduced by
Joos (1960), seems very apt for this stylistic level. The degree of
spontaneity or warmth in the replies of individuals may vary greatly,
but the relation of their careful speech to the speech of less formal
contexts is generally constant. Careful speech will then be defined
as that speech which occurs in Context B, and will be designated
StyJe B.
It is from Context
a relatively simple matter to shift the context
B in a more formal though there are a number of ways
direction,
of refining this procedure. In the following discussion, we will pursue
the definition and control of more formal styles to its ultimate
conclusion, before attempting to move in the opposite direction.

Context C. Reading Style

After the main body of the interview, which might last anywhere
from half an hour to an hour, the informant is asked to read two
standard texts. One of these is designed to concentrate the main
phonological variables in successive paragraphs, and the other to
juxtapose minimal pairs in a text. Both are written in a colloquial
style, to get as smooth a flow of language as possible, and to involve
the reader as much as possible in the story line. This involvement
gives us a maximum spread between Style C and the more formal
inquiries to follow, without any danger of reducing the distance
between B and C: the most formal conversational style will still be
sharply differentiated from reading style in the phonological varia-
bles. Secondly, the involvement in the story insures that there will
be a continuous flow of speech, with appropriate sandhi rules. It

The formal interviews on the Lower East Side were conducted as research of
2.

the "American Language Survey," which provided a framework for the study of
reading, of word lists, of attitudes towards language, and subjective reaction tests.
Our more recent studies do not take language as the overt topic of the research, but
a broader subject which includes language —
such as "common-sense learning." How-
ever, the stylistic constraints are roughly the same; the basic situation is that questions
are being asked by one person and answered by another. The more casual or vernacu-
lar style is used primarily with those who share the most knowledge together, where
the minimum amount of attention is paid to speech.
The Isolation of Contextual Styles 81

might have been possible to standardize in a different direction, by

urging the subject to read carefully and slowly, but very slow reading
is accompanied by special phonetic characteristics which would

make it difficult to compare conversation and reading style. For

—
example, the variable (r##v) final (r) followed by another word
—
beginning with a vowel, as in four o'clock may become hard to code
if the tempo is very slow. In normal speech, a pronunciation in which

no consonant occurs between four and o'clock would be entered as

a violation of the rule followed by most New Yorkers which pre-
serves [r] in this position. But such a rule begins to break down if
speech is slow enough. Then too, at a very slow tempo of reading
the minimal pairs are more likely to be noticed by the reader.
Therefore the overall design of the two texts is to encourage a
reasonably fast reading style.
The instructions given to the reader are designed to establish a
set towards the colloquial end of the reading style; but the effect
is slight, since people have little conscious control over their use of

the variables in reading style. The actual content of the test is more
influential. has been found in the construction of a number of such
It

readings that a text which is written as a narrative of a teenage boy

seems to lend itself to the least artificial performance of most people.
In such a framework, it was possible to incorporate such phrases
as, "He was a funny kid, aU'right." Elderly women might balk at

such a phrase if it were placed in the mouth of an adult, but as the

utterance of a teenage boy, it made natural reading for them.
The content of the readings carries this point further by focusing
on two main themes: the teenager's traditional protest against the
restrictions of the adult world, and his exasperation at the foibles
and inconsistencies of the girls he dates. In this context, adult readers
find it easy to handle colloquial phrases like "got her finger in the
pie," which they might not use in their own speech.
The first reading, "When I was nine or ten ..." consists of five
paragraphs in which the chief variables are successively concen-
trated (Labov 1966a:597). The first paragraph is a zero section, in
which none of the variables being studied are to be found. The
second paragraph concentrates (oh), beginning "We always had
chocolate milk and coffee cake around four o'clock." (Occurrences
of the variable are italicized here, but not in the actual text used.)
The third is concerned with (eh), as in "One man is IT: you run past
him as fast as you can, and you kick a tin can so he can't tag you."
82 SOCIOLINGUISTIC PATTERNS

The fourth concentrates on (r), as in "He darted out about four feet

before a car, and he got hit hard." The last paragraph has a high
concentration of (th) and (dh), e.g. "There's something strange about
that— how I can remember everything he did this thing, that thing, —
and the other thing." The text has a double purpose. First, it allows
us to measure in Context C the speaker's use of all five variables
as efficiently as possible. The close juxtaposition of many examples
gives us a fatigue factor not present in word lists, which differentiates
the speaker's use of a recently learned "superposed" form from the
vernacular forms produced without effort. Secondly, this reading
contains the sentences that are used in the Subjective Reaction Test
is given in Ch. 6). The subjects who have read the text
(the full text
themselves will be clear when they hear others read them that they
are judging the form of speech rather than the content.
The second reading, "Last Saturday night I took Mary Parker to
the Paramount Theatre ," is designed to juxtapose a number of
. . .

words that form minimal pairs, including those involving the phono-
logical variables studied in "When I was nine
pairs or ten ..." The
are italicized in the text given below, but not, of course, as the
informant reads them.

Last Saturday night I took Mary Parker to the Paramount Theatre. I would
rather have gone to see the Jazz Singer myself, but Mary got her finger in
the pie. She hates jazz, because she can't carry a tune, and besides, she
never misses a new film with Cary Grant. Well, we were waiting on line
about half an hour, when some farmer from Kansas or somewhere asked
us how to get to Palisades Amusement Park.
Naturally, I told him to take a Authority Garage on 8th
bus at the Port
Avenue, but Mary right away said no, he should take the I.R.T. to 125th
St., and go down the escalator. She actually thought the ferry was still

running.
"You're certainly in the dark," I told her. "They tore down that dock ten
years ago, when you were in diapers."
"And what's the source of your information, Joseph?" She used her
sweet-and-sour tone of voice, like ketchup mixed with tomato sauce. "Are
they running submarines to the Jersey shore?"
When Mary starts to sound humorous, that's bad: merry hell is sure to
break loose. I remembered the verse from the Bible about a good woman
being worth more than rubies, and I bared my teeth in some kind of a smile.
"Don't tell this man any fairy tales about a ferry. He can't go that way."
"Oh yes he can/" she said. Just then a little old lady, as thin as my
grandmother, came up shaking a tin can, and this farmer asked her the same
The Isolation of Contextual Styles 83

question. She told him to ask a subway guard. My god/ I thought, that's
one sure way to get lost in New York City.
Well, I managed to sleep through the worst part of the picture, and the

stage show wasn't too hard to bear. Then wanted to go and have a bottle
I

of beer, but she had to have a chocolate milk at Chock Full O'Nuts. Chalk
thisup as a total loss, I told myself. I bet that farmer is still wandering
around looking for the 125th St. Ferry.

In this reading, the minimal contrasts are brought as close together

as possible, under comparable stress, so the analyst can compare
their pronunciation without editing, but naturally enough so that the
reader is not aware of making the contrast overtly. The examples
with (r) illustrate the technique. In "You're certainly in the dark! They
tore down that dock" we can determine if the contrast of dock and
dark is by length alone [a— a:] or by length and backing [a— d:]. In
"she told him to ask a subway guard. My god! I thought" we have
close to the optimum juxtaposition of guard to god, which can be
identical, or differ in any of the three ways shown above. Less elegant
is the collocation of "source of your information" with "tomato

sauce". Here /ohr/ in source is compared to /oh/ in sauce; unless

the /r/ is realized, these two words are generally reported as homo-
nyms. 3 In these three cases, we have an opportunity to observe the
careful but unreflecting use of /r/ to differentiate words which
otherwise can be homonyms, and we make a direct comparison with
the same contrast in minimal pairs (see below). This reading also
gives us potential contrasts of /n — rjg/ in Singer^ finger,
/ehrV— erV— aerV/ in Mary— merry and Cary^carry, and
/airy— ferry, /ehr— ihr/ in bear— beer, /en— in/ in ten— tin,
/oy— ahr/ in voice— verse, /9— t/ in thin^tin, /aeh— ehr/ in
bad^bared, /ohr— uhr/ in shore— sure, /ae— aeh/ in can[N]—
can[AUX], and /a— oh/ in chock^chalk~~chocolate.
The style used in reading under Context C will be designated
Style C.

3. Our recent spectre-graphic studies of this data show that source and sauce are
usually not homonyms, even though the speaker thinks so and reports them as "the
same." The second formant of the nucleus of the vowel in source is usually lower,
(further back in terms of the normal articulatory correlate), and in connected speech
the first formant may also be lower (that is, the vowel is higher). During the minimal
pair test, the vowels are brought closer together, but second formant differences persist.
The phonetic differentiation of these nuclei is the same as that normally found in
r-pronouncing dialects.
84 SOCIOLINGUISTIC PATTERNS

Context D. Word Lists

A further step in the direction of a more formal context is to

consider the subject's pronunciation of words in isolation. There are
three types of word lists which are used for the investigation of the
variables (r), (eh), and (oh). One is a list which the subject knows
by heart: the days of the week and the months of the year. A second
type is a printed list of words with the same or similar segment. One
of these contains the (eh) variable, alternating lax with tense. A
reading pattern which followed the basic vernacular for this word
list would show:

Lax Tense Lax Tense

bat can
bad half
back past
bag ask
batch dance
badge have
bath has
bang razz «- razz
pat jazz - jazz
pad hammer
pass hamster
pal fashion
cash national
family ~~ family

This list therefore gives us, first, the height of the vowel in formal
pronunciation of the tense forms, and second, any disturbance
through social correction of the New York City vernacular form of
the tensing rule. 4 The (oh) list has no such complexity, since the
raising rule affects all members of the /oh/ and /ohr/ class. One
—
member of the /a/ class chock is included in that list: Paul, all,
ball, awful, coffee, office, chalk, chocolate, chock, talk, taught, dog,
forty-four.
The third type of word list continues the phonemic investigation
begun in the "Last Saturday night I took Mary Parker ." reading. . .

4. For a detailed study of this rule, see Cohen The Lower East Side study
1970.
was concerned with the extent of raising of the tense vowel, and not the selection
of environments by the tensing rule. Variation in the latter seems to be immune from
social correction, and shows geographic and idiolectal variation of a very complex
nature, controlled to a degree by the implicational ordering of the environments.
The Isolation of Contextual Styles 85

The subject is shown a list of words containing most of the minimal

pairs which occurred in that reading, and a few more:

dock dark
pin pen
guard god
"I can" tin can

The subject is asked to read each pair of words aloud, and then say
whether they sound the same as or different from the way he usually
pronounces them. Thus in addition to the unreflecting contrasts of
Style C, we have the subject's considered performance in Style D,
and his subjective reaction to that performance. Eventually, all of
this data is to be used for a structural analysis of the system; here
the mean values of the variables in the word lists (except (r) in
minimal pairs —
see below) give us the index values for Style D.

Context D'. Minima] Pairs

For the variable (r), it is useful to extend the spectrum of formality
one stage further. In the word lists of Context D, (r) occurs in two
situations. In one, the pronunciation of (r) is seemingly incidental,
as in the reading of hammer and hamster in the (eh) list, or the names
of the months ending in -er, or with such minimal pairs as finger
and singer, mirror and nearer. Here (r) is pronounced in the formal
context of a word list, but it does not receive the full attention of
the reader. But in minimal pairs such as dock and dark, guard and
god, source and sauce, bared and bad, (r) is the sole differentiating
element, and it therefore receives maximum attention. We will
therefore single out this subgroup of Style D for (r) as Style D'.

The Problem of Casual Speech

Up to this point, we have been discussing techniques for extending

the formal range of the interview by methods which fall naturally
into the framework of a discussion about language. But even within
the interview, we must go beyond the interview situation if we can.
We must somehow become witnesses to the everyday speech which
the informant will use as soon as the door is closed behind us: the
style inwhich he argues with his wife, scolds his children, or passes
the time of day with his friends. The difficulty of the problem is
86 SOCIOLINGUISTIC PATTERNS

considerable; yet the rewards for its solution are great, both in
furthering our present goal, and in the general theory of stylistic
variation.
First, it is important to determine whether we have any means of
knowing when we have succeeded in eliciting casual speech. Against
what standard can we measure success? In the course of the present
study of New York City speech, there are several other approaches
to casual speech that have been used. In the exploratory interviews,
I recorded a great deal of language which is literally the language

of the streets. This material included the unrestrained and jubilant

activity of a great many small children, and also some recordings
of street games among young men, 18 to 25 years old, where I was
an anonymous bystander. It may be that none of the conversation
within the interview will be as spontaneous and free as this material.
But if the informants show a sudden and marked shift of style in
this direction, we will be justified in calling this behavior casual
speech.
Another check is random and anonymous observation such as the
department-store survey discussed in Ch. 2, in which the bias of the
linguist's presence disappears completely. Here we can judge
whether the type of alternation which is found within the interview
gives us a range of behavior comparable to that which is found under
casual conditions in everyday life.
The immediate problem, then, is to construct interview situations
in which casual speech will find a place, or which will permit
spontaneous speech to emerge, and then set up a formal method for
defining the occurrence of these styles. By casual speech, in a narrow
sense, we mean the everyday speech used in informal situations,
where no attention is directed to language. Spontaneous speech
refers to a pattern used in excited, emotionally charged speech when
the constraints of a formal situation are overridden. Schematically:

Context: Informal Formal

Style: Casual Careful/Spontaneous

We do not normally think of "spontaneous" speech as occurring in

formal contexts: yet, as we will show, this frequently happens in
the course of the interview. Spontaneous speech is defined here as
the counterpait of casual speech which does occur in formal con-
texts, not in response to the formal situation, but in spite of it.
While there is no a priori reason to assume that the values of the
The Isolation of Contextual Styles 87

variables will be the same

spontaneous as in casual speech, the
in
results of this investigation show
that they can be studied together.
At a later point, as we examine more deeply the mechanism of
stylistic variation, it will be possible to suggest an underlying basis
for this identification. For the moment, either term will be used
according to the nature of the context, but they will both be meas-
ured under the heading of Style A, or casual speech in general.
The formal definition of casual speech within the interview re-
quires that at least one of five contextual situations prevail, and also
at least one of five nonphonological cues. We will first discuss
the contextual situations, which will be identified as Context A t
through A 5 .

Context Av Speech Outside the Formal Interview

There are three occasions within the larger context of the interview
situation which do not fall within the bounds of the formal interview
proper, and in these contexts, casual speech is apt to occur.
Before the interview proper begins, the subject may often address
casual remarks to someone else in the household, his wife or his
children, or he may make a few good-natured remarks to the inter-
viewer. Although this is not the most common context for a good
view of casual speech, the interviewer will not hurry to begin formal
proceedings if there seems to be any opportunity for such an ex-
change. In several cases, where a housewife took time to wash the
dishes, or a family to finish dinner, the interviewer overheard casual
speech in some quantity.
After the interview begins, there may be interruptions, when
someone else enters the room, or when the informant offers a glass
of beer or a cup of coffee. In the following example, the three
paragraphs represent, 1, speech in the formal interview directly
before the break, 2, speech used while opening a can of beer for
the interviewer, and 3, the first sentences spoken on the resumption
of the formal interview.

1 If them the same. There

you're not careful, you will call a lot of
are a couple ofthem which are very similar; for instance, width
and with. [What about guard and god?l That's another one you
could very well pronounce the same, unless you give thought
to it.
88 SOCIOLINGUISTIC PATTERNS

2 ... these things here —

y'gotta do it the right way otherwise —
[laughter] you'llneed a pair of pliers with it You see, what . . .

actually happened was, I pulled it over to there, and well ... I

don't really know what happened Did it break off or get stuck
. . .

or sump'm?
. . . just the same as when you put one of these keys into a can of
sardines or sump'm — and you're turning it, and you turn it lop-
sided, and
end you break it off and you use the old fashioned
in the
opener always have a spoon or a fork or a screw driver
. . . but I

handy to wedge into the key to help you turn it [laughter] I . . .

always have these things handy to make sure.

3 [How do you make up your mind about how to rate these people?]
Some people — suppose perhaps it's the result of their training and
I

the kind of job that they have — they just talk in any slipshod
manner. Others talk manner which has real finesse to it, but
in a
that would be the executive type. He cannot [sic] talk in a slipshod
manner to a board of directors meeting.

The shift in style from 1 to 2 and back to 3 is quite evident even

in conventional orthography. The prosodic channel cues, and the
phonological variables point in the same direction as the shifts in
lexicon, syntax, and content.
The interviewer may make every use of this opportunity by mov-
ing away from his chair and tape recorder, and supporting the
emergence of casual conversation. One great advantage of such a
break is that it occurs in close juxtaposition with very careful speech,
and the contrast is very sharp, as in this example given above. The
sudden occurrence of radically different values of the variables is
particularly marked in this example. The word otherwise in extract
2 has (dh) in medial position; this is rarely [d] in the careful speech
of this subject, but [d] does occur here and makes a sharp impression
on the listener.
The most frequent place for casual speech to emerge in Context
A 1
is at the end of the interview. It is perhaps most common when

the interviewer has packed away his equipment, and is standing with
one hand on the door knob. 5

5. The interviewer is not a passive agent in any of these circumstances. By his

participation in the developing informality, he can help casual speech to emerge. At
the termination of the interview, he can also terminate his role as interviewer, and
behave like any other tired, hot, or sleepy employee who has now finished his job
and is free to be himself.
The Isolation of Contextual Styles 89

Context A2 . Speech with a Third Person

At any point in the interview, the subject may address remarks
to a third person and casual speech may emerge. One of the most
striking examples occurred in an interview with a black woman, 35,
raised in the Bronx, and then living on the Lower East Side in poor
circumstances as a widow with six children. The following three
sections illustrate the sharp alternations between the careful, quiet,
controlled style used in talking to the interviewer, and the louder,
higher-pitched style used with her children. Again, the grammatical
and stylistic differences shown in conventional orthography illustrate
the shift of style.

1 ... Their father went back to Santo Domingo when they had
the uprising about two years ago that June or July ... he got
killed in the uprising ... I believe that those that want to go
and give up their life for their country, let them go. For my part,
his place was here with the children to help raise them and give
them a good education .that's from my point of view.
. .

2 Get out of the refrigerator, Darlene! Tiny or Teena, or whatever

your name is! . Close the refrigerator, Darlene!
. . . What
. .

—
pocketbook? I don't have no pocketbook if he lookin' for money
from me, dear heart, I have no money!

3 I thought the time I was in the hospital for three weeks, I had

peace and quiet, and I was crying to get back home to the
children, and I didn't know what I was coming back home to.

Interruptions of the interview by telephone calls sometimes pro-

vide unusually good opportunities to study casual speech. In one
interview, the telephone interrupted the proceedings at the very
middle. The informant, Dolly R., had just returned from a summer
spent in North Carolina, and one of her cousins was anxious for news
of the family. I left the room with her nephew, and continued to
talk to him quietly in another room; for twenty minutes, the inform-
ant discussed the latest events in a very informal style, and we thus
obtained an excellent recording of the most spontaneous kind of
speech. The contrast is so sharp that most listeners cannot believe
it is the same person talking. The style that Dolly R. used with me

was friendly, relaxed, seemingly informal and casual: in talking

about common sense she said:
90 SOCIOLINGUISTIC PATTERNS

Smart? Well, I mean, when you use the word intelligent an' smart
I mean
—
you use it in the same sense? (Laughs) So some . . .

— —
people are pretty witty I mean yet they're not so intelligent!

Although the laughter and informality of this passage would seem

to place it in a "casual" category, no absolute judgment can be made

without contrasting it with other styles. And the values of the lin-
guistic variables are suspiciously remote from the vernacular (r) —
is almost consistently [r], and there is only one nonstandard (dh),

in they're. Here on the other hand is a passage from the telephone

conversation:

5 Huh? . . . Yeah, go downThis is. So you know what

'e(r)e to stay.

Carol Ann what Carol Ann say. Carol Ann say,

say? Listen at
'An' then when papa die, we can come back" [belly laugh] . . .

Ain't these chillun sump'm [falsetto]? An' when papa die, . . .

can we come back? [laughs]. . . .

The laughterof this passage is very different from 4: it is a full-bodied

performance that begins low and ends high, shaking from somewhere
down deep. Listening to it, we realize that the laughter of 4 is forced
—
by comparison a "Ha ha ha" drawn from a white repertoire. The
voice quality and personality of 5 are also very different, and the
intonation contours are dramatically opposed. 6 The phonological and
grammatical variables are altogether different. The contrast is so
dramatic in the case of Dolly R. that we are forced to recognize the
limitations of our other methods of eliciting the vernacular: for some
speakers, at least, our best techniques within the interview situation
way toward the vernacular but there
will shift the speaker part of the
no
is guarantee that we have covered the major part of the distance.
We have defined a direction but not the destination.

Context A3 . Speech Not in Direct Response to Questions

In some types of interview schedules, it is necessary to cut off long,
rambling replies, or sudden outbursts or rhetoric, in order to get

6. We used these two passages cited here in a Family Background test in our
interviews with adults in south-central Harlem (Labov et al. 1968 2:4.7). Many of the
subjects were acutely embarrassed by 5; they shifted in their chairs as they listened.
They assumed, naturally, that it was a performance done to order for the tape recorder,
and for anyone to use this intimate family style in such a public situation is clearly
playing "Uncle Tom." They could not know, of course, that Dolly R. did not realize
at the time that she was being recorded, and that she assumed that the conversation
she heard from the other room was the interview proper.
The Isolation of Contextual Styles 91

through with the work. In this interview program, the opposite policy
prevailed. Whenever a subject showed signs of wanting to talk, no
obstacle was interposed: the longer he digressed, the better chance
we had of studying his natural speech pattern. Some older speakers,
in particular, pay little attention to the questions as they are asked.
They may have certain favorite points of view which they want to
express, and they have a great deal of experience in making a rapid
transition from the topic to the subject that is closest to their hearts.
Context A 3 forms a transition from those contexts in which casual
speech is formally appropriate, to those contexts in which the emo-
tional state or attitude of the speaker overrides any formal restric-
tions, and spontaneous speech emerges.

Context A4 . Childhood Rhymes and Customs

This is one of the two topics within the interview itself which is
designed to provide the context in which spontaneous speech is
likely to emerge. The atmosphere or tone required for such a shift
is provided by a series of questions which lead gradually to the topic

of jump-rope rhymes, counting-out rhymes, the rules of fighting, and

similar aspects of language drawn from the preadolescent period
when the youngster participates in a culture distinct from that of
adult society. Rhymes, for example, cannot be recited correctly in
Style B of careful conversation. Both the rhyme itself, and the tempo,
would be wrong if Style B were used in

Cinderella,
Dressed in yella
Went downtown to buy some mustard,
On way her girdle busted,
the
How many people were disgusted?
10, 20, 30 . . .

The following song, which is popular in New York City schools, does
not permit the r-pronunciation which would creep into Style B:

Glory, glory, Hallelujah,

The teacher hit me with the ruler,
The ruler turned red,
And the teacher dropped dead,
No more school for me.
Equally r-less pronunciation is implied in the traditional
92 SOCIOLINGUISTIC PATTERNS

Strawberry short cake, cream on top

Tellme the name of your sweetheart . . ?

If the compulsion of these rhymes demanded a return to a childhood

pronunciation which was no longer normal, their use as evidence

would be wrong. However, the pattern which is used in Context A 4
is quite comparable to that which is used in the four other contexts

which are utilized. There is no necessity for the following rhyme

to assume any particular value of (oh), yet (oh-1) is very common:

I won't go to Macy's any more, more, more

There's a big fat policeman at the door, door, door,
He you by the collar
pulls
And makes you pay a dollar,
I won't go to Macy's any more, more, more.

The nine examples of (oh) in this rhyme provide an efficient means

of studying that variable.
Even in counting-out rhymes, where meter and rhyme are less
compelling for the informant, we find that Style B is inadequate for

My mother and your mother were hanging out the clothes,

My mother punched your mother right in the nose.
What color blood came out?
[Green.] G-R-E-E-N spells green and you are not IT.

or for the much simpler

Doggie, doggie, step right out.

Men as well as women will be able to repeat counting-out rhymes

such as "Eeny meeny miny moe," or "Engine, engine, number nine."
Lacking this, spontaneous speech is often obtained from men in the
rules for playing marbles, the complex New York City game of
skelley, punchball, or stickball.

Context A5 . The Danger of Death

Another series of questions in a later section of the interview leads
to the following:

7. The acceptable half-rhyme used here implies a pronunciation of -heart as [hat],

with a fairly short vowel. Such shortenings are not rare in the city, especially in
polysyllables.
The Isolation of Contextual Styles 93

Have you ever been in a situation where you thought

you were in serious danger of being killed— where
you thought to yourself, "This is it?"

If the informant answers yes, the interviewer pauses for one or two
seconds, and then asks, "What happened?" As the informant begins
to reply, he is under some compulsion to showthat there was a very
real danger of his being killed; he stands in a very poor light if it
appears that there was no actual danger. Often he becomes involved
in the narration to the extent that he seems to be reliving the critical
moment, and signs of emotional tension appear. One such example
occurred in an interview with six brothers, from 10 to 19 years old,
from a lower-class Irish-Italian household. While most of the boys
had spoken freely and spontaneously in many contexts, the oldest
brother, Eddie, had been quite reserved and careful in his replies.
He had given no examples of casual or spontaneous speech until
this topic was reached. Within a few short sentences, a sudden and
dramatic shift in his style took place. At the beginning of Eddie's
account, he followed his usual careful style:

6 IWhat happened The school I go to is Food and Mari-

to you?]
time — that's — and I was up in the masthead,
maritime training
and the wind started blowing. I had a rope secured around me
to keep me from falling — but the rope parted, and I was just
hanging there by my fingernails.

At this point, the speaker's breathing became very heavy and irregu-
lar; his voice began to shake, and sweat appeared on his forehead.
Small traces of nervous laughter appeared in his speech.

7 I never prayed to God so fast and so hard in my life [What . . .

happened?] Well, I came out all right Well, the guys came
. . .

up and they got me. [How long were you up there?] About ten
minutes. [I can see you're still sweating, thinking about it.] Yeh,
I came down, I couldn't hold a pencil in my hand, I couldn't

touch nuttin'. I was shakin' like a leaf. Sometimes I get scared

thinkin' about it but
. . uh
. well, it's training.
. . . .

All of the phonological variables in 7 shift towards the forms most

typical of casual speech, including (th), (dh), and (ing). At the very
end, Eddie returns to his careful style with an effort: "Well, it's

training!" The effect of probing for the subject's feelings at the

94 SOCIOLINGUISTIC PATTERNS

moment of crisis can be effective even with speakers who are quite
used to holding the center of the stage. One of the most gifted story
tellers and naturally expressive speakers in the sample was Mrs.
Rose B. She was raised on the Lower East Side, of Italian parents;
now in her late 30's, she recently returned to work as a sewing
machine operator. The many examples of spontaneous narrations
which she provided show a remarkable command of pitch, volume
and tempo for expressive purposes.
8 —
... And another time that was three times, and I hope it never
—
happens to me again I was a little girl, we all went to my aunt's
farm right near by, where Five Points is and we were thirteen . . .

to a car. And at that time, if you remember, about 20 or 25 years

ago, there wasn't roads like this to go to Jersey — there was all
dirt roads. Well, anyway, I don't know how far are I don't —
remember what part we were one of the wheels of the car came —
off— and the whole car turned, and they took us all out. They
hadda break the door off. And they took us out one by one. And
I got a scar on my leg here 'ats the on'y thing [When. . . . . .

the car turned over, what did you think?]

... it was upside dow— you know what happened, do you
know how remember anything. This is really the
I felt? I don't
truth — till could tell that to anybody, 'n' they don't
today, I

believe me, they think I'm kiddin' 'em. All I remember is —

thought I fell asleep, and I was in a dream ... I actually saw
stars . you know, stars in the sky y'know, when you look
. .
—
up there and I was seein' stars. And then after a while, I
. . .

felt somebody pushing and pulling —

you know, they were all on
top of each other — and they were pulling us out from the bottom
of the car, I was goin' "Ooooh."
and
And when came — you know — to, says to myself, "Ooooh,
I I

we're in a car accident,"— and that's all I remember — as clear

as day — don't remember the car turning or anything. All I know
I

is I thought I went to sleep. I actually felt went to sleep. I

Channel Cues for Casual Speech

The five contexts just described are only the first part of the formal
criteria for the identification of Style A in the interview. It is of
course not enough to set a particular context in order to observe
casual speech. We also look for some evidence in the type of lin-
guistic production that the speaker is using a speech style that
The Isolation of Contextual Styles 95

contrasts with Style B. To use phonological variables would involve

a circular argument, because the values of these variables in Styles A
and B are exactly what we are trying to determine by the isolation
of styles. The best cues are channel cues: modulations of the voice
production which affect speech as a whole. 8 Our use of this evidence
must follow the general procedure of linguistic analysis: the absolute
values of tempo, pitch, volume, and breathing may be irrelevant, but
contrasting values of these characteristics are cues to a differentiation
of Style and Style B. A change in tempo, a change in pitch range,
A
a change volume or rate of breathing, form socially significant
in
signs of shift towards a more spontaneous or more casual style of
speech.
Whenever one of these four channel cues is present in an appro-
priate context, the utterance is marked and
which contains them
measured under Style A. The
channel cue is another modulation
fifth
of voice production: laughter. This may accompany the most casual
kind of speech, like the nervous laughter in the example from Eddie
D., and is frequently heard in the description of the most dramatic
and critical moments in the danger-of-death narration. Since laughter
involves a more rapid expulsion of breath than in normal speech,
it is always accompanied by a sudden intake of breath in the follow-

ing pause. Though this intake is not always obvious to the listener
in the interview situation, the recording techniques being used in
this study detect such effects quite readily; it is therefore possible
to regard laughter as a variant type of changes in breathing, the
fourth channel cue. 9
The question now arises, what if a very marked constellation of
channel cues occurs in some Context B? Intuition may tell us that
this is spontaneous speech, but the formal rules of this procedure
instruct us to consider it Style B. This is a necessary consequence
of a formal definition. The situation may be schematized in this way:

8. These would be considered modifications of the Message Form rather than the

Channel in the terminology used by Dell Hymes (1962). In the framework suggested
by Hymes, the more formal styles of reading would represent a shift in the channel;
the elicitation of casual speech would be encouraged by shifts in the Setting and Topic,
and the phonological variables appear as variations in the Code.
9. The case of Dolly R., noted above, shows that we may also have to take laughter

as a contrastive cue —a change in the form of laughter is more important than laughter
itself.
96 SOCIOLINGUISTIC PATTERNS

intuitive Careful speech Casual speech

observations
formal definition Style B Style A
and measurement

As this diagram indicates, Style B formally defined overlaps casual

speech intuitively observed. Some examples of casual speech will
occur outside the five contexts given, conditioned by some less
prominent context we have not considered, and these will be lost
by the formal definition. However, since the body of careful speech
bulks much larger than casual speech, this small amount of compar-
atively casual speech now included under Context B and Style B
will not seriously distort the values for careful speech. If, on the other
hand, there should be overlap in the other direction, with a definition
which specified the contexts of careful speech, the resulting admix-
ture in the smaller bulk of casual speech would be a source of serious
distortion. By leaving careful speech as the unmarked category, we
are protected from such distortion.
What are the actual proportions in our material of casual and
careful speech as defined? We can obtain a good estimate from the
records which show the total occurrences of (r) and (dh) in the styles
as defined above, since these very frequent variables give us a
measure of the total volume of speech. Ten percent of the adult
interviews from the Lower East Side survey were randomly selected,
and the relative volume of speech in each style was measured by
a combined index of the total incidence of (r) and (dh). 10 The average
percentages for this sample are, for Style A, casual speech: 29 per-
cent; and for Style B, careful Speech: 71 percent.
An alternative approach to identifying casual speech would have
been to rely only upon channel cues, without reference to the con-
text. This would have been far less reliable, for in many contexts,
the channel cues vary continuously, and to determine where contrast

10. The use of (dh) or (r) alone would have produced serious bias. For some speakers,
primarily lower-class white and black speakers, (r) is not a variable, and is not
recorded as such on the transcription forms. For others, primarily middle-class
speakers, (dh) is always a fricative, and is not tabulated. There are no speakers in
the sample for whom neither of these features is a variable. It is interesting to note
that the (dh) variable gives a somewhat higher percentage for casual speech: 33 percent
as against 26 percent for (r). This is probably a reflection of the greater spontaneity
and more casual approach of many working-class speakers.
The Isolation of Contextual Styles 97

occurred and where it did not would have often been very difficult.
The interview as now constructed provides for sudden shifts of
contexts which have sharp boundaries. These shifts thus enable us
to observe sudden contrasts in the channel cues. Another alternative
would be to adopt certain sections of the interview as casual speech,
without regard to channel cues or any other measure. Obviously
this would weaken our approach to the vernacular, since there is
no technique which is certain to relax the constraints of the inter-
view situation for everyone. 11 It is not contended that Style A and
Style B are natural units of stylistic variation: rather they are for-
mal divisions of the continuum set up for the purposes of this study,
which has the purpose of measuring phonological variation along
the stylistic axis. The discovery of natural breaks in the range of
stylistic phenomena would have to follow a very different procedure.
It isnot unlikely that results of the present work, yielding sensitive
indexes to linguistic variation, may eventually be applied to this end.
The validity of this method may be tested by a comparison with
other means of recording casual speech. A number of anonymous
observations in public places were made on the Lower East Side
which match quite closely the characteristics of casual speech as
obtained in the interview (see Appendix B in Labov 1966a for the
Punch Ball Game and the Lunch Counter). We can also approach
validation and explanation from the experimental direction. Mahl
has conducted a series of studies on the effect of removing subjects'
ability to monitor speech (1972). This was done by feeding random
noise through earphones at a volume high enough to prevent the
subject from hearing his own speech. In addition, the subject was
sometimes facing away from the interviewer so that he could not
see the interviewer's face. The speech of each subject was then
studied during three interviews under four conditions: with white
noise, facing or not facing; and without white noise, facing or not
facing. In many cases the loss of audio-monitoring produced sharp
changes in pitch, volume, intonation, and in the length of responses;
in several cases, there were changes in speech patterns which seemed
to Mahl to resemble differences in social dialects. In cooperation with
Mahl, I examined some of these recordings to see if the style shifting

11. This is in fact the approach taken by Trudgill (1971). Whatever weaknesses this
technique may have, it did not prevent Trudgill from showing a regular differentiation
between Casual and Careful speech.
 98 SOCIOLINGUISTIC PATTERNS

could be measured objectively by linguistic variables. An explora-

tory study of other New Haven speakers developed a list of socially
significant variables: the most important of these for the style shifting
of Mahl's subjects was (dh). Fig. 3.2 shows the percentage of the
standard fricative form [5] in the speech of one subject whose style
shifting was most striking. The horizontal axis shows the series of
three interviews, with the four different conditions in the order that
they were administered. There is an overall familiarization effect,
in which the percentage of standard forms drops steadily. The four
conditions are also clearly differentiated: both the loss of audio-
monitoring and the loss of visual monitoring of the addressee in-
terfered with the subject's control over the (dh) variable. We can
infer that a consistent production of fricative forms is part of a
superposed dialect not characteristic of this speaker's early ver-
nacular pattern, and requires a certain amount of attention paid to
speech which is facilitated by audio-monitoring of the self and
stimulated by visual monitoring of the other.
The parallel with style shifting of (dh) in our interviews is quite
striking. We note another phenomenon in our interviews which

Facing interviewer, no noise

Facing away, no noise

100
Facing interviewer, white noise

Facing away, white noise

90

80

3 70

60 -

50
Interview 1 Interview 2 Interview 3

Fig. 3.2. Effect of white noise and orientation on one subject's

use of (dh).
The Isolation of Contextual Styles 99

points in this direction. We ask subjects to start to say ten and tell

us where the tip of their tongue is. Even though this occurs in the
formal context of the discussion of speech, the forms used in this
reply are shifted strongly towards casual speech. Attention directed
to the location of the tongue seems to interfere with attention to
articulation in the answer that follows.
We can therefore put forward the hypothesis that the various styles
of speech we are considering are all ranged along a single dimension
of attention paid to speech, with casual speech at one end of the
continuum and minimal pairs at the other. If future research suc-
ceeds in confirming this hypothesis, and quantifying attention paid
to speech, we will then have a firmer foundation for the study of
style shifting,and more precise relations can be established in the
study of sociolinguistic structures as a whole.

The Array of Stylistic Variation

Given the techniques for isolating styles outlined above, we can

now ask how this stylistic dimension correlates with our dependent
variables. For this purpose we can lay out the following array of
the five main phonological variables:

Word Minimal
Casual Careful Reading lists pairs
Variable A B C D D'

(0 X > X > X > X > X

(eh) X > X > X > X >
(oh) X > X > X > X
(th) X < X < X
(dh) X < X < X

A separate style D' for minimal pairs is shown only for (r). The (th)
and (dh) variables are not studied in word lists, but only in reading
style. We then have altogether 19 points where the mean values of
the variables can be placed in a stylistic array. If their use is cor-
related with the stylistic continuum as we expect, we should find
that the first three are at a maximum for Style A, and decline steadily
for B, D, C, and D'; and the last two are at a minimum for A and
rise regularly for B and C.
The first native New Yorker to whom this method was applied
100 SOCIOLINGUISTIC PATTERNS

was Miss Josephine P., 35, who lived with her Italian-born mother
in the same Lower East Side tenement apartment where she was
born. Miss P. attended high school on the Lower East Side, and
completed almost four years of college. At the time of the interview,
she worked as a receptionist at Saks 5th Avenue. Josephine P.'s style
of speech is lively and rapid; she seems to be an outgoing person
who has no difficulty in making friendly contact with strangers. Her
careful conversation, in Context B, seems at first to be equivalent

to the casual conversation of most speakers. Two short samples of

casual speech were recorded, which contrasted with her speech in
Context B. We thus have the complete array of average values of
the variables for this speaker.

STYLISTIC ARRAY FOR JOSEPHINE P.

Variable A B C D D'

(r) 00 03 23 53 50
(eh) 25 28 27 37
(oh) 21 23 26 37
(th) 40 14 05
(dh) 34 09 09

The (r) values for Josephine P. rise from 00 to 50 as we would

expect; (eh) and (oh) also though B and C are essentially the
rise,
same for (eh). There is a sharp upturn in Style D which is generally
characteristic of lower-middle-class speakers (see Ch. 5 in this vol-
ume). The (th) and (dh) variables are at a low level in Style A, and
fall to a very low point in more formal styles, as we would expect
from a speaker of her background.
The two sections of casual speech which were recorded in contrast
to Style B occurred in Context A v extra-interview. In one section,
Josephine P. talked with some emotion about her dead father, as she
remembered him from her childhood, and about the dolls he brought
her from the factory where he worked. The associated channel cues
were laughter, increase in tempo, and a change in the rate of breath-
ing. The second section was a burst of irritation at the behavior of
other tenants in the building, with increased pitch and volume. Both
of thesewere recorded after the interview, as I sat having coffee with
Josephine P. and her mother.
In the course of a normal dialectological interview, the whole
conversation of Josephine P. would have been accepted as free and
The Isolation of Contextual Styles 101

spontaneous; but since the present procedure assumes that the

speech of Context B cannot be truly casual, all of the contexts
relevant to Style A were examined. The emergence of a very differ-
ent speech pattern in the measurements of the five variables under
—
Style A especially for (th) and (dh) —
confirms our expectation.
Without the sections of casual speech, we would have to report that
Josephine P. rarely used affricates or stops for these variables.
In the overall pattern, there are two departures from the expected
array, both less than 5 percent in magnitude. This is remarkable
when we consider the irregular fluctuations of the variables that
seem to mark the individual sections of speech. For example, here
are the occurrences of (th) in casual speech, in the order that they
occurred: 1221221111; and here a continuous series in careful speech:
221111111111112121. There seems to be no pattern or system within
this sequence —yet it fits into the larger pattern shown in the array
of styles. The total number of items upon which the array of Jose-
phine P. was based is not large; a relatively small number of occur-
rences establish the progressions, despite the variations within each
style. The number of instances in each cell are given in the frequency
array.

FREQUENCY ARRAY FOR JOSEPHINE P.

Variable A B C D D'

(r) 18 66 44 15
(eh) 4 4 28 13
(oh) 10 11 19 11
(th) 10 29 20
(dh) 26 65 35

This array of frequencies shows three weak points, at (r) D', and at
(eh) A and B, where there were only four occurrences of the variable
in each cell. This limitation of the data allows errors in perception
and transcription to affect the final result significantly, as well as
the inherent variation of the individual. If this array is now compared
with the table of the average values of the variables given on Jose-
phine P. above, it appears that the low points of frequency coincide
exactly with the points where small deviations from the overall
pattern were found. The implication of this finding is that if more
occurrences of (eh) A and B and (r) D' were introduced, the behavior
of the subject might be seen as perfectly regular.
102 SOCIOLINGUISTIC PATTERNS

The next New Yorker who was

interviewed by this procedure was
Abraham high-school graduate, native of the Lower East
G., 47, a
Side, of Polish-Jewish parents. He lived in a public housing project,
and drove a taxi for his regular income. In contrast to Josephine P.,
this informant was immediately and obviously a multiple-style
speaker. In Context B, he used a fluent but self-conscious style, which
reflected his experience in many committee meetings head of his
as
American Legion chapter. His Style B, which employed such phrases
as the armed forces for 'army' and fair and equitable for 'fair', was
obviously not his casual style. He even managed to tell several long
and exciting stories of near-hold-ups, in the danger-of-death section,
without losing the elevated manner of Style B. However, midway
through the interview, he stopped to offer me a can of beer, and
delivered the humorous monologue quoted on page 88, which is the
main basis for the Style A column in his array. The blank spot in
this array, at (th) A, is the point where the single occurrence of (th),
as a stop, could not be used for a rating. The only apparent irregu-
larity is the change of direction at (oh) D: as further studies showed,
this is not uncommon.

STYLISTIC ARRAY FOR ABRAHAM G.

Variable A B C D D'

(r) 12 15 46 100 100 -

36 39 40 8 60 39 7
(eh) 35
6 22 18 13
(oh) 10 18 29 20
3 11 16 11
(th) 17 00 1 20 20
(dh) 72 33 05 18 78 35

In most cases, the interview procedure isolates Style A in more

than one context. The case of Mrs. Doris H., 39, is typical. She is
black, raised on Staten Island, a high-school graduate; her husband
is a New York City policeman. Mrs. H. showed a wide range of

stylistic behavior, from the careful, well-reasoned, highly organized

replies in Context B, to sudden outbursts of spontaneous humor that
marked her as a person of considerable wit and charm. Her chart
shows spontaneous speech in Context A 2 (speech to a third person)
as she rallied her 13-year-old son on his tendency to show off; in
Context A 3 (not in direct response) as a long account of the tactless
behavior of some of her friends, with direct quotations; in four cases
within Context A 4 (childhood rhymes) and in Context A 5 (danger
The Isolation of Contextual Styles 103

of death). In these seven sections of Style A, the

most prominent
channel cues are sudden increase in volume, and laughter; occa-
sionally there was an increase in tempo and in rate of breathing.
The resulting array of the variables is quite regular in its left to right
progression except for (eh). Part of the reason for the irregularity
of (eh) is A
represented only by three vowels. We do
that Style is

find that small numbers

(r) in Style D' are usually quite regular,
of
even when there are only four instances. The overriding effect of
the formality of the context seems to provide quite uniform results.
But in all other contexts, three or four items seem to be insufficient
to provide values that fit into a regular array. This problem disap-
pears as we begin to sum the arrays of individuals to obtain values
for social groups. The other deviation at (eh) D, is based on sufficient
evidence, and indicates again that a reversal at (eh) D and (oh) D
is more common than a reversal in the pattern anywhere else. The

great range in (r-1) pronunciation which is seen here, from 00 to 100,

is a frequent characteristic of the linguistic class of speakers to which
Mrs. H. belongs — the lower middle class (see Ch. 5 in this volume).

Variable

(r)
ABC
STYLISTIC

00
ARRAY FOR DORIS

31 44
H.

DO'
69 100
(eh) 30 26 32 29 29 64 55 19
3 10 25 13
(Oh) 18 21 23 25
16 21 18 11
(th) 80 24 12 5 29 24
(dh) 50 22 28 85 42

The three New Yorkers considered so far are typical of the speech
community in their concern with language and their overt rejection
of the New
York vernacular. But the pattern of style shifting is not
directlygoverned by overt values; even when the explicit norms
expressed by the individual are reversed, the pattern is the same.
The case of Steve K. will illustrate this crucial point. He was a very
intense young man of 25, a copyreader's assistant, living in a fifth-
floor walk-up on the East Side. He had come to the Lower East Side
only three years before from Brooklyn, where he was raised, a
third-generation New Yorker. His grandparents were Jewish immi-
grants from Eastern Europe.
Steve K. might be considered a deviant case in many ways. He
studied philosophy for four years at Brooklyn College, but left
104 SOCIOLINGUISTIC PATTERNS

without graduating; he had turned away from the academic point

of view, and as an intense student of Wilhelm Reich, sought self-
fulfillment in awareness of himself as a sexual person. 12 His attitude
towards language was much more explicit than that of most people.
He was unique among the informants in being aware of all five of
the chief variables, and believed that he was able to control or at
least influence his own usage. He had consciously tried to reverse
his college-trained tendency towards formal speech, and to reinstate
the natural speech pattern of his earlier years. In other words, he
deliberately rejected the pattern of values reflected in the array of
numbers shown in the preceding examples; in his own words, he
wanted to "go back to Brooklyn."
Steve K.'s self-awareness and his set of values might prepare us
to find a radically different pattern in the array of the variables — if

we believed that the linguistic and social forces operating here are
subject to conscious manipulation. But as a matter of record they
are not. Except for the fact that the (th) and (dh) patterns operate
at a low level, his array is quite similar to that of Abraham G. The
only deviation from a regular progression is that at (eh) D.

STYLISTIC ARRAY FOR STEVE K.

Variable A B C D D'
N
(0 00 06 08 38 100
(eh) 28 33 34 30 37 70 49 16 3
6 16 25 13
(oh) 22 23 25 30
5 27 18 11
(th) 09 00 00 11 12 24
(dh) 15 06 05 34 55 42

For New Yorkers of Steve K.'s age, all of these variables will remain
variables in normal speech, no matter what conscious adjustments
are attempted. Not one speaker in the sample who was raised in New
York City was able to use 100 percent (r-1) in conversation, and this
includes a great many speakers who were consciously aiming in that
direction after (r) had been discussed. For example, Steve K. claimed
that his present performance was a deliberate step backward from
his college days, when he had pronounced all or most (r) as (r-1).
I then asked him to reread the r paragraph from "When I was nine

or ten," and pronounce all (r) as (r-1).

12. Steve K.'s definition of a successful man puts his point of view very concisely:
"a man who is fully aware of himself ... of his own sexuality and of his emotions
. . . who always knows what he feels towards each person he meets."
The Isolation of Contextual Styles 105

His first attempt was a complete failure, and his second no better.
I asked him to read a little more slowly. He continued and produced
an (r) index of 33. A third try produced a step upward to 45. A fourth
attempt gave 61, and in a fifth trial, he seemed to level off at 69. He
then confessed that he probably could not have pronounced that
much (r-1) when he was in college.
Steve K.'s inability to deal with a few sentences containing only
thirteen (r)'s suggests that the original reading score of 38 is probably
very close to the pattern which was solidified in his college days.
Despite his profound shift in ideology, the speech pattern dictated
by equally profound forces remains constant. It is not likely that he
could, by his own efforts, return to zero or reach much higher than
38 in extended reading style.
Many similar tests could be cited. The most consistent and highly
controlled speaker in the survey was Warren M., 27, a social worker
and graduate student. At college he had been intensively trained in
speaking technique, had done a great deal of acting, and was justly
proud of the control he could exert over his voice. His original
reading of the r paragraph was at an index of 68. After a thorough
discussion of (r), he read again to produce a perfectly consistent
version. A very slow reading gave 90; fast, 56; more careful, 80; a
repeat, 80; again, concentrating on voice quality 63; he then recited
Jabberwocky at 88. 13
Merwin same age, was able
M., a less sophisticated speaker of the
to improve his performance from (r)-28 to (r)-50. There is reason to
think that older speakers would have less ability to shift, and that
only very young ones, just emerging from their preadolescent years,
would be able to make radical changes in their pattern by conscious
attention.
Martha S., a very careful, Jewish middle-class speaker of 45, was

asked to read several paragraphs after discussion.

Variable Original reading Conscious effort

(r) 45 47
(eh) 40 40
(oh) 28 29

13. appears here, as indicated in fn. 5, that a high concentration of (r) words
It

makes more difficulties than a long text with the (r)'s dispersed. A similar effect was
noted in the (th) paragraph; some speakers saw the phrase this thing, that thing, and
the other thing, some even took a breath before attempting it, but by the time they
reached the fifth or sixth item, fatigue set in, and with it, (dh-3).
106 SOQOLINGUISTIC PATTERNS

The (eh) index was already at the point preferred by the speaker,
but the (oh) items still fluctuated considerably, and the small in-
creases in both (r) and (oh) show her inability to attain the desired
result. On the other hand, her daughter, Susan S., 13, was able to
read with an index of 50, and after discussion, reach as high as
(r)

75. Her normal (oh) index of 15 was shifted to 28 as she imitated

her mother.
An even more dramatic case was that of Bonnie R., 10 years old.
Whereas her parents used no more than 5 or 10 percent (r-1) in
reading, she was able to go from an (r) index of 14 to (r)-64 after
this variable was discussed in the family interview.
The compelling nature of the pattern of stylistic alternation ap-
pears to operate at the extremes of the social scale, as well as in
the center. Below, we may compare the record of two New Yorkers
of radically different education and social status. On the left is the
performance of Bennie N., 40, a truck driver who finished only the
first term of high school. On the right is the record of Miriam L.,

35, who graduated from Hunter College and St. John's Law School,
and is now practicing law on the Lower East Side (heading for styles
and variables as before).

STYLISTICARRAY STYLISTIC ARRAY

FOR BENNIE N. FOR MIRIAM L.

00 00 13 33 33 32 47 39 56 100
19 21 26 22 28 38 40 39
15 20 24 20 20 26 30 30
168 81 58 00 00 00
153 96 38 25 04 02

The absolute values of these variables are as totally opposed as any

pair of speakers we might choose. But the structure of stylistic
variation is essentially the same. In this comparison, one can find
a statement of the theme which will dominate this study of social
stratification of language: that New York City is a speech community,
united by a common evaluation of the same variables which serve
to differentiate the speakers. The structures seen above are concrete
manifestation of that evaluation.
The differences between the speakers are, of course, very real.

Bennie N. uses no (r-1) in conversation; at her most casual, Miriam L.

uses large numbers of (r-1) variants. The (eh) sound for Bennie N.
is normally that of where; Miriam L. aims for the sound of that
The Isolation of Contextual Styles 107

and bat and usually reaches it. For Bennie N., stops are practically
normal forms of (th) and (dh); Miriam L. never uses anything but
the prestige form for (th), and only a few affricates for (dh) except
in the most casual style. At this point, one might ask whether the
difference may be in large part that Miriam L. recognizes the formal
situation of the interview, and never uses her casual style in this
interview, while Bennie N. doesn't care that much about making a
good impression. Perhaps Miriam L.'s true casual style, outside of
the interview, is not so different, after all.

The record of the survey in general shows that this is not the case.
In this particular case, can resolve a part of the doubt since I spent
I

fifteen minutes waiting in Miss L.'s office while she discussed busi-
ness affairs with a client. The client seemed to be an old friend, and
in any case, Miss L. did not know who I was, and language had not
entered the picture. We may compare the record of this conversation
with the Style A
and Style B of the interview: the former appears
to lie somewhere in between Style A and Style B, perhaps closer
to B. In any case, the casual style elicited by the interview is consid-
erably less formal than that which Miss L. uses in the daily execution
of her business affairs.

With Client Style A Style

(r) 40 32 47
(eh) 30 28 38
(oh) 27 20 26
(th) 00 00 00
(dh) 00 25 04

The Structure of Stylistic Variation

In the study of the Lower East Side, we proposed

to reduce the
irregularity in the linguistic behavior of New York speakers by going
—
beyond the idiolect the speech of one person in a single context.
We first isolated the most important variables which interfered with
the establishment of a coherent structure for these idiolects. After
defining and isolating a wide range of styles in highly comparable
interview situations, we were able to discover a regular pattern of
behavior governing the occurrence of these variables in the speech
of many individuals.
The term structure has been used so often in linguistic discussion
that it sometimes slipsaway from us, or becomes fixed in denoting
a particular kind of unit which was originally analyzed by structural
108 SOCIOUNGUISTIC PATTERNS

considerations. Thus a list of phonemes may be taken as a structural

statement, though no structure uniting the list is given, other than
the fact that each unit is different. The excellent definition of Web-
ster's New International Dictionary (2nd Edition): "structure, the
interrelationship of parts as dominated by the general character of
the whole" describes the pattern of stylistic variation which has been
shown in the foregoing pages. But in addition to this description,
20th-century linguistics has added the requirement that linguistic
structures be composed of discrete units, which alternate in an
all-or-none relationship. 14
The dimensions of stylistic variation that have been illustrated
—
cannot satisfy this requirement at least, not by the evidence that
has been presented. The sharp contrasts among Styles A through D'
are in part artifacts of the procedure. If this dimension is thought
of as a continuum, then the method of dividing that continuum used
here is perfectly adequate; if one suspects that natural breaks in the
continuum exist so that in natural situations one does not pass evenly
and continuously from careful to increasingly casual speech, this
must be demonstrated by other methods.
If contrast exists between casual and careful styles, and the varia-
bles which we are using play a significant role in that contrast, they
do not seem to operate as all-or-none signals. The use of a single
—
variant even a highly stigmatized one such as a centralized diph-
—
thong in "boid" for bird does not usually produce a strong social
reaction; it may only set up an expectation that such forms might
recur, so that the listener does begin to perceive a socially significant
pattern. Every speaker occasionally begins a (dh) word with a sharp
onset, which can be interpreted as an affricate, [dS]. However, in
the prestige form of speech, these forms recur so seldom that they
are negligible. Any pattern of expectation set up by them dies out
before the next is heard. It is the frequency with which Bennie N.
uses such forms that has social significance, and it is essentially one
level of frequency which contrasts with another level in the struc-
tures outlined above.
Are there breaks in the continuum of possible frequencies? This
varies from one variable to another, as the overall study of strati-

14. Thus the phonological structure is built with discrete units, phonemes that are
themselves the products of the natural economy of the language. The structural units
of the vowel systems are not artifacts of the analytical procedure; the categorizing
procedure which breaks the continuum into highly discrete units can be tested and
observed.
The Isolation of Contextual Styles 109

fication in New York City showed. However, the very clearcut type
of all-or-none reaction which is characteristic of phonemic units will
be found not in performance so much as in evaluation (see Ch. 6).
But whether or not we consider stylistic variation to be a continuum
of expressive behavior, or a subtle type of discrete alternation, it
is clear that it must be approached through quantitative methods.

We are in no position to predict exactly when a given speaker will

produce a fricative, or when he will produce a stop. A complex of
many factors operates to obscure stylistic regularities at the level
of the individual instance. The remarkable fact is that the basic unit
of stylistic contrast is a frequency set up by as few as ten occurrences
of a particular variable.
The methods developed here for the isolation of contextual styles
were preliminary to the general analysis of social and stylistic strati-
fication in New York City. But they are quite general in their scope
and have since been successfully used in many other contexts. The
techniques for extending the formal end of the stylistic range have
been used more widely than the techniques for isolating casual
speech but both directions have been followed (see Shuy, Wolfram
and Riley 1967; Wolfram 1969; Cook 1969; Sankoff and Cedergren
1971; Trudgill 1971). The methods for bypassing the constraints of
the interview situation are of course only one way of obtaining a
view of casual speech, and not necessarily the most definitive. In
more recent work we have relied more upon group sessions, in which
the interaction of members overrides the effect of observation, and
gives us a more direct view of the vernacular with less influence
of the observer (Cf. Gumperz 1964; Labov et al. 1968; Legum et al.
1971). However, individual face-to-face interviews will always be
needed for the large body of accurately recorded speech that we need
for a detailed study of the speech of a given individual. Individual
interviews were used for a random sample of 100 adult speakers in
south-central Harlem, and the techniques developed here were used
to isolate casual speech (Labov et al. 1968). In recent instrumental
studies of sound change in progress, individual interviews were
required to produce the large body of continuous speech needed to
chart the vowel system of each individual in full. In a series of
exploratory interviews in various regions of the United States and
England, we have been developing further the techniques for elicit-
ing the vernacular in face-to-face situations. It therefore seems likely
that the principles behind the methods outlined here will provide
a foundation for future sociolinguistic studies.
 The Reflection of
4 Social Processes in
Linguistic Structures

IN the last chapter, we presented five phonological variables

of the New York City system and showed how a wide range of styles
could be isolated within the individual interview. When styles are
organized along a single dimension according to the amount of
attention paid to speech, it can be seen that most speakers follow
a regular pattern of style shifting in the same direction. We will now
examine a second dimension of linguistic variation: the differentia-
tion of speakers by their social status. 1 Our debt to sociology and
survey methodology is very great: the main findings are based on
a secondary survey of the Lower East Side, utilizing the careful and
exhaustive enumeration of the population by the Mobilization for
Youth research staff and their construction of a random sample; the
clarity of the results depends upon a sophisticated index of socio-
economic status, itself the product of a long line of sociological
development. This chapter is therefore addressed directly to sociol-
ogists and those who are interested in the more systematic, quanti-
tative aspects of class stratification. But we will also consider other
aspects of social differentiation of language which emerged from the
Lower East Side survey. Our focus will be frankly and directly
sociolinguistic, and the conclusion will consider more generally the
ways in which the study of language and the study of society may
interact.
The procedures of descriptive linguistics are based upon the
conception of language as a structured set of social norms. It has

1.This chapter first appeared in Joshua Fishman, ed., Readings in the Sociology
of Language (The Hague: Mouton, 1968), pp. 240-51. It has been revised for publication
here.

110
Reflection of Social Processes in Linguistic Structures 111

been useful in the past to consider these norms as invariants, shared

by all members of the speech community. However, closer studies
of the social context in which language is used show that many
elements of linguistic structure are involved in systematic variation
which reflects both temporal change and extralinguistic social proc-
esses.
As a form of social behavior, language is naturally of interest to
the sociologist. But language may have a special utility for the
sociologist as a sensitive index of many other social processes.
Variation in linguistic behavior does not in itself exert a powerful
influence on social development, nor does it affect drastically the
lifechances of the individual; on the contrary, the shape of linguistic
behavior changes rapidly as the speaker's social position changes.
This malleability of language underlies its great utility as an indicator
of social change.
—
Phonological indexes based upon the elements of the sound
—
system of a language are particularly useful in this respect. They
give us a large body of quantitative data from relatively small sam-
ples of speech: 50-200 occurrences of a single item in a half-hour
conversation. To a large extent, the variation on which these indexes
are based is independent of conscious control of the subject. Finally,
phonological systems show the highest degree of internal structure
of all linguistic systems, and thus provide the investigator with an
extensive series of parallel and convergent results.
The five phonological variables described in Ch. 3 were studied
in a population drawn from a linguistic survey of the Lower East
Side. This survey was based upon a primary survey of social atti-
tudes of Lower East Side residents, carried out by Mobilization for
Youth in 1961. The original sample of the population of 100,000
consisted of 988 adult subjects. Our target sample was 195 of these
respondents, representing about 33,000 native English speakers who
had not moved within the previous two years. Through the assistance
of Mobilization for Youth, and the Columbia School of Social Work,
we had available a large body of information on the social charac-
teristics of the informants, and we were able to concentrate entirely
on their linguistic behavior in this secondary survey. Eighty-one
percent of the target sample was reached in the investigation of
language on the Lower East Side.
New York City presents some exceptionally difficult problems for
the study of linguistic systems. As we have seen in Ch. 3, New
112 SOCIOLINGUISTIC PATTERNS

Yorkers show a remarkable range of stylistic variation, as well as

such an extent that earlier investigators failed
social variation, to
to findany pattern, and attributed many variables to pure chance.
To study social variation, it was first necessary to define and isolate a
range of contextual styles within the linguistic interview (see Ch. 3).
Since the context of the formal interview does not ordinarily elicit
casual or spontaneous speech, the methods which were developed
to overcome this limitation were crucial to the success of the inves-
tigation. Our success in defining and eliciting casual conversation
is evident in the convergence of these results with other studies
which utilized anonymous observations (Ch. 2), and also in the
consistency of the patterns of stylistic variation which were found.
Although there is a great range in the absolute values of these
variables as used by New Yorkers, there is great agreement in the
pattern of stylistic variation. Almost 80 percent of the respondents
showed patterns of stylistic variation consistent with the status of
(r-1) as a prestige marker, and stops and affricates for (th) as stigma-
tized forms.
This pattern of stylistic variation isprimarily of concern to lin-
guists and to students of the ethnography of speaking. However, it
is closely associated with the pattern of social stratification which
pervades many aspects of urban society. The pattern of stylistic
variation and the pattern of social variation enter into the complex
and regular structure which we see in Fig. 4.1.
Fig. 4.1 is a class stratification diagram for (th), derived from the
behavior of 81 adult respondents, raised in New York City (Labov
1966a, Ch. 7). This is the first of the sociolinguistic patterns which
are the main focus of this and the following two chapters. The
vertical axis is the scale of average (th) index scores. The horizontal
axis represents the four contextual styles, A-D, established in Ch.
3. The horizontal lines connecting the values show the progression
of average index scores for socioeconomic class groups. These groups
are defined as divisions of the ten-point socioeconomic scale con-
structed by Mobilization for Youth on the basis of their data in the
original survey. The socioeconomic index is based on three equally
weighted indicators of productive status: occupation (of the bread-
winner), education (of the respondent), and income (of the family). 2

2. The socioeconomic index, as developed by Mobilization for Youth,

original
breadwinner rather than of the respondent. The net result
utilized the education of the
of both approaches in the study of social stratification is the same.
Reflection of Social Processes in Linguistic Structure 113

Style

Fig. 4.1. Class stratification of a linguistic variable with stable

social significance: (th) in thing, through, etc. Socioeconomic
class scale: 0-1, lower class; 2-4, working class; 5-6, 7-8, lower
middle class; 9, upper middle class. A, casual speech; B, careful
speech; C, reading style; D, word lists.

Informally, we may describe these class groups as follows: group

0-1, lower class; 2-4, working class; 5-8, lower middle class; 9, upper
middle class. Classes 2 and 5 are marginal groups, which sometimes
show the linguistic behavior of the next lower group, and sometimes
that of the next higher group.
Fig. 4.1 is an example of what we may call sharp stratification.
The five strata of the population are grouped into two larger strata
with widely different use of the variable. The parallel variable (dh)
shows the same kind of sharp stratification (Labov 1966a:253). Fig.
4.2 is a class stratification diagram which shows a somewhat different
114 SOCIOLINGUISTIC PATTERNS

Style

Fig. 4.2.Class stratification of a linguistic variable in process of

change: (r) in guard, car, beer, beard, board, etc. SEC (Socio-
economic class) scale: 0-1, lower class; 2-4, working class; 5-6,
7-8, lower middle class; 9, upper middle class. A, casual speech;
B, careful speech; C, reading style; D, word lists; D', minimal
pairs.

type of stratification. The vertical axis is the phonological index for

(r), in which 100 represents a consistent r-pronouncing dialect, and

00 a consistent r-less dialect. The horizontal axis shows five, not four,
stylistic contexts, including, at D', the reading of word pairs in which
(r) is the sole focus of attention: guard vs. god, dock vs. dark. This
structure an example of what we may call fine stratification: a
is

great many divisions of the socioeconomic continuum in which

stratification is preserved at each stylistic level. Other investigations
carried out in New York City, including the department-store survey,
support the general hypothesis on the fine stratification of (r) ad-
vanced in Ch. 2: any groups of New Yorkers that are ranked in a
Reflection of Social Processes in Linguistic Structures 115

hierarchical scale by nonlinguistic criteria will be ranked in the same

order by their differential use of (r).

The status of (r-1) as a prestige marker is indicated by the general

upward direction of all horizontal lines we go from informal to
as
formal contexts. At the level of casual, everyday speech, only the
upper middle class (9) shows a significant degree of r-pronunciation.
But in more formal styles, the amount of r-pronunciation for other
groups rises rapidly. The lower middle class, in particular, shows
an extremely rapid increase, surpassing the upper-middle-class level
in the two most formal styles. This crossover pattern appears at first
sight to be a deviation from the regular structure shown in Fig. 4.1.
It is a pattern which appears in other diagrams: a similar crossover

of the lower middle class appears for two other phonological in-
—
dexes in fact, for all those linguistic variables which are involved
in a process of linguistic change under social pressure. On the other
hand, the social and stylistic patterns for (th) have remained stable
for at least 75 years, and show no sign of a crossover pattern. Thus
the hypercorrect behavior of the lower middle class is seen as a
synchronic indicator of linguistic change in progress. We will con-
sider this hypercorrect pattern in much greater detail in the next
chapter, and its significance for the study of change in progress will
be examined in Ch. 9.
The linear nature of the ten-point scale of socioeconomic status
is confirmed by the fact that it yields regular stratification for many

linguistic variables, grammatical as well as phonological. The lin-

guistic variables have been correlated with the individual social
—
indicators of productive status occupation, education and in-
come — and appears that no single indicator is as closely correlated
it

with linguistic behavior as the combined index. However, an index

which combines occupation and education neglecting income—
gives more regular stratification for the (th) variable. For education,
there is one sharp break in linguistic behavior for this variable: the
completion of the first year of high school. For occupation, there are
sharp differences between blue-collar workers, white-collar workers,
and professionals. If we combine these two indicators, we obtain
four classes which divide the population almost equally, and stratify
(th) usage regularly. This classification seems to be superior to the
socioeconomic scale for analysis of variables such as (th) which
reflect linguistic habits formed relatively early in life (Labov 1966a).
However, the combined socioeconomic index, utilizing income, does
 116 SOCIOLINGUISTIC PATTERNS

show more regular stratification for a variable such as (r). Since (r-1)
is a recently introduced prestige marker in New York City speech,
it seems consistent — almost predictable — that it should be closely
correlated with a socioeconomic scale which includes current in-
come, and thus represents most closely the current social status of
the subject.
Fig. 4.3 shows the distribution of (r) by age levels, a distribution

100 r
Style A

SO

(r) 60 -

20-29 yrs
40 30-39 yrs
JT — Class 9

20 40-49 yrs
50-75 yrs ^_^

=-b^—crEr_ -
I

0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8

SEC
Fig. 4.3. Development of class stratification of (r) for casual
speech (Style A) in apparent time. SEC = socioeconomic class
scale.

in apparent time which indicates a sudden increase in real time of

the social stratification of (r) in everyday, casual speech. For the two
oldest age levels, there is little indication of social significance of
(r-1).But beginning with those under 40 years old, there is a radically
different situation, with (r-1) acting as a prestige marker of upper-
middle-class usage only. This sudden change in the status of (r) seems
to have coincided with the events of World War II.
So far, we have been considering only one aspect of social strati-
fication: the differentiation of objective behavior. In our studies of
New York City, the complementary aspect of social stratification was
alsoexamined: social evaluation. A subjective reaction test was
developed to isolate unconscious social responses to the values of
individual phonological variables. In these tests, the subject rated
a number from the speech of other New Yorkers
of short excerpts
on a scale of occupational suitability (see Fig. 6.1), and cross com-
Reflection of Social Processes in Linguistic Structures 117

parison of these ratings enabled us to isolate the unconscious sub-

jective reactions of respondents to single phonological variables. This
subjective dimension will be examined
in detail in Ch. 6: most
striking is the uniformity of New
Yorkers under 40 years old. All
subjects between 18 and 39 agreed in their positive evaluation of
(r-1) (Fig. 6.2) despite the fact (as shown in Fig. 4.3) that the great
majority of these subjects do not use any (r-1) in their everyday
speech. Thus sharp diversification of (r) in objective performance
is accompanied by uniform subjective evaluation of the social sig-

nificance of this feature. On the other hand, the subjects over 40 years
old, who show no differential pattern in their use of (r), show a very
mixed pattern in their social evaluation of this variable.
This result is typical of many other empirical findings which
confirm the view of New York City as a single speech community,
united by a uniform evaluation of linguistic features, yet diversified
by increasing stratification in objective performance.
The special role of the lower middle class in linguistic change has
been illustrated here in only one example, the crossover pattern of
Fig. 4.2. We more detailed discussion of the
can look ahead to the
next chapter and consider other evidence for the special behavior
of this group. When Fig. 4.3 is replicated for increasingly formal
styles, we see that in each age level, the lower middle class shows
the greatest tendency towards the introduction of r-pronunciation,
and in the most formal styles, goes far beyond the upper-middle-class
level in this respect. A great deal of evidence shows that lower-
middle-class speakers have the greatest tendency towards linguistic
insecurity, and therefore tend to adopt, even in middle age, the
prestige forms used by the youngest members of the highest-ranking
class. This linguistic insecurity is shown by the very wide range of
stylistic variation used by lower-middle-class speakers; by their great
fluctuation within a given stylistic context; by their conscious striv-
ing for correctness; and by their strongly negative attitudes towards
their native speech pattern.
A simple yet accurate measure of linguistic insecurity was ob-
tained by an independent approach, based on lexical behavior. The
subjects were presented with 18 words which have socially signifi-
cant variants in pronunciation: vase, aunt, escalator, etc., and were
asked to select the form they thought was correct: [veiz - va:z], [ae nt
- a • nt], [eskaleit^ - They were then asked to indicate
eskjuleitar-], etc.

which form they usually used themselves. The total number of cases
118 SOCIOLINGUISTIC PATTERNS

inwhich these two choices differed was taken as the Index of

(ILI). By this measure, the lower middle class
Linguistic Insecurity
showed much the greatest degree of linguistic insecurity (see Table
5.1).
Social stratification and its consequences are only one type of
social process which is reflected in linguistic structures. The inter-
action of ethnic groups in New York City — Jews, Italians, blacks, and
—
Puerto Ricans is also reflected in these and other linguistic varia-
bles. For some variables, New York City blacks participate in the
same structure of social and stylistic variation as white New Yorkers.
For other variables, there is an absolute differentiation of white and
black which reflects the process of social segregation characteristic
of the city. For example, there is a southern phonological charac-
teristic which merges the vowels /i/ and /e/ before nasals: pin and
pen, since and sense, are homonyms: "I asked for a straight [pin]
and he gave me a writing [pin]." In New York City, this phonological
traithas been generalized throughout the black community, so that
the younger speakers, whether or not they show other southern
characteristics in their speech, regularly show this merger. Thus this
linguistic characteristic acts as an absolute differentiator of the black
group, reflecting the social processes which identify the racial group
as a whole. Similar phonological characteristics can be found mark-
ing the Puerto Rican group. 3
Segregation of black and white may be seen in aspects of linguistic
behavior quite distinct from the phonological system. Our investi-
gation of New York City speech includes a number of semantic
studies: one of the most fruitful of these concerns the semantic
structures which revolve about the term common sense. This term
lies at the center of one of the most important areas of intellectual
activity for most Americans. It is a term frequently used, with con-
siderable effect; its meaning is often debated, and questions about
common sense evoke substantial intellectual effort from most of our
subjects. Blacks use the term common sense, but also an equivalent

3. Most New Yorkers differentiate the vowels of can as in "tin can" from those
of can in "I can." None of the Puerto Rican subjects interviewed showed a consistent
use of this phonemic distinction. Puerto Rican speakers also show patterns of conso-
nant cluster simplification which are different from those of both black and white
New Yorkers. Clusters ending in -rd are simplified, and preconsonantal r is treated
as a consonant: a good car' game. This does not fall within the range of variations
open to other New Yorkers, who treat -r- as a vowel tactically, quite different from
-I- in good oV Mike.
Reflection of Social Processes in Linguistic Structure 119

term which is not a part of the native vocabulary of any white

speakers. This term is mother-wit, or mother-with [itiaoVwiG]. For
a few white speakers, mother-wit is identified as an archaic, learned
term: but for blacks, it is a native term used frequently by older
members of the household, referring to a complex of emotions and
concepts that is quite important to them. Yet blacks have no idea
that white people do not use mother-wit, and whites have no inkling
of the black use of this term. Contrast this complete lack of commu-
nication in an important area of intellectual activity, with the smooth
and regular transmission of slang terms from black musicians to the
white population as a whole.
The process of social segregation springs from causes and mecha-
nisms which have been studied in detail. However, the opposing
process of social integration is less obvious, and on the plane of
linguistic structure, it is not at all clear how it takes place. Consider
the semantic structure of common sense. When we analyze the
semantic components of this term, its position in a hierarchical
taxonomy, and its relation to coordinate terms in a semantic para-
digm, we see great differences in the semantic structures used by
various speakers.
I can best illustrate this diversity by contrasting two types of
responses to our questions on common sense, responses which
usually fall into two consistent sets. Respondent A may think of
common sense as just 'sensible talk.' If he understands the cognitive
content of an utterance, that to him is common sense. Respondent
B considers common sense to be the highest form of rational activity,
the application of knowledge to solve the most difficult problems.
Do most people have common sense? A says yes, B says no. Who
has a great deal of common sense? A thinks that doctors, lawyers,
professors have the most. B thinks that uneducated people are more
apt to have common sense, and immediately calls to mind some
highly educated people with no common sense at all. If we say "two
and two make four," is that an example of common sense? A says
yes, B says no. Can we say that a person is intelligent, yet has no
common sense? A says no, because intelligence is the same as com-
mon sense. B says yes, common sense and intelligence are quite
different. A believes that if someone can be called smart, he would
also have common sense; B sees no connection between smartness
and common sense. Can one have wisdom, and yet no common
sense? A says yes, B says no.
The extreme differences between types A and B, which are not
120 SOCIOLINGUISTIC PATTERNS

independent of social stratification, lead us to question the possibility

of semantic integration. Can such individuals, who have radically
opposed semantic structures for common sense, be said to under-
stand one another? Can the term common sense be used to commu-
nicate meaning between these speakers? Some writers (particularly
the followers of General Semantics) feel that native speakers of
English usually do not understand one another, that such opposing
structures inevitably lead to misunderstanding. My own results lead
me to infer the opposite. People do understand one another: semantic
integration seems to take place through a central set of relations of
equivalence and attribution upon which all English speakers agree.
With only a few exceptions, all subjects agree that common sense
falls under the superordinate judgment: it is 'good judgment'. Equally
high agreement is found in the collocation of practical, or everyday,
with common sense. We have no simple term to describe the quality
of 'not being learned from books', yet there is also a very high degree
of agreement in this attribute of common sense.
semantic integration takes place, it must be by a social process
If

in which extreme variants are suppressed in group interaction at the

expense of central, or core values. Further studies are required to
determine if such a mechanism does in fact operate.
This discussion has presented a number of aspects of language
behavior in which linguistic structures are seen to reflect social
is a wide range of benefits which
processes. In the overall view, there
may be drawn from the interaction of sociological and linguistic
investigations. These may be considered under three headings, in
order of increasing generality:

1. Linguistic indexes provide a large body of quantitative data which

reflect the influence of many independent variables. does not
It

seem impractical for tape-recorded data of this type to be col-

lected and analyzed by social scientists who are not primarily
linguists. Once the social significance of a given linguistic variant
has been determined, by methods such as those outlined above,
this variable may then serve as an index to measure other forms
of social behavior: upward social aspirations, social mobility and
insecurity, changes in social stratification and segregation.
2. Many of the fundamental concepts of sociology are exemplified
in the results of these studies of linguistic variation. The speech
community is not defined by any marked agreement in the use
Reflection of Social Processes in Linguistic Structures 121

of language elements, so much as by participation in a set of

shared norms; these norms may be observed in overt types of
evaluative behavior, and by the uniformity of abstract patterns
of variation which are invariant in respect to particular levels of
usage. Similarly, through observations of linguistic behavior it is

possible to make detailed studies of the structure of class strati-

fication in a given community. We find that there are some lin-
guistic variables which are correlated with an abstract measure
of class position, derived from a combination of several non-
isomorphic indicators, where no single, less abstract measure will
yield equally good correlations.
3. If we consider seriously the concept of language as a form of social

behavior, evident that any theoretical advance in the analysis

it is

of the mechanismof linguistic evolution will contribute directly

to the general theory of social evolution. In this respect, it is
necessary for linguists to refine and extend their methods of
structural analysis to the use of language in complex urban socie-
ties. For this purpose, linguistics may now draw upon the tech-

niques of survey methodology; more importantly, many of the

theoretical approaches of linguistics may be reinterpreted in the
light of more general concepts of social behavior developed by
other social sciences. Thus the main achievements of linguistic
science, which may formerly have appeared remote and irrelevant
to many sociologists, may eventually be seen as consistent with
the present direction of sociology, and valuable for the under-
standing of social function and social change.
 Hypercorrection by the
5 Lower Middle Class as a Factor
in Linguistic Change

A LARGE part of our approach to language concerns the

isolation of invariant functional units, and the invariant structures
which each other. 1 Considerable progress has
relate these units to
been made through this. approach to the analysis of language, but
in many areas we have reached a point where a different approach
is required, in which the variable features of language become the

primary focus of attention, rather than the constant features. The

empirical study of linguistic variants shows us that linguistic struc-
ture is not confined to the invariant, functional units such as pho-
nemes, morphemes, or tagmemes; rather, there is a level of variant
structure which relates entire systems of functional units, and which
governs the distribution of subfunctional variants within each func-
tional unit. This type of variant structure thus becomes a new type
of invariant at a more refined level of observation.
The study of social variation in language simply one of many
is

aspects of the study of variant linguistic structures. One motivation

for the linguist to study such structures is that they provide empirical
evidence to resolve alternate structural analyses at the functional
level,providing empirical solutions to problems which are otherwise
meaningless. Secondly, variant structures are defined by quantitative
methods which allow the detailed studies of linguistic changes in
progress. The central theoretical problem with which the present
report is concerned is the mechanism of linguistic change, in which
the dynamics of social interaction appear to play an important part.

1. This chapter is based on a paper first given at a conference on Sociolinguistics

held at the University of California, Los Angeles, in 1964, and first published in William
Bright, ed., Sociolinguistics (The Hague: Mouton, 1966).

122
Hypercorrection as a Factor in Linguistic Change 123

The process of linguistic change may be considered as having three

stages (cf. Sturtevant 1947:Ch. 8). In the origin of a change, it is one
of innumerable variations confined to the use of a few people. In
the propagation of the change, it is adopted by such large numbers
of speakers that it stands in contrast to the older form along a broad
front of social interaction. In the completion of the change, it attains
regularity by the elimination of competing variants. In this discus-
sion, the second stage will be of primary concern: at this stage, it
appears that social significance is inevitably associated with the
variant and with its opposition to the older form.

Social forces exerted upon linguistic forms are of two distinct

types, which we may call pressures from above, and pressures from
below. By below is meant "below the level of conscious awareness."
Pressures from below operate upon entire linguistic systems, in
response to social motivations which are relatively obscure and yet
have the greatest significance for the general evolution of language.
In this presentation, we will be concerned primarily with social
pressures from above, which represent the overt process of social
correction applied to individual linguistic forms. In this process the
special role of the lower middle class, or more generally, the role
of the second-ranking status group, will be the principal focus of
attention.
The role of hypercorrection in the propagation of linguistic change
was indicated in the Martha's Vineyard study (Ch. 1). Here we will
examine the hypercorrect behavior of a single class group in the
speech community of New York City, and the consequences of this
behavior for the process of linguistic change.
Most of the evidence to be presented is based upon the quantitative
measurement of phonological indexes, although lexical and gram-
matical behavior is also considered. The methods of selecting, de-
fining, and measuring phonological variables, and of defining and
isolating contextual styles, have been presented in Ch. 3. These
methods were applied in a survey of the Lower East Side of New
York City, an area with a population of 100,000, using informants
from the Mobilization for Youth survey discussed in Ch. 4. A total
of 207 adults and children were interviewed; most of the data to be
presented here is based on the speech of 81 adults raised in New
York City, whose speech was studied in the greatest detail, but
information from several other subgroups will be utilized as well.
The methods of quantitative analysis were applied to the problem
124 SOCIOLINGUISTIC PATTERNS

of describing the phonological structure of the community as a

whole, as opposed to the speech of individuals. Indeed, it was found
that the speech of most individuals did not form a coherent and
rational system, but was marked by numerous oscillations, contra-
dictions and alternations which were inexplicable in terms of a
single idiolect. For this reason, previous investigators had described
large parts of the linguistic behavior of New Yorkers as being a
product of pure chance, "thoroughly haphazard" (Hubbell 1950:48;
cf. Bronstein 1962:24). But when the speech of any one person in

any given context was charted against the overall pattern of social
and stylistic variation of the community, his linguistic behavior was
seen to be highly determined and highly structured.
Such a structure of social and stylistic variation was seen in Fig.
4.1, (p. 113)showing the stratification of (th). Let us return to the
extraordinarily regular properties of this pattern (for convenience,
Figs. 4.1and 4.2 are repeated here).
Figure 4.1 shows a complex pattern of regular relations. Each
value for a given class, in a given style, is lower than the value for
the next most informal style, and higher than the next most formal
style; it is also lower than the value for the next lower status group,
and higher than the next higher status group (with one exception).
From this figure, it is plain that the fricative form of (th) is the
prestige form in New York City (as it is throughout the United States),
and the stops and affricates are stigmatized forms. All class groups
agree in their gradual reduction of the use of stops and affricates
in more formal styles, and in each style there is a clearcut stratifica-
tion of the variable. In other words, we have a structure consisting
of two invariant sets of relations.

Hypercorrect Behavior of the Lower Middle Class

By contrast, Fig. 4.2 of Ch. 4 shows a different type of structure.

At one extreme we see one class group shows any de-
that only
gree of r-pronunciation in casual speech; that is, in everyday life,
(r-1) functions as a prestige marker of the highest-ranking status

group. The lower middle class, 6-8, shows only the same negligible
amount of r-pronunciation as the working class and lower class. But
as one follows the progression toward more formal styles, the lower
middle class shows a rapid increase in the values of (r), until at Styles
D and D', it surpasses the usage of the upper middle class. (See Ch.
 Style

Fig. 4.1. Class stratification of (th) (see p. 113).

SEC.
80

60

01
"? 40

20

— i

C
Style

Fig. 4.2. Class stratification of (r) (see p. 114).

126 SOCIOLINGUISTIC PATTERNS

3 for definitions of the contextual styles of A-D'). The crossover

pattern appears to be a deviation from the regularity of the behavior
shown by the other classesand Fig. 4.2 thus departs from the regu-
larity of Fig. 4.1. To describe this phenomenon, the term hyper-
correction will be used, since the lower-middle-class speakers go
beyond the highest-status group in their tendency to use the forms
considered correct and appropriate for formal styles. (This is of
course an extension of the usual use of the term to indicate an
irregular misapplication of an imperfectly learned rule, as in the
hypercorrect case marking of whom did you say is calling?) If the
deviation shown in Fig. 4.2 were found only in one case, it would
—
be difficult to interpret an irregularity in the method or in linguistic
behavior. However, we find a similar crossover pattern in several
other structures. Fig. 5.1, for example, is the class stratification
diagram for (eh), the height of the vowel in bad, ask, dance, etc.
as defined in Ch. 3.
The class stratification shows four class groups, with the lower
middle class crossing over the value for the upper middle class in
the most formal style. That is, in reading a word list such as bat,
bad, back, bag, batch, badge, the lower-middle-class speakers show
the greatest tendency to say [baet, bae:d, baek, bae:g ], although
. . .

in everyday speech, they are more likely to say [be: a d], homonymous
with bared: "I had a [be: a d] cut; I [be: a d] my arm."
We find a similar hypercorrect pattern for the socially significant
variable (oh). Fig. 5.2 is a style stratification diagram for this variable.
The vertical axis in Fig. 5.2 represents the phonological index,
formed from ratings which are parallel to those for (eh) (see page
76). For each socioeconomic class, the value of the index in each
contextual style is plotted on Fig. 5.2, and the values for the same
style are connected along horizontal lines. This diagram preserves
all of the information present in the original data, and allows us to

follow the differentiation of stylistic behavior in great detail. For the

lower class, 0-2, the variable (oh) has no social or stylistic signifi-
cance, as shown by the chaotic distribution of values in the four
contextual styles. There is an intimate connection between stylistic

and and as we see here, deviations from one axis

social variation,
of variation are usually accompanied by deviations from the other.
The lower class shows no trace of the stratified pattern of the three
higher-ranking classes.
Hypercorrection as a Factor in Linguistic Change 127

15—,

SEC

Style

Fig. 5.1 Class stratification for (eh). SEC = socioeconomic class

scale. A, casual speech; B, careful speech; C, reading style; D,
word lists.

The upper section of the working class begins to show the high
values of (oh-1) — a
[u: f] — in casual speech, values which are charac-
teristic of the lower middle class as well. However, the social signifi-
cance for the working class is not exactly the same as for the lower
middle class, since the working-class speakers show only a slight
tendency towards correction of this vowel to more open values in
formal style. The lower middle class, on the other hand, shows an
extreme tendency towards correction, reading a list of words such
as Paul, all, ball, awful, office, with a determined, though inconsist-
ent effort towards (oh-S^po^l, d1, ofal, od:^1, ufis, du: a g ]. The . . .

highest-ranking group, the upper middle class, shows a more moder-

ate tendency in both casual and formal styles; as a result, we see
the familiar crossover pattern in Fig. 5.3, which is the class strati-
fication diagram for the variable without the lower class.
(oh),
If we now compare Figs. 4.2, 5.1, and 5.3, it is evident that the same

hypercorrect behavior of the lower middle class is displayed. This

deviation from regular structure is not, in fact, a deviation, but rather
a recurrent aspect of a regular structure quite distinct from that of
Fig. 4.1. Fig. 4.1 represents the structure typical of variables not
involved in the process of linguistic change. A similar pattern may
 128 SOCIOLINGUISTIC PATTERNS

15 -7

20 Style

25-

30-

35

SEC
Fig. 5.2. Style stratification of (oh). SEC = socioeconomic class
scale. A, casual speech; B, careful speech; C, reading style; D,
word lists.

be found in the parallel case of (dh), the initial consonant of this,

that, etc., or the morphological variable of the suffix -ing. On the
other hand, the pattern of (r), (eh), (oh) is that of phonological varia-
bles involved in the process of linguistic change, as we shall see.
The special role of the lower middle class in this pattern is apparent
in the sensitivity of this group to social pressures from above. We
may gain further understanding of the behavior of this class by
studying other types of evidence which go beyond the tabulation
of objective performance.

Hypersensitivity of the Lower Middle Class

in Subjective Reactions

In the study of the social stratification of language, we need not be

confined to the evidence of objective differentiation of behavior.
Hypercorrection as a Factor in Linguistic Change 129

15-i

SEC

Style

Fig. 5.3. Class stratification for (oh). SEC = socioeconomic class

scale; A, casual speech; B, careful speech; C, reading style; D,
word lists.

two aspects: differentiation, on the one hand,

Social stratification has
and on the other. In the survey of the Lower East
social evaluation,
Side, methods were devised to measure unconscious subjective
reactions to individual values of the phonological variables. These
subjective reaction tests are the principal subject of Ch. 6; here we

will draw from them just the information relevant to the hyper-
correct pattern of the lower middle class.
In the subjective reaction test (SR test), a tape-recording of 22
sentences played for respondents, who are asked to rate the speech
is

heard in each sentence on a scale of occupational suitability. Only

five different speakers are represented, all reading the same standard
text, in which the values of the main phonological variables are
successively concentrated. Since the respondent cannot know ex-
actly how he has marked any given speaker in earlier sentences in
the series, each rating is effectively independent. It is therefore
possible to compare the rating given to a speaker for a given treat-
ment of a phonological variable with that given in a "zero section"
which contains none of the variables. The effects of voice quality,
130 SOCIOLINGUISTIC PATTERNS

reading style, etc., are cancelled out, and the unconscious reactions
to values of a single variable are isolated.
In the test for subjective reactions to (oh), the respondents heard
three different speakers use high vowels in the sentences, "We
[o.^we-z] had [tjo* :klit] milk and [ku: 8 fi] cake around four o'clock."
If the respondent's ratings of all three speakers are equal or lower

to ratings given by him to the same speakers in the zero section, his
response is termed (oh)-negative. The percentage of respondents who
show (oh)-negative response for each class group is as follows:
Lower class 0-2 24%
Working class 3-5 61
Lower middle class 6-8 79
Upper middle class 9 59

In this table, we see the subjective correlate of a number of objective

patterns discussed before. The lower class,which does not partici-
pate in a pattern of social or stylistic variation for (oh), shows no
sensitivity to (oh) in the SR test. Secondly, there is here evidence
of a very general principle that pervades all of the results of the SR
test: those who show the greatest use of a stigmatized feature in
casual speech show the greatest sensitivity to this feature in subjec-
tive reactions. Thus, for example, Italian working-class men, who
show the highest use of stops for (th), also show the greatest sensi-
tivity to this stigmatized feature in the speech of others. Here, the
working class, and particularly the lower middle class, who show
high (oh) vowels in casual speech, show the most negative reactions
to high (oh) in the SR test. Finally, we see that the lower middle
class shows hypersensitivity, in that it goes beyond the reaction of
the upper middle class in its negative response to high (oh).
A similar pattern emerges from subjective reactions to (eh), differ-
ing from that of (oh) only in that we see a higher level of sensitivity
on the part of the two lower classes: the following percentages of
(eh)-negative response were shown by the four class groups:

Lower class 0-2 63%

Working class 3-5 81
Lower middle class 6-8 86
Upper middle class 9 67

Here again we see the hypersensitive behavior of the lower middle

Hypercorrection as a Factor in Linguistic Change 131

class, which reacts more sharply to high (eh) vowels than the high-
est-status class (and is joined here by the working class).
In the case of (r), the lower middle class shows a level of sensitivity

to this variable which goes beyond the reaction of the upper middle
class, even though the behavior of the upper middle class in everyday
speech is closer to the norm. The test for subjective reactions to (r)
is described in detail in the next chapter: essentially, it measures

the degree to which a respondent's reactions are consistent with the

status of (r-1) as a prestige marker. The percentages of (r)-positive
response for the various class groups again show the hypersensitive
pattern of the lower middle class:

Lower class 0-1 50%

Working class 2-5 53
Lower middle class 6-8 86
Upper middle class 9 75

If this hypersensitive behavior is indeed associated with linguistic

change in progress, and consistent with the findings on objective

performance, then we should find that the lower middle class does
not exceed the upper middle class in reaction to the variable (th)
(see Ch.6). The percentage of (th)-sensitive responses for the several

classgroups show that this is indeed the case, as there is no evidence

here for a hypersensitive response of the lower middle class:

Lower class 0-2 58%

Working class 3-5 76
Lower middle class 6-8 81
Upper middle class 9 92

So far, we have been considering evaluation of the speech of others.

In another section of the linguistic interview, we explored the re-
spondent's evaluation of his own usage of the variables. He was
asked to choose which of four alternate pronunciations of a given
word (cards, chocolate, pass, etc.) was closest to the way he usually
pronounced the word himself. This self-evaluation test shows quite
clearly that the extraordinary agreement in subjective response to
the speech of others is not matched by an accurate perception of
the respondents' own speech production. On the contrary, the re-
spondents identified their own speech with those subjective norms
which governed the direction of stylistic variation. For example,
most respondents reported themselves as using variants of (eh) and
132 SOCIOLINGUISTIC PATTERNS

(oh) which were lower than their own speech production, even in
the most formal style.
It appears from this and other evidence that the New York speaker

perceives his own phonic intention, rather than the actual sounds
he produces; in this sense, the pattern which governs the direction
of stylistic variation is determined by a structured set of social
norms. It is phonemic, in the broadest sense.
The behavior of the lower middle class in this test is consistent
with the hypercorrect pattern already shown. Lower-middle-class
speakers show the greatest tendency to report very low values of
(eh) —
that is, they reported that they said [pae:s] instead of [pe: a s]
and by far the greatest tendency to report low (oh) vowels claiming —
that they said [tsoklit] when they actually said [tjbi^klit]. On the other
hand, the lower middle class showed no hypercorrect tendency
towards inaccurate reporting in the self-evaluation of (th).

Linguistic Insecurity of the Lower Middle Class

The great fluctuation in stylistic variation shown by the lower middle

class, their which they
hypersensitivity to stigmatized features
themselves use, and the inaccurate perception of their own speech,
all point to a high degree of linguistic insecurity for these speakers.

Linguistic insecurity may be measured directly by several methods

which are independent of the phonological indexes. In Ch. 4, we
discussed the Index of Linguistic Insecurity (ILI) which contrasts
"own use" and "correctness" for 18 lexical items. Table 5.1 shows
the percentage distribution of scores for the four class groups, for
four levels of the ILI. The lower middle class shows a much higher
concentration at the highest levels of the index than any other
—
group a conclusion reinforced by the attitude of respondents in this
class to the language of New York City.
In general, New Yorkers show a strong dislike for the sound of
New York City speech. Most have tried to change their speech in
one way or another, and would be sincerely complimented to be
told that they do not sound like New Yorkers. Nevertheless, most
of the respondents have been identified by their speech as New
Yorkers whenever they set foot outside of the metropolitan area.
They firmly believe that outsiders do not like New York City speech,
for one reason or another. Most New Yorkers show a strong belief
in correctness of speech, and they strive consciously to achieve such
correctness in their careful conversation.
Hypercorrection as a Factor in Linguistic Change 133

TABLE 5.1

DISTRIBUTION OF INDEX OF LINGUISTIC

INSECURITY SCORES BY SOCIOECONOMIC CLASS

Socioeconomic class
ILI

scores 0-2 3-5 6-8 9

44% 50% 16% 20%

1-2 25 21 16 70
3-7 12 25 58 10
8-13 19 04 10 —
100 100 100 100
N= 16 28 19 10

In all these respects, lower-middle-class speakers exceed all other

New The profound linguistic insecurity of the New York
Yorkers. 2
City speech community is most clearly exemplified in the conscious
statements of the lower-middle-class respondents, as well as their
unconscious behavior.

The Role of the Lower Middle Class in Linguistic Change

We can now approach the question of linguistic change, particu-

larly as it affects the phonological variables. In this discussion, the
principal approach to change will be through internal evidence, in
the distribution of linguistic behavior through the various age levels
of the population. This distribution forms a dimension which we
may apparent time, as opposed to real time. In the complete
call
report of the New York City survey, the relations of apparent time
and real time have been analyzed in detail, and the following cases
considered (Labov 1966a, Ch. 9):

1. A stigmatized feature A. not in process of change

B. in process of change
2. A prestige marker A. not in process of change
B. in process of change
3. Change from below A. early stage
B. later stage, with correction
from above

2. With the exception that a number of lower-middle-class respondents stated that

they had been identified by outsiders as not coming from New York City.
134 SOCIOUNGUISTIC PATTERNS

We can find empirical data for examples of each of these types,

which in general confirm the analysis provided. In particular, the
variable (th) corresponds to the case of a stigmatized feature not
involved in change; (oh) to an early stage of change from below; and
(eh) to a later stage of change from below, with correction from
above. The latter cases are complicated by the fact that the principal
dynamic does not lie in the area of class stratification, but rather
in the contrastive behavior of ethnic groups. In the remainder of this
discussion, we will (r), a prestige marker recently
consider the case of
introduced into New York City speech. This variable is of course
more than a simple phonemic alternation: to the extent that (r-1) is
introduced, a whole series of shifts occur which reverse the main
line of evolution of the New York City vowel system.
To begin with, we must consider the relations of real time to
apparent time for a prestige marker in the process of change. For
the highest-ranking group, which shows the greatest linguistic secu-
rity, these two dimensions match most closely. The oldest members

of the upper middle class would tend to preserve their older prestige
forms, as solidified relatively early in their development, and the
younger members would show the adoption of the newer prestige
form. When we consider the next highest ranking group, usually the
lower middle class, the reverse situation prevails. The great linguistic
insecurity of these speakers would lead to fluctuation in their norms
for formal contexts, and even in middle age they tend to adopt the
latest prestige markers of the younger upper-middle-class speakers.
In this respect, they would surpass the younger members of their
own group, who would not have had as wide an exposure to the
structure of social stratification, and its consequences.
The working-class respondents might be expected to follow the
same general lines of behavior as the lower middle class, but to a
less pronounced degree. Only the lower class would be largely
immune from the tendency to follow the latest prestige norms. On
the basis of these general considerations, we would expect the
following scheme of relations between younger and older sections
of the four class groups, as far as their use of the prestige feature
is concerned:

Lower Working Lower middle Upper middle

class class class class

Younger low [lower] lower high

Older low [higher] higher low
Hypercorrection as a Factor in Linguistic Change 135

We can now consider the empirical data which corresponds to this

diagram. Fig. 4.3 (p. 116) showed the distribution of (r) values for
casual speech. For the two older age groups, there is no indication
of a social significance of (r); but for all those under 40 years old,
there is sharp stratification, in which (r-1) has become the mark of
the highest-status group only. This sudden break corresponds exactly
with a change in the social evaluation of (r), as shown in Table 6.1
(see p. 149).While those over 40 years old show a very mixed pattern
in their subjective response to (r), those from 18 to 39 show a com-
plete unanimity in their positive evaluation of this prestige feature.
Note that in the older group, the lower middle class shows the highest
degree of (r)-positive response.
Table 6.1 has important implications for our general view of
society as well as the development of language within society. We
see the New York City speech community moving towards an in-
creased diversification of speech performance (in Fig. 4.3); but at the
same time, in Table 6.1, there has developed complete agreement in
subjective evaluation. This sudden change may precede and outrun
the parallel development in overt behavior.
The data for successively more formal styles of speech appear in
Fig. 5.4. Beginning with Style B, careful speech, there is an increasing
tendency for all class and age groups to use more (r-1). However,
it is the middle-aged members of the lower-middle-class group who

show the greatest tendency to increase their use of (r-1) in formal

styles, until in Styles D and D', they far surpass the level of the upper
middle class. The hatched areas in Fig. 5.4 represent the degree by
which a given index score exceeds the level of the upper middle
class.
complex structure, relating four variables. From top
Fig. 5.4 is a
to bottom, there is a regular increase in (r-1) with increase of formal-
ity of contextual style; from left to right, there is a pattern of class
behavior, repeated many times, in which the lower middle class leads
the working class and lower class in the use of (r-1). Finally, the
larger left-to-right pattern shows a complex relationship of age level
to (r), in which the younger class 9 speakers show more and more
(r-1), while the reverse pattern holds for the correlation of age and

(r-1) in the three lower classes.

With the background of all of the data presented in this discussion,
the complex structure of Fig.5.4 seems comprehensible, and indeed,
almost predictable. This data made it possible to account for earlier
results that seemed puzzling —
the unexpected configurations that
136 SOCIOLINGUISTIC PATTERNS

emerged from the study of three New York department stores (in
Ch. 2). In Fig. 2.5 we saw that the highest-ranking store, Saks, showed
the expected increase of (r-1) with decreasing age. However, the
middle-ranking store, Macy's, showed the reverse relation, at a lower
The lowest-ranking store, Kleins, showed no clearcut pattern.
level.
When we rearranged Fig. 5.4 in the form of Fig. 2.5, we obtained
Fig. 2.6, Style B (p. 60). Point by point, feature by feature, the two
diagrams matched. This convergence is the strongest possible veri-
fication of the findings of both surveys, for they approached the data
by completely opposite methods. The possible sources of error are
complementary: in each area where the department-store survey is
liable to error, the Lower East Side survey is most reliable, and
vice-versa.

The convergence of the two surveys establishes the validity of the

analysis given above of the relations of apparent time and real time
for a prestige marker which has recently been introduced into the
linguistic structure of a community.

The Role of Hypercorrection in the

Mechanism of Linguistic Change
There is evidence to support the view that the introduc-
sufficient
tion of (r-1) into New
York City is indeed a relatively recent process,
which showed a sharp qualitative increase in the years immediately
following World War II. Traces of a similar development may be
found in other r-less areas, but nowhere is the tendency so great as
in New York City, and nowhere has it penetrated so deeply through-
out the structure of the community. We can, of course, look to the
influence of the broadcast media, where r-less patterns have all but
disappeared, but this is a factor which affects all sections of the
United States. In order to account for the special development of
New York City, it is necessary to look to social and linguistic mecha-
nisms which are peculiar to the metropolitan area.
New York City, as a speech community, may be regarded as a sink
of negative prestige. This is not a recent pattern: the prestige dialect
seems to have been a borrowed one for as long as we can trace it,
and the process we are witnessing here is primarily one of substi-
tuting a northern-midwestern r-pronouncing dialect for the older
r-less prestige dialect borrowed from eastern New England. Yet the
(r)

80
Style B

60 - 20-29 yrs
Class 9
30-39 yrs
40-49 yrs
40 -

r
20 - 50-75 yrs

0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8

80 r-
Style C
20-29 yrs
60 30-39 yrs
l
— —— . Class 9

40-49 yrs

50-75 yrs F
20 -

0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8

80 r-
20-29 yrs

Class 9

0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8

80 r- 40-49 vrs 20-39 yrs

Style D' Class 9

60 50-75 yrs

40

20

0-1 2-5 6-8 0-1 2-5 6-8 0-1 2-5 6-8

Fig. 5.4.Class stratification of (r) in apparent time for Styles B-D'.

B, careful speech; C, reading style; D, word lists; D\ minimal
pairs. Nos. 0-9 indicate socioeconomic class scale.
138 SOCIOLINGUISTIC PATTERNS

differences between the newer and the older forms are profound
from the point of view of phonological structure. All of the New
York City respondents had grown up speaking an r-less dialect: since
they had acquired r-pronunciation long after their primary speech
pattern had been established, it was not possible for them to achieve
consistency in the use of (r-1), even in the most formal context.
The evidence of this study as a whole indicates the following
sequence of events in the genetic development of the complex
linguistic structures that have been displayed here. The child's first
experience in the use of English, at 2 to 3 years old, is usually
dominated by the example of his parents. But from about 4 to 13
years old, his speech pattern is dominated and regulated by that of
the preadolescent group with which he plays. These are the peers
who are able, by their sanctions, to eliminate any deviations from
the dialect pattern of the group. It appears that this preadolescent
period is the age when automatic patterns of motor production are
set:as a rule, any habits acquired after this period are maintained
by audio-monitoring in addition to motor-controlled patterns.
It is in the first year of high school that the speaker begins to

acquire the set of evaluative norms which have been displayed in

this presentation. He becomes sensitive to the social significance of
his own form and other forms; complete familiarity with
of speech,
the norms of the community seems to be attained at the age of 17
or 18. On the other hand, the ability to use prestige forms of speech,
such as r-pronunciation, is not acquired until relatively late: the
youngster seems to begin this process at 16 or 17. A working-class
or lower-middle-class youth never attains the security in the use of
this prestige form which the youngster from an upper-middle-class
family does: as we have seen, even at the age of 30 or 40, the lower-
middle-class speaker may be intent on changing his careful style,
shifting his concept of the prestige norm to meet the most recent
standards. In contrast, the college graduate has attained a certain
degree of security in his use of English, partly through extensive
contact with prestige speakers, and partly through the approval of
his fellow students. Despite the fact that he may thus depend upon
an acquired secondary prestige pronunciation, his use of this form
may remain relatively constant from that time forth.
From the Lower East Side study we can derive systematic evidence
to show the gradual acquisition of these sociolinguistic norms, and
the differential rate of acquisition of different class groups. Fig. 5.5
Hypercorrection as a Factor in Linguistic Change 139

is a composite chart of all the forms of linguistic behavior which

reveal the recognition of such norms. In regard to any given variable,
a subject can show a regular pattern of style shifting towards the
prestige norm, self-report biased towards the norm, subjective reac-
tion tests sensitive to the norm, and overt recognition of the socio-
linguistic feature as a stereotype. The vertical axis is the total score
for the acquisition of sociolinguistic norms, and the horizontal axis
is the age of the subject. All the subjects shown in Fig. 5.5 are
children of the main subjects of the Lower East Side survey. Hori-
zontal lines connect brothers and sisters, so that the upward direction
of these lines shows the rate of acquisition of the norms by age.
When we examine the socioeconomic status of each family, it im-
mediately becomes apparent that there is sharp differentiation by
social class. Children of upper-middle-class families start higher on
the scale and show a more complete response to sociolinguistic
norms than lower-middle-class children, and so on down the line.
While every family is moving in the same general direction, lower-
class youth are still located at a relatively low point on the scale
at the age of 18 or 19. It follows that by the time they acquire a good
knowledge of sociolinguistic norms, they are no longer able to mod-
ify their basic vernacular to achieve consistent productive control.
The problem which must be faced now is: how can such a mecha-

nism lead to the solidification of r-pronunciation as a native speech

pattern, for careful or casual speech? The period when primary
speech patterns are solidified is separated by a gap of at least four
or five years from the period when secondary prestige forms are
learned. The preadolescent may be influenced in his speech pattern
by those who are one or two years older, but it is hard to imagine
that he would be in close contact with those who are four or five
years older. It would then seem that this mechanism can lead only
to permanent stratification, and that r-pronunciation would never
penetrate to the preadolescent period.
Yet one must consider that the original solidification of New York
City as an r-less area must have followed the same pattern as that
we are now witnessing, but in reverse. R-less speech was originally
a prestige form, modeled on the fashion of Southern British speech,
and the present configuration of r-less areas, surrounding Boston,
New York, Richmond, and Charleston must represent the successful
introduction of a prestige form into the primary native speech pat-
tern. If the process was completed once, it can be completed again.
£
« o
e "5 E
01
O
"o
O
XM
u «
+ *
Fig. 5.5. Acquisition of prestige norms in families with two or
more children. Reprinted by permission of the National Council
of Teachers of English.
Hypercorrection as a Factor in Linguistic Change 141

The key to this puzzle may lie in the hypercorrection of the lower
middle class. We have seen that middle-aged, lower-middle-class
speakers tend to adopt the formal speech pattern of the younger,
upper-middle-class speakers. This tendency provides a feed-back
mechanism which is potentially capable of accelerating the intro-
duction of any prestige feature. Instead of a gradual, genera-
tion-by-generation spread of a feature from the highest-ranking
group to the lowest-ranking group, we have here a means by which
the process can be brought to an entirely different tempo. The
lower-middle-class youth (and to a lesser extent the working-class
youth) is in contact with the new prestige pronunciation on two
fronts. On the one hand, he is familiar with the speech of those who
are going to college, whether or not he belongs to this group. On
the other hand, his parents (and his teachers) also use this prestige
pattern in the most formal circumstances. Normally, the dialect used
by his parents has little obvious effect upon his own native dialect
form: it makes no difference whether they come from Maine or
Brooklyn, as far as his own speech is concerned. However, it is not
impossible that repeated use of (r-1) by the parents in the earliest
stage of language learning may lay the groundwork for automatic
and consistent r-pronunciation. Such an effect is not strong in New
York City today, except perhaps in some upper-middle-class homes.
But it may well be that in another half-generation or so, the use of
r-pronunciation in formal styles on the part of adults may have
increased to the point that children will acquire this pattern among
their earliest, motor-controlled habits. One might think that parents
would use only casual speech patterns in speaking to their young
children; but on the contrary, I have heard respondents using the
most careful, r-pronouncing forms when scolding their children.
Since r-pronunciation has been adopted as the norm for the most
careful type of communication, it has perhaps become appropriate
for many types of interaction between parent and child. Hyper-
—
correctness is certainly strongest in women and it may be that the
lower-middle-class mother, and the grade-school teacher, are prime
agents in the acceleration of this type of linguistic change.
The existence of the hypercorrect pattern in New York City has
been established beyond any reasonable doubt. The suggested role
of hypercorrection in the acceleration of linguistic change has been
put forward with the expectation that further empirical studies may
confirm or refute this possibility. Similar investigations may profit-
142 SOCIOLINGUISTIC PATTERNS

ably be carried out in other cities, perhaps in those which do not

show as great a range of stylistic variation in the speech community.
Furthermore, it is necessary to explore more deeply the social moti-
vation which underlies the more systematic and obscure process of
change from below. All of these investigations will help in the
illumination of the important problem of establishing the mechanism
of linguistic change.
 Subjective Dimensions
6 of a Linguistic
in Progress
Change

TRADITIONAL studies of regional dialects in the United

States have shown that isolation leads to linguistic diversity, while
the mixing of populations leads to linguistic uniformity. Yet as we
turn to the study of language differences in metropolitan centers,
a new and different situation appears. 1 Instead of horizontal, spatial
differentiation, we have a vertical cross-section which does not
presuppose isolation of the linguistic strata. On the contrary, groups
living in close contact may participate in rapid linguistic changes
which lead to increased diversity, rather than uniformity.
Studies of the social stratification of language in New York City
show two overall directions of change in the phonological system
contributing to such diversity. One direction is the further develop-
ment of the traditional New York City pattern, continuing the lines
of its earlier evolution, with new phonemic mergers and chain shifts
comparable to the Great Vowel Shift itself. The other large-scale
change is the superposition of a new prestige pattern, almost directly
opposed to the traditional dialect in most of its variable elements.
This chapter will deal with one variable of the new prestige
pattern: (r), the presence or absence of consonantal [r] in final and
preconsonant position in beer, beard, car, card, etc. We have already
dealt with this variable in many ways. Its use in everyday interaction
was examined in the rapid and anonymous survey of department
stores in Ch. 2; the pattern of style shifting of (r) was delineated in
Ch. 3; the pattern of social and stylistic stratification of (r) was
1. This is a revised version of a paper originally given before the Linguistic Society
of America's Winter Meeting in Chicago, Illinois, in December, 1963. It incorporates
some of the more detailed treatment of the subject in Chapter 11 of The Social
Stratification of English in New York City (1966).

143
144 SOCIOLINGUISTIC PATTERNS

outlined in Ch. 4; and it was seen

one element of the hypercorrect
as
pattern of the lower middle class in Ch. 5. At various points we have

considered results of the tests which trace the subjective dimension

of this variable: in this chapter, subjective reactions will be our
primary focus. We will trace a sudden change in the subjective
evaluation of the new prestige pattern by native New Yorkers: the
development of a uniform attitude toward (r), a norm which tran-
scends class differences, in sharp contrast to the increasing differen-
tiation of the objective performance of speakers.
Many patterns of linguistic variation that we find in New York
run parallel to the dominant and increasingly rigid pattern of social
stratification. Techniques for studying linguistic differentiation by
quantitative indexes first developed in the study of Martha's Vine-
yard (Ch. 1) were applied in the Lower East Side study to construct
indexes for five phonological variables. But the other side of the
problem, measuring subjective evaluation to particular variables, had
to be solved for the first time in New York City. Reactions to pho-
nological variables are inarticulate responses, below the level of
conscious awareness, and occur only as a part of an overall reaction
to many variables. There is no vocabulary of socially meaningful
terms with which our informants can evaluate speech for us. Tests
for reactions to social dialects as a whole have been developed by
Lambert and his associates for discussion of the principles involved
(see below and Ch. 8). For our linguistic purposes, we need to elicit
a kind of evaluative behavior that is sensitive enough to register the
effect of a particular linguistic feature —
and a way of presenting a
stimulus which concentrates that feature. The method developed in
this survey appears to answer such a requirement; it will be dis-
cussed here in its application to the social evaluation of (r).

This r-pronunciation is the chief manifestation of the new prestige

pattern which prevails in New
The introduction of /r/
York City.
into this previously r-less dialect is not a mere phonetic change,
substituting (r-1) for (r-0). It has widespread phonemic consequences.
With (r-1), a speaker differentiates guard and god, source and sauce,
bared and bad, which are homonyms 2 in traditional New York City

2. As indicated here, God and guard can be differentiated by the quality of the
vowel, with a back raised [o:] being used where an underlying /r/ follows. However,
the most recent developments in the New
York City system show /a/ before a voiced
final shifting to the class of [a] is now opposed to God, gart-
guard, so that got with
and guard with [t>:]. The classes of source and sauce are also reported to be homonyms
Subjective Dimensions of a Linguistic Change 145

a a
speech: [ga:d] and [ga:d], [sd:s] and [sd:s], [be d] and [be d]. Along
with the use of (r-1) we have the reversal of chain shifts in the low
and mid vowels: as for example, the movement of farmer [fa:ma]
to [fa^ma] and then to [fD:ma] is reversed with the introduction
of (r-1).

The evidence presented in Chs. 3 and 4 for the social differentiation

of this feature is conclusive. The evidence
change in the use of
for
(r) is also quite definite. We know that New
York City was an
r-pronouncing region in the 18th century, and became completely
r-less in the 19th. This change seemed to follow the influence of
London speech, where the r-less pattern was overtly observed by
Walker in 1791. New York followed the example of Boston, Charles-
ton, Savannah, and other Eastern seaboard cities, so the model may
have been simultaneously that of the prestige pattern of New Eng-
land and London. It seems to be one of our best examples of a
—
"change from above" originating in the highest social group— which
eventually spread to the entire speech community and became the
vernacular form. Our first documented evidence for r-less pronun-
ciation in New York City dates to the middle of the 19th century;
Richard Norman has observed that the New York poet Frederick
Cozzens rhymed shore and pshaw in 1856. 3 Babbitt's study of 1896
was the first linguistic report, and it showed that the r-less speech
was the regular vernacular pattern of the city. Babbitt's report as
well as Linguistic Atlas interviews of the 1930's show a completely
r-less dialect for the 21 speakers of the city proper (Frank 1948).
Hubbell's study (1950) and Bronstein's observations (1962) both
mention an increasing use of (r-1) by college students. Fig. 4.2 (p.
00) showed that (r) has now become for most New Yorkers a function
of both the formality of the context and the social status of the
speaker. Our question now is: what is the subjective correlate of this
social differentiation?
The "matched guise" technique developed by Lambert (1967) is

the basic tool now

widely utilized for the study of subjective reac-
tions towards language. The essential principle which emerges from

by most New Yorkers. But recent spectrograph^ work shows that this is not the case
for most speakers. Even when they do not pronounce /r/, the nucleus of source is
higher and/or further back than sauce, although it appears to be "the same" to native
speakers (Labov 1970, Labov, Yaeger, and Steiner 1972).
3. Richard Norman, "An ear to New York," unpublished manuscript.
146 SOCIOLINGUISTIC PATTERNS

Lambert's work is that there exists a uniform set of attitudes towards

language which are shared by almost all members of the speech
community, whether they use a stigmatized or a prestige form of
that language. These attitudes do not emerge in a systematic form
if the subject is questioned directly about dialects; but if he makes

two sets of personality judgments about the same speaker using two
different forms of language, and does not realize that it is the same
speaker, his subjective evaluations of language will emerge as the
differences in the two ratings.
Thus, we find that the first step is to expose each informant to
utterances with contrasting values of the variable in which all other
variables would be held constant. This might be done with synthetic
speech, or with utterances of a trained phonetician. But then we
would have to prove that the phonetic detail of the variant was
equivalent to that of the natural variants, and also, that the artifi-
ciality of the utterance did not itself introduce a new variable that
disturbed subjective reactions. It seems preferable to approach the
problem from above, by using natural utterances of native speakers.
In casting the net a little wider, we may dredge up some extraneous
variables, but we will be certain of our main object, the natural
occurrences of (r).
On Martha's Vineyard we could not follow this procedure, because
in such a small community there were no anonymous native voices
(see Ch. 1). In New York, this was no problem, and we began with
40 versions of a standard reading, "when I was nine or ten .", . .

collected in exploratory interviews on the East Side. The full text

is given in Appendix A to this chapter; the first paragraph is the

"zero" level for calibration of the speaker's voice and reading style,
with none of the five phonological variables used. In each succeeding
paragraph, the five variables are successively concentrated; (eh), (oh),
(eh), (r), and the two th variables together, (th), and (dh). We will
be concerned here with the fourth paragraph, beginning, "I remem-
ber where he was run over, not far from our corner ..." Each
occurrence of (r) is underlined in the text. From five readings of this
paragraph by women speakers, I chose 22 sentences and assembled
them on a test tape in which each was heard twice in succession.
The first five sentences are from the zero paragraph; then sentences
are taken from successive paragraphs in the reading, with speakers
occurring in mixed order.
 Subjective Dimensions of a Linguistic Change 147

This tape was played to the subjects near the end of the interview,
when only the direct discussion of speech remained. The subject had
already read the text himself; now he was asked to register his
feelings about the speech of these other native New Yorkers on a
scale of occupational suitability, shown as Fig. 6.1. He was asked
to place himself in the position of a personnel manager, and use this

Zero ection '

Speech would be •—
2 T. A 5 h a 9 io 11
acceptable for: —
Television personality

Executive secretary

Receptionist
(b - ii

Switchboard operator

Salesgirl
() "
t i ii

\
Factory worker , i \ 1 I ii ii • i

None of these \
i

\
\

1 2 1 3 1 4 1 1 i i i
i
2 i
2 2 2

i l~

k:
1

Television personality 4

"*"-»,
Executive secretary t ~__^

Receptionist
^ > /
Switchboard operator ;
-< p

H
;
\
t

Salesgirl '/ v
.

Factory worker

None of these

consistent /r/ inconsistent /r/ th-section

Fig. 6.1. Subjective evaluation form and patterns for (r) and (th).
148 SOCIOLINGUISTIC PATTERNS

scale to rate the speakers as if they were candidates for jobs. A mark
across the line meant that the speaker could hold that job, and all
below, but did not speak well enough to hold any of the jobs above.
The marks on this chart represent the median performance of
younger middle-class speakers.
Altogether, 200 tests were completed. Some subjects realized that
voices were recurring in the test, but no one could know exactly
how he had ranked the speaker in earlier occurrences. Each rating
is therefore effectively independent. As a result, we can compare

ratings given to a speaker as she was first heard in the zero paragraph,
with ratings given to her in other sentences marked by many occur-
rences of a particular variable. We then use this difference as a
measure of the subject's evaluation of that particular value of that
variable. Voice quality, reading style, intonation, contribute the same
effect to each rating, and their influence may then be cancelled out.
Fig. 6.1 also shows the structure of the test in regard to (r). The
speaker who is first heard as No. 2 in the zero section is later heard
as No. 14 pronouncing all (r) as (r-1) —
usually a weakly constricted
|V|, but indicated here more broadly as [r]:

He [dartid] out about [h&] feet [bifo^] a [kar], and he got

hit [hard].

But in No. 18, she is heard repeating the same sentence with one
(r-0) at the end:

He [dartid] out about [foa^] feet [bifW] a [kar] and he got

hit [ha* :d].

The same alternation is shown by the speaker who heard as

is first

No. 4. In No. 15 she reads the sentence "He darted out ." with
. .

a precise articulation of all (r) as (r-1), but then in No. 19 is heard

saying

We didn't have the [hart] to play ball [dt] [ka :dz] all
[mDrnin].

We thus have two consistent examples of (r-1) in Nos. 14 and 15,

followed at a short distance by two inconsistent versions in Nos. 18
and 19. Those who are sensitive to (r-1) as a mark of prestige pro-
nunciation show a remarkably fine ability to discriminate between
these two pairs, rating 18 a rank or two lower than 14, and 19 one
or two ranks below 15. Only a few especially knowledgeable subjects
Subjective Dimensions of a Linguistic Change 149

are aware of this inconsistent (r): most people cannot explain why
they lower the ratings for 18 and 19, though they are quite firm in
their opinion.
How shall we
reduce these ratings to a single index? Let us con-
sider that there are two possible responses consistent with the rec-
ognition of (r-1) as a prestige marker: rating 18 and 19 lower than
14 and 15 respectively, or in view of the fact that they are after all
the same speakers, rating 18 as the same as 14, 19 the same as 15.
Either of these reactions, combination, we will treat as
or a
(r)-positive. If in either case, the subject followed a contrary direc-
tion, rating 18 higher than 14 or 15 higher than 19, we will call his
reaction (r)-negative.
Table 6.1 shows the percentages of (r)-positive response to the
two-choice test for four age levels, and five divisions of the socio-
economic scale (the same divisions used for the original class strati-
fication of (r) in Ch. 4). In this table, our attention is immediately

TABLE 6.1

(r)-POSITIVE RESPONSE TO THE

TWO-CHOICE TEST BY SOCIOECONOMIC
CLASS AND AGE

Age 0-1 2-3 4-5 6-8 9 Total

N
8-17 16% 57% 67% 89% (50)% 61%
6 14 12 9 2
18-19 100 100 100 100 100
2 2 1 3
20-39 100 100 100 100 100 100 3 6 7 11 5
40- 63 67 50 70 57 62 8 18 8 10 7

taken by a regularity more absolute than any that has been encoun-
tered so far. One hundred percent of the speakers from age 20 to
39 showed (r)-positive reactions to the two-choice test, but only 62
percent of those over 40. Furthermore, this regularity is extended
to the respondents 18 and 19 years old. A simple four-cell table shows
a remarkable distribution of respondents who show (r)-positive and
(r)-negative response for two age levels:

Age (r)-positive (r)-negative

18-39 40
40- 31 20
150 SOCIOLINGUISTIC PATTERNS

This is a startling result. beyond a doubt that there

It establishes
is two age groups, and
a great difference in the behavior of these
it implies that the recognition of (r-1) as a prestige marker has

reached the stage of absolute regularity which we associate with

completed linguistic changes.
In Table 6.1, class differences have largely disappeared, and only
differences in age level stand out. This is a particularly striking fact,
since in Chs. 2 and 4, (r) showed the finest and most regular class
stratification of all of the variables in the Lower East Side Study.
We now uniform stratification of (r) in performance
find that this
is accompanied by a uniform evaluation of the prestige norm by

younger speakers of all classes. There has been no overall increase

in the use of (r-1) in the vernacular of everyday speech; but there
has been a sudden increase in the stratification of (r). Fig. 6.2a shows
—
the social stratification of (r) in Style A casual speech for those —
under 40 and those over 40 years old in the Lower East Side sample.
In the older age level there is no particular difference in the use of
the four socioeconomic groups. But for younger speakers, there is
a sharp differentiation between the highest group the upper middle—
class —
and the rest. The total amount of (r-1) used may actually have
—
decreased what has increased is the differentiation of social class.
We can compare this development with the subjective evaluations,
shown in Fig. 6.2b. Here we shift from more or less random response
for the older level to a uniform positive evaluation for the younger.
Fig. 6.2 thus demonstrates the general principle that social stratifica-
tion in the use of a variable is correlated with a uniform subjective
evaluation of it.

A more may now be constructed to include the two

difficult test
original sentences 2 and 4. Consistent recognition of (r-1) as a prestige
marker should lead to the rating of sentences 14 and 15 equal or
higher than the zero level of speakers 2 and 4. Instead of a two-choice
test, a four-choice test will be used to establish an (r)-positive re-

sponse. For an (r)-positive rating the subject must rate the consistent
use of (r-1) equal or higher than the zero level, and the inconsistent
use equal or lower than the consistent use. A reversal in any one
of these four choices will give the subject an (r)-negative rating. Table
6.2 shows the data for the four-choice test which corresponds to
Table 6.1 for the two-choice test.

The more difficult four-choice test reduces the overall percentages

slightly, but preserves the relationships intact. The results of the
 —
O 00
0> CM

cc o

(D
DC
<

<

I 1 i
1
r— "T 1
—1—~l 1

o
LO
o o
ro
O
«—
o
CN
O o
o O O
CO
o
K
o
sO
o
LO
Tf CTi

sajoDs xapu; (j) aSejaAy ssuodsaj 9Ai}!sod-(j)

Fig. 6.2.(a) Class stratification of (r) by age in casual speech

(b)Subjective evaluation of (r) by age and class groups.
Nos. 0-9 indicate socioeconomic class scale.
152 SOCIOLINGUISTIC PATTERNS

TABLE 6.2
(r)-POSITIVE RESPONSE TO THE
FOUR-CHOICE TEST BY SOCIOECONOMIC
CLASSIFICATION AND AGE

Socioeconomic class

Age 0-1 2-3 4-5 6-8 9 Total

N
8-17 00% 36% 33% 67% (50)% 37
6 14 12 9 2
18-19 50 100 100 100 88
2 2 1 3
20-39 75 84 86 100 100 87 3 6 7 11 5
40- 38 44 25 70 57 48 8 18 8 10 7

four-choice test are more impressive in several ways. If we take the

total number of choices which the respondents had to make, for
sentences 14, 15, 18, and 19, the following contrast between age
groups appears:

Choices consistent No differences Choices inconsistent

with recognition of from zero with recognition of
Age prestige marker level prestige marker
18-39 128 35 5
40- 108 48 48

The consistency of the younger group is the more remarkable when

one considers that sentences 14 and 15 are widely separated from
sentences 2 and 4 in the course of the SR test. Only five deviations
from the pattern of (r)-positive responses appear for younger
speakers. Furthermore, these deviations were all in class 4 and below,
so that it is evident that minor differences in sensitivity to (r) still
exist among the several class groups. 4

4. Differences exist in the fineness of reaction to sentences 18 and 19. For all of
the variables, the average values of the absolute differences in ratings of the same
speakers are correlated with class. In the present case, the higher-ranking classes seem
to hear the difference between sentences 14 and 18, 15 and 19, as slight differences;
the ratings of the speakers drop one or two ranks only. Lower-ranking respondents
react as a rule in an exaggerated fashion, and penalize the inconsistent utterances
by rating them much lower than sentences 14 and 15. If we sum the absolute differences
between 14 and 18, 15 and 19, for all respondents between 18 and 39, we obtain the
following progression:

Working class and Lower middle Upper middle

lower class 0-5 class 6-8 class 9
3.9 3.5 3.1

N= 20 12 8
 Subjective Dimensions of a Linguistic Change 153

From Tables 6.1 and 6.2, there can be no doubt that the age differ-
ences in (r)-positive response are well established. There is little
room for differences of sex or ethnic group, or even socioeconomic
class, in the face of such a general change in apparent time. We see
that socioeconomic differentiation, obscured in the two-choice test,
reappears to some extent in the four-choice test. The differences in
age groups are repeated in every class, however, they are larger in
magnitude than any difference between classes.
The break is actually sharper than it appears in Tables 6.1 and
6.2. Fig. 6.3 shows the percentages of (r)-positive response for nine

age groups. The break seems to come exactly with those who were
born in 1923 as far as our sample is concerned. No particular direc-
tion for those over 40 is shown in this figure, while at the other end
of the scale, it seems to be just about at the age of 18 that young
people learn to recognize the social significance of this feature.

100 ——— — \ \
\ \
/
/ /
/
\
\
\
\
90 - / \ \
/
/ / \ \
/ / \ \
80 / / \ \
/ / \ \
/ / \ \
/ / \ \
70 \ \
// \
\
\
\
a 1
\ \
60 -
1
1
1
\
\
\
\
^^
1

50 - 1 \
1 \
\.
1
1
^*"«««.

40

1 1 1 1
1 1
1

& LD

m c
I I

Age group
Fig. 6.3. Development of social evaluation of (r) in two subjective
reaction tests. Solid line, two-choice test; dotted line, four-
choice test.
154 SOCIOLINGUISTIC PATTERNS

These results are clear, but there are still doubts to resolve. First,
the consistent response may not be governed by variation in (r), but
some other variables instead; second, the difference in response
between age groups may be spurious if the younger and older groups
differ in other respects besides age.
It is true that there is a class bias in the age groups. The younger

group has more education, and better prospects, as a natural product

of the upward movement of succeeding generations on the Lower
East Side. And the higher classes show somewhat better response
in the recognition of (r-1). Let us then divide both age groups down
the middle, into two overall groups: upper middle combined with
lower middle on the one hand, working class combined with lower
class on the other. We find that the difference in age groups applies
to both upper and lower halves equally. For the four-choice test, in
terms of percentage of (r)-positive response:

Upper half Lower half

over 40 63 39
18-39 95 79

If now we see that the difference in age levels is preserved in both

class groups, it is still possible that it is due to the fact that the
younger group hears better, or has more interest in the test, or shows
less fatigue. We can check this question by correlation with another
variable: the perception of the initial consonants of thing and then.
Figure 6.1 also shows the structure of the test in regard to (th) and
(dh) in sentences 20, 21, 22. In 20, the speaker of 3 is heard with a
typical mixture of fricative and stop and affricate forms for (th) and
(dh).

20. [SJere's some[9]ing strange about [d]at — how I can re-

member every[0]ing he did: [S]is [t0]ing, [5]at [0]ing, and
[d]e o[d]er [0]ing.

In sentence 21, the speaker of sentence 4 is heard pronouncing the

same sentence with all fricative forms: [9] and [3]. And finally, the
speaker of 3 and 20 is heard again in sentence 22 with

I suppose it's [S]e same [t]ing wi[t] most of us.

Any pattern in which the middle sentence, 21,

is rated higher than

both 20 and 22, is consistent with the recognition of the fricatives

in [Gin] and [Sen] as higher in prestige than the affricates in [tOirj]
Subjective Dimensions of a Linguistic Change 155

and [d<5en] or the lenis stops in [tin] and [den]. As far as we know,
the status of these variables is not changing, and usage
divided is

on class lines alone. There

evidence for the stigmatization of the
is

stop form in America going back to as early as 1783. 5 Since these

sounds are harder to hear than most, and occur at the end of the
test, we can say that those who do respond positively with a low-

high-low pattern are roughly comparable in hearing, span of atten-

tion, and sensitivity to prestige forms of speech.
For the subgroup of respondents who showed (th)-sensitive re-
sponse, the factors of fatigue, loss of hearing, or lack of interest
should be considerably less than for the (th)-insensitive group. If
there is a connection between these factors and (r)-response, the
difference between age levels in (r)-response will appear significantly
reduced for respondents who were uniformly (th)-sensitive (see Table
6.3).

TABLE 6.3
(r)-POSITIVE RESPONSE TO THE FOUR-CHOICE TEST
BY AGE AND (th)-SENSITIVITY

(th)-sensitive (th)-insensitive
Age All respondents respondents respondents
N
20-39 87% 92% 80% 34 ^9 5
40- 48 46 50 51 35 16

Itappears that (r)-positive response is independent of (th)-sensi-

tivity.The pattern of (r)-positive response by age levels is repeated
for the (th)-insensitive group as well as the (th)-sensitive group. Only
5 younger respondents showed (th)-insensitive ratings: 4 of them
were (r)-positive on the four-choice test. Sixteen older respondents
were (th)-insensitive: 8 of these were (r)-positive, and 8 (r)-negative.
We may conclude that the (r)-positive response which was meas-
ured is a function of age, and that the factors of hearing loss, fatigue,
or lack of interest in the test are not likely to have played a part
in this result.
We may now consider the possibility that other variables associ-

5. As Richard Norman points out, Noah Webster marked the use of stops for (th)
and (dh) as nonstandard in the late 18th century. documents we have
The earliest
stigmatizing this feature in New
York date from the middle of the 19th century
travelers' —
accounts and plays about the Bowery Boys but we can assume that the
social stereotype arose earlier.
156 SOCIOLINGUISTIC PATTERNS

ated with sentences 14 and 15, 18 and 19 were responsible in whole

or in part for the differential reaction of the age levels to the pairs
of sentences.
When a speaker shows an inconsistency in (r), she is likely to show
other pronunciation features which are less typical of careful speech.
For example, in sentence 19, speaker 4 hesitated after the word ball
[trouble with her eyesight]; her consonants were not formed or
released as forcefully in 19 as in 15: she did not, for example, pro-
nounce the final /t/ in sentence 19, and one or two respondents
noticed this.
Sentence 18 was taken from a first reading of the text by speaker
2, and in this reading she was a little further away from the micro-

phone than in the second reading, from which sentence 14 is taken.

Such differences as these may account for a part of the reaction
which placed sentences 14 and 15 higher than sentences 2 and 4,
18 and 19. However, if this is true, there is no reason to suspect that
out-of-town speakers would react any differently from native New
Yorkers to the text. They should be able to hear such differences
as preciseness of articulation, speed of reading, or distance from the
microphone, just as well as New Yorkers. We may therefore turn
to the 32 speakers in our survey who came to New York after they
were 13 years old, to check this point.
Table 6.4 shows the out-of-town percentages of (r)-positive re-
sponse to the four-choice test. The older out-of-town speakers show
about the same response as New Yorkers did, but the younger
speakers, instead of showing more (r)-positive response, actually
show Jess. This relationship we would expect if the
is exactly what
test does measure the special New York response to (r-1) as a new
prestige marker. The older out-of-town respondents have had about
as much exposure to the new prestige form in New York City as
the native New York respondents. But the younger out-of-town
subjects were raised outside of New York, away from this influence,

TABLE 6.4
RESPONSE TO THE
(r)-POSITIVE
FOUR-CHOICE TEST FOR NEW YORK
AND OUT-OF-TOWN RESPONDENTS
Age New Yorkers Out-of-town
N
20-39 87% 40% 14 10
40- 48 50 51 22
Subjective Dimensions of a Linguistic Change 157

and have had only a brief exposure to it. The distribution of

(r)-positiveresponse among out-of-town speakers therefore confirms
the fact that it is the variable (r) which is the focus of subjective
reactions. 6
We can use the out-of-town speakers to check this question in
another way. If (r) is indeed the variable which is being measured
in the SR test, then speakers who come from an r-pronouncing region
should have more tendency to show (r)-positive response than those
who come from an r-less region, where an r-less dialect has prestige.
This is indeed the case. For the four-choice test, out-of-town respond-
ents show very different results depending on whether they come
from an r-less or an r-pronouncing region. 7

Out-of-town Respondents

From r-pronouncing From r-less

region region

(r)-positive 10 5
(r)-negative 7 10

The evidence that we have presented shows that the reactions to

sentences 14 and 15, 18 and 19, are indeed reactions to the use of
(r). The evidence for a sudden change in the norms of r-pronuncia-

tion cannot be explained by the presence of associated variables.

The original presentation of subjective reactions to (r) in this chapter
showed a sudden increase in (r)-positive response in apparent time,
and this increase points to a corresponding change in the structure
of the New York City speech community in real time. As we have

6. A majority of the out-of-town respondents were black. This group is therefore

not comparable to the New York respondents, and it is possible that the special (r)

response of black subjects was responsible for the difference. However, when we
compare only black out-of-town subjects with only black New York subjects, the
difference in (r) response holds. The younger New York black respondents showed
even more consistent (r)-positive response than the younger New York white respond-
ents.
7. In this case, we do not expect to find that the low use of (r-1) among younger
speakers is associated with high sensitivity to this prestige feature. The younger
out-of-town subjects who were raised in an area where the prestige norm was an
r-less dialect would have no reason to stigmatize sentences 18 and 19, or award high
ratings to 14 and 15 on the basis of consistent (r-1) pronunciation— except in so far
as they have absorbed the New York City standard. It seems natural that they could
apply this standard less accurately than those who had been born and raised in the
city.
158 SOCIOLINGUISTIC PATTERNS

noted before, the change seems to be closely associated with the

period of World War II: all those in the sample who were raised
during and after the war show a uniform (r)-positive response in the
test.

These results support other evidence that bears

on the definition
of a speech community. As Fig. 6.2 shows, a speech community
cannot be conceived as a group of speakers who all use the same
forms; it is best defined as a group who share the same norms in
regard to language. In this sense, older and younger speakers in New
York City belong to slightly different speech communities, with a
fairly distinct discontinuity for those speakers born in the mid 1920's.
Our conclusion should be sufficient to prove that we have indeed
found a means of measuring subjective reactions to a single pho-
nological feature, despite the co-occurrence of other socially mean-
ingful variables. This measurement has shown us an astonishing
regularity in the evaluation of (r) by younger speakers the same —
speakers who are sharply divided along class lines in their use of
(r) everyday speech. Thus we see a society moving towards greater
in
linguistic diversityamong subgroups that are in close contact, and
indeed share a common set of linguistic norms. This may be a
characteristic intermediate stage as a language change moves to-
wards completion. Or we may witness a hardening of the situation
into a permanent stratification of language. Since we are now in a
position to observe further developments along two dimensions
rather than one, we can expect that the resolution of this question
will add considerably to our understanding of linguistic evolution.
In the next chapter we apply some of these findings to the general
problem of elucidating the mechanism of linguistic change.

Appendix A
Standard Reading for Five Phonological
Variables of New York City Speech
zero
When was nine or ten, had a lot of friends who used to come
I I

over to my house to play. remember a kid named Henry who

I

had very big feet, and I remember a boy named Billy who had
no neck, or at least none to look at. He was a funny kid, all right.
Subjective Dimensions of a Linguistic Change 159

hi
We always had chocolate milk and coffee cake around four
o'clock. My dog used to give us an awful lot of trouble. He jumped
all over us when he saw the coffee cake. We called him Hungry

Sam.
/ae/
We used play Kick-the-can. One man is "IT": you run past
to
him as fast as you can, and you kick a tin can so he can't tag you.
Sammy used to grab the can and dash down the street. We'd chase
him with a baseball bat, and yell, "Bad boy! Bad! Bad!" But he
was too fast. Only my aunt could catch him. She had him do tricks,
too: she even made him ask for a glass of milk, and jump into a
paper bag.
/r/
I remember where he was run over, not far from our corner.
He darted out about four feet before a car, and he got hit hard.
We didn't have the heart to play ball or cards all morning. We
didn't know we cared so much for him until he was hurt.

th
There's something strange about that — how I can remember
everything he did: this thing, that thing, and the other thing. He
used to carry three newspapers in his mouth at the same time.
I suppose it's the same thing with most of us: your first dog is like

your first girl. She's more trouble than she's worth, but you can't
seem to forget her.
 On theMechanism of
7 Linguistic Change

IN our studies of Martha's Vineyard and the New York City

speech community, regular relations have been found where previ-
ous reports showed chaotic oscillation or massive free variation.
These findings have enabled us to state a number of sociolinguistic
principles concerning the relations of stylistic variation, class strati-
fication, and subjective evaluation. In other studies, such data have
been applied to general problems of linguistic structure: particularly,
the characterization of linguistic rules (see Ch. 8). The most general
issues that have been raised so far concern the explanation of lin-
guistic change, and in this chapter we will attempt to focus the data
assembled so far on the mechanism of linguistic change, specifically,
sound change.

The Problems of Linguistic Evolution

Despite the achievements of 19th-century historical linguistics,

many avenues to the study of linguistic change remain unexplored.
In 1905, Meillet (1921:16) noted that all of the laws of linguistic
history that had been discovered were merely possibilities:

... it remains for us to discover the variables which permit or incite the
possibilities thus recognized.

The problem as we face it today is precisely that which Meillet

outlined over 60 years ago, for little progress has been made in
ascertaining the empirical factors which condition historical change.
The chief problems of linguistic evolution might be summarized as
five questions:

1. Is there an overall direction of linguistic evolution?

160
On the Mechanism of Linguistic Change 161

2. What are the universal constraints upon linguistic change?

3. What are the causes of the continual origination of new linguistic
changes?
4. By what mechanism do changes proceed?
5. Is there an adaptive function to linguistic evolution? 1

One approach tolinguistic evolution is to study changes completed

in the past. This has of course been the major strategy of historical
linguistics, and it is the only possible approach to the first two
—
questions the direction of linguistic evolution, and the universal
constraints upon change. On the other hand, the questions of the
mechanism of change, the inciting causes of change, and the adaptive
functions of change are best analyzed by studying in detail linguistic
changes in progress. The mechanism of linguistic change will be the
chief topic of the discussion to follow; however, many of the con-
clusions will plainly be relevant to the questions of inciting causes
and adaptive functions of change, and it will be apparent that more
complete answers to these questions will require methods similar
to those used here.
An essential presupposition of this line of research is a uniformi-
tarian doctrine: that is, the claim that the same mechanisms which
operated to produce the large-scale changes of the past may be
observed operating in the current changes taking place around us.

A Strategy for the Study of Linguistic Changes in Progress

Although answers to the five questions given above are the

ultimate goals of our current research, they do not represent the
actual strategy used. For the empirical study of changes in progress,
the task can be subdivided into three separate problems which
jointly serve to answer the questions raised above.

1. The transition problem is by which one stage

to find the route
of a linguistic change has evolved We wish
from an earlier stage.
to trace enough of the intervening stages so that we can eliminate
all but one of the major alternatives. Thus questions of the regu-

1. This question is all the more puzzling when we contrast linguistic with biological
evolution. It is difficult and animal kingdoms
to discuss the evolution of the plant
without some reference to adaptation to various environments. But what conceivable
adaptive function is served by the efflorescence of the Indo-European family? On this
topic, see Hymes 1961 and Ch. 9.
162 SOCIOLINCUISTIC PATTERNS

larity of sound change, of grammatical influence on sound change,

of "push chains" versus "pull chains," of steady movement versus
sudden and discontinuous shifts, are all aspects of the transition
problem.
2. The embedding problem is to find the continuous matrix of social
and linguistic behavior in which the linguistic change is carried.
The principal route to the solution is through the discovery of
correlations between elements of the linguistic system, and be-
tween those elements and the nonlinguistic system of social be-
havior. The correlations are established by strong proof of con-
comitant variation: that is, by showing that a small change in the
independent variable is regularly accompanied by a change of the
linguistic variable in a predictable direction. 2
3. The evaluation problem is to find the subjective (or latent) cor-
relates of the objective (or manifest) changes which have been
observed. The indirect approach to this problem correlates the
general attitudes and aspirations of the informants with their
linguistic behavior. The more direct approach is to measure the
unconscious subjective reactions of the informants to values of
the linguistic variable itself.

With tentative solutions to these problems in hand, it would be

possible to provide an explanation of a linguistic change which
answers the three questions of inciting cause, mechanism, and
adaptive function. 3 As in any other investigation, the value of an
explanation rises in relation to its generality, but only to the extent
that it rests upon a foundation of reliable and reproducible evidence.

2. The concept of the linguistic variable is that developed in Labov 1966a and
reflected in Ch. 2. It is further developed in the notion of variable constraints on
variable rules (Ch. 8). The definition of such a variable amounts to an empirical
assertion of covariation, within or without the linguistic system. appears that the It

fundamental difference between an explanation of a linguistic change, and a descrip-

tion, is that a description makes no such assertion. In terms of a description of change,
such as that provided by Halle 1962 there is no greater probability of the change
taking place in the observed direction, as in the reverse direction. Note that the
embedding problem is presented here as a single problem, despite the fact that there
are two distinct aspects: correlations within the linguistic system, and with elements
outside the system. The main body of this chapter provides justification for this
decision.
3. To these three problems we can add the constraints problem and the actuation

problem as developed in Weinreich, Labov, and Herzog 1968. The search for general
On the Mechanism of Linguistic Change 163

The Observation of Sound Change

The simplest data that will establish the existence of a linguistic

change is a set of observations of two successive generations of
—
speakers generations of comparable social characteristics which
represent stages in the evolution of the same speech community.
Hermann (1929) obtained such data at Charmey in 1929, by develop-
ing Gauchat's (1905) original observations of 1899. We have such data
for Martha's Vineyard, adding the 1961 observations to the 1933 data
of the Linguistic Atlas (Kurath et al. 1941). For New York City, we
add the current data of 1963 to the Linguistic Atlas data of 1940; in
addition, we have many other reports, including the excellent ob-
servations of Babbitt in 1896 to add further time depth to our analysis
(Frank 1948; Kurath and McDavid 1951; Hubbell 1950; Babbitt 1896).
Solutions to the transition problem proposed here will depend
upon close analysis of the distribution of linguistic forms in apparent
time — that is, along the dimension formed by the age groups of the
present population. Such an analysis is possible only because the

original simple description of change in real time enables us to

distinguish age-grading in the present population from the effects
of linguistic change (Hockett 1950).
The evidence obtained in the research reported here indicates that
the regular process of sound change can be isolated and recorded
by observations across two generations. This process is characterized
by a rapid development of some units of a phonetic subsystem, while
other units remain relatively constant. It affects word classes as a
whole, rather than individual words: yet these classes may be defined
by a variety of conditions, morphophonemic and grammatical as well
as phonetic. It is regular, but more in the outcome than in its incep-
tion or its development. Furthermore, it appears that the process of
sound change is not an autonomous movement within the confines
of a linguistic system, but rather a complex response to many aspects
of human behavior.
Some comment is required on the possibility of observing regular
sound change, since arguments inherited from the neo-grammarian

constraints on sound change is one theme in our current work on sound change in
progress (Labov, Yaeger, and Steiner 1972). Some progress on the actuation riddle
is suggested by the conclusion of this chapter.
164 SOCIOUNGUISTIC PATTERNS

controversy have impeded the progress of empirical research in this

arela. The inheritors of the neo-grammarian tradition, who should
be 'most interested in the empirical study of regular change in prog-
ress, have abandoned the arena of meaningful research in favor of
abstract and speculative arguments. Indeed, Bloomfield (1933:347,
365) and Hockett (1958:439,444) have maintained that phonetic
change cannot in principle be observed by any of the techniques
currently available. 4 Hockett has proceeded to identify sound change
with a level of random fluctuations in the action of the articulatory
apparatus, without any inherent direction, a drift of the articulatory
target which has no cognitive, expressive or social significance. 5 All
of the empirical observations of change in progress which have been
reported are explained as the results of a complex process of bor-
rowing, and are relegated to a type of linguistic behavior known as
the fluctuation or conflict of forms. No claims are made for the
regularity of this process, and so the basic tenet of the regularity
of sound change has been deprived of all empirical significance.
Furthermore, the changes which actually are observed are regarded
as unsystematic phenomena, to be discussed with anecdotal evi-
dence, subject to forces "quite outside the linguist's reach," factors
which "elude our grasp," fluctuations "beyond our powers" to record
(Bloomfield 1933:343-68).
The evidence of current research suggests that this retreat was
premature, that the regular process of sound change can be observed
by empirical methods. The refinements in methodology called for
are not the mechanical elaborations suggested by the writers cited
above; for the mere multiplication of data only confounds analysis

Hockett writes: "No one has yet observed sound change: we have only been able
4.

to detectit via its consequences. We shall see later that a more nearly direct observa-

tion would be theoretically possible, if impractical, but any ostensible report of such
an observation so far must be discredited." His theoretical proposal is that "over a
period of fifty years we made, each month, a thousand accurate acoustic records . . .

all from the members of a tight-knit community." The suggestion to multiply the data
in this way is not necessarily helpful, as the experience of sociological survey analysts
has shown: for relatively small numbers are needed to measure change in a population
if the bias of selection is eliminated or minimized. Otherwise, we merely multiply
the errors of measurement.
5. According to Hockett, the variables responsible for sound change include "the
amount of moisture in the throat, nose and mouth of the speaker, random currents
in his central nervous system, muscular tics the condition of the hearer's outer
. . .

ear [presence of wax or dirt] ..." (1958:443-444).

On the Mechanism of Linguistic Change 165

and perpetuates the bias of selection. It is rigor in the analysis of

a population and in the selection of informants which is required.
Furthermore, we need ingenuity in the resolution of stylistic varia-
tion, to go beyond the sterile method of endless dissection into
idiolects. With such techniques, we find that regularity emerges
where only confusion was seen before. Random fluctuations in
articulation can certainly be found: indeed, this is the level of "noise"
which prevents us from predicting the form of every utterance which
our informants will make. But it would be an error to ascribe a major
role to such fluctuations in the economy of linguistic change. The
forces which direct the observed changes appear to be of an entirely
different order of magnitude, and the changes take place much more
rapidly than any process of could account for. 6
random drift
A single example of a sound change observed will be used to
illustrate the general approach to solving the transition, embedding
and evaluation problems. This example is one of the simplest
—
cases that of the centralization of (aw) on Martha's Vineyard,
discussed in Ch. 1. In the development of this case, some new evi-
dence will be presented on the mechanism of sound changes.

The Centralization of (aw) on Martha's Vineyard

We begin with a clearcut case for the existence of a linguistic

change from observations in real time. In 1933, Guy Lowman found
no more than the barest trace of centralization of /aw/; the signifi-
cant variation observed was the fronting of /aw/ from [au] to [aeu].
In 1961, a comparable set of older eighth-generation descendants of
Yankee settlers from the same villages showed a very pronounced
centralization of /aw/ — now clearly the variable (aw).
The transition problem is studied through a detailed examination
of the distribution of forms through apparent time — that is, through
the various age levels in the present population. The first step in the
analysis is to construct a quantitative index for discrete values of
the variable:

6. Thus the following table contrasts the two points of view:

Neo-grammarian: sound change fluctuation of ultimate
forms regularity
Present discussion: sublinguistic sound change ultimate
fluctuations regularity
166 SOCIOLINGUISTIC PATTERNS

aw-0 [au]
aw-1 [a"u]
aw-2 [BU]
aw-3 [au]

The index of centralization was constructed from this scale by

averaging the numerical values assigned to each variant. Thus
(aw)-OO would mean no centralization at all, while (aw)-300 would
mean consistent centralization at the level of [au]. This index was
applied to interviews with 69 informants by rating each of the words
in which (aw) occurred. The first approach to the transition problem
can then be made by correlating average (aw) index scores for these
interviews with the age level of the speakers. The first three columns
of Table 7.1 (a rearranged version of Table 1.2) show a regular
correlation, in which the centralization index rises regularly for four
successive age levels.
The overall tendency of Table 7.1 represents an amalgamation of
many different types of speakers and many different trends in the
use of (aw). Fig. 7.1 presents a more detailed analysis of the transition
problem for a critical subgroup. Here are displayed the percentage
distribution of lexical items for eight individuals from 92 to 31 years
of age.The horizontal axes show the four coded levels of the variable
(aw).The vertical axes are the percentages of lexical items used with
each variant. The vocabulary is broken into two sections that are
tabulated separately: the solid line represents words in which (aw)
isfollowed by a voiceless obstruent, as in out, house, about, mouth;
the broken line represents all other words (and principally those

ending in a nasal, as in town, found, or with no consonant final, as

in now, how, etc.) 7
The first diagram in Fig. 7.1 is not that of an individual, but shows
the composite results for the four Linguistic Atlas informants inter-
viewed in 1933. They show only the barest trace of centralization.
The second diagram, b, is that of the oldest informant of 1961, a man
7. The phonetic conditioning of (aw) was actuallymore complex than this: both
following and preceding consonants are involved. We have returned to the phonetic
mechanism of the change in our recent studies of sound change in progress. Spectro-
graphic studies confirm the gradual development of phonetic conditioning, the gradual
rise in importance of a following voiceless obstruent, and the final development of
discrete allophonic conditioning (Labov 1972c). The spectrographs displays suggest
that there may not be a retrograde movement of the voiced environments, but rather
that they are heard as the raising in voiceless environments reaches mid position.
On the Mechanism of Linguistic Change 167

TABLE 7.1

CENTRALIZATION INDEXES BY AGE LEVEL

Generation Age (aw) (ay)

la over 75 22 25
lb 61-75 37 35
Ma 46-60 44 62
lib 31-45 88 81

92 years old. The average age of the Atlas informants was 65 years;
Mr. H. H. Sr. would have been 64 years old in 1933, and so he is
of the same age group. His centralization profile is quite similar to
that of the Atlas informants in a. In c, we have an 87-year-old woman
who shows only a slight increase in centralization. Fig. d, Mr. E. M.,
83 years old, indicates a small but distinct increase in the occurrence
of variant (aw-2). Mr. H. H. Jr., in e, is considerably younger; he is
61 years old, the first representative of the next generation, since he
is the son of Mr. H. H. Sr. Here we have a marked increase in

centralization, with both classes of words centered about a norm

of (aw-1). In /, Mr. D. P., 57 years old, shows a distinct difference
between words ending in voiceless obstruents and all others; the first
are now centered upon a norm of (aw-2), while the second group
is concentrated at (aw-1). This process is carried further in the speech

of Mr. P. N., 52 years old, who shows perfect complementary dis-

tribution. Before voiceless obstruents, /aw/ has an allophone which
is almost always (aw-2), while before other terminals it is usually

uncentralized. And at this point, there is no overlap in the distribu-

most extreme case of centralization, we see
tion. Finally, in h, the
an even sharper separation: this is Mr. E. P., 31 years old, the son
of Mr. D. P. in f.
On the right hand side of Fig. 7.1 are the figures for the actual
numbers of lexical items observed, and the composite index scores
for each of the eight cases. It may be noted that (aw) is only one-third
as frequent as (ay), and the regularity which appears here does not
require a vast corpus of observations. The regularity emerges through
the controlled selection of informants, methods of elicitation, and
of recording the data.
The eight diagrams of Fig. 7.1 represent the most homogeneous
type of population. All of the speakers are Yankee descendants of
the original settlers of the island, all are interrelated, many from the
 (aw-) (aw-)
1 2 12 3

Four LANE _C°

informants, else.
av. 65 yrs. YYYttSSTTrrT^.

Mr. H. H., Sr.

92 yrs. else.

Mrs. S. H., 8 2 _C°

87 yrs. 12 3 else.

Mr. E. M. _C°
else.
83 yrs.

Mr. H. H., Jr.

_C°
60 yrs. else.

Mr. D. P. 1 3 10 _C°
57 yrs. 9 15 3 else.

Mr. P. N. 17 2 _C°
52 yrs. 10 else.

Mr. E. P.
9 9
else.
31 yrs.

Fig. 7.1. Stages in the centralization of (aw) on Martha's Vine-

yard, Mass. Solid line: (aw) followed by voiceless obstruent;
broken line: all other words.
On the Mechanism of Linguistic Change 169

same with similar attitudes towards the island. All had rural
families,
upbringing, and worked as carpenters or fishermen, with one excep-
tion. Thus the continuous development of centralization represents
the very model of a neo-grammarian sound change, accomplished
within two generations.
The embedding problem was approached by correlating the
first

centralization of the obviously related variables (ay) and (aw) that —

is, the change of (aw) was embedded in the system of upgliding

diphthongs. The Atlas records indicate a moderate degree of central-

ization in the 1930's, so that we know that the centralized forms of
(ay) preceded the rise of (aw). The fourth column of Table 7.1 shows
a close correlation of the two variables, with (ay) slightly in the lead
at first, but (aw) becoming more dominant at the end. This pattern
was repeated when the variables were correlated with a number of
independent extralinguistic factors: the occupation, education, and
geographic location of the speaker, and most importantly, the ethnic
group to which he belonged. The significant differences in the tran-
sition rates of these various subgroups allowed the following state-
ment of a solution to the embedding problem:
The centralization of (aw) was more general change which began
part of a
with the centralization of (ay). change proceeded from a moder-
This initial
ate level of (ay) centralization which was probably a regional and recessive
trait inherited from the original settlers of the island. The increase of
centralization of (ay) began in a rural community of Yankee fishermen
descended directly from these original settlers. From there, it spread out-
ward to speakers of the same ethnic group in other occupations and in other
communities. The structurally symmetrical variable (aw) began to show
similar tendencies early in this process. The change was also adopted by
the neighboring Indian group at Gay Head, and a generation later, spread
to the large Portuguese group in the more settled sections of the island. In
these two ethnic groups, centralization of (aw) overtook and surpassed
centralization of (ay).

would lead us to believe that the phonetic environment

Fig. 7.1
of (aw)was a powerful factor in the initiation of the sound change.
Moreover, we can observe that the centralization of (ay) also showed
a strong tendency towards phonetic conditioning in Generation lb,
similar to that displayed for (aw) in Generation lib. 8 However, pho-
8. This phonetic conditioning is more in the nature of a continuum than that for

(aw). Fig. 1.4, p. 19 gives the complete data for a speaker of the same age and back-
ground as Mr. H. H. Jr. of Fig. 7.1.
170 SOCIOLINGUISTIC PATTERNS

netic restriction onwas overridden in the following generation,

(ay)
so that Generation shows a uniform norm for (ay) in all phonetic
II

environments. This development would support the view that pho-

netic conditioning does not play a significant role as an inciting cause
of the centralization of (aw), but acts rather as a conditioning factor
which may be eliminated by further change.
On Martha's Vineyard, the evaluation problem was approached
by analyzing a number of clues to the subjective attitudes towards
island life which appeared in the course of the interviews. Attitudes
towards summer tourists, towards unemployment insurance, towards
work on the mainland, towards other occupational and ethnic
groups, were correlated with data obtained from community leaders
and historical records, and then with the linguistic variables. It
appeared that the rise of (aw) was correlated with the successive
entry into the main stream of island life of groups that had previously
been partially excluded. It was concluded that a social value had
been, more or less arbitrarily, associated with the centralization of
(ay) and (aw): to the extent that an individual felt able to claim and
maintain status as a native Vineyarder, he adopted increasing cen-
tralization of (ay)and (aw). Sons who had tried to earn a living on
the mainland, and afterwards returned to the island, developed an
even higher degree of centralization than their fathers had used. But
to the extent that a Vineyarder abandoned his claim to stay on the
island and earn his living there, he also abandoned centralization
and returned to the standard uncentralized forms.
The solution to the evaluation problem is a statement of the social
significance of the —
changed form that is, the function which is the
direct equivalent on the noncognitive level of the meaning of the
form on the cognitive level. In the developments described here, the
cognitive function of /ay/ and /aw/ has remained constant. It is
plain that the noncognitive functions which are carried by these
phonological elements are the essential factors in the mechanism of
the change. This conclusion can be generalized to many other in-
stances of more complex changes, in which the net result is a radical
change of cognitive function. The sound change observed on
Martha's Vineyard did not produce phonemic change, in which units
defined by cognitive function were merged or split. But many of the
changes in progress that have been observed in New York City did
produce such mergers and splits on the level of the bi-unique
phoneme. One such change is the raising of (oh), the vowel of law,
On the Mechanism of Linguistic Change 171

more, etc., which will serve to illustrate many aspects of

talk, off,
the mechanism of linguistic change not relevant to the simpler
example on Martha's Vineyard.

The Raising of (oh) in New York City

It was not possible to make a direct attack upon the transition

problem in New Although the records of the Linguistic
York City.
Atlas showed sporadic raising of (oh) at a fairly low level, the Atlas
informants in New York City were not selected systematically
enough so that we could construct a comparable sample in 1963. 9
Furthermore, an overall comparison of the usage of this variable by
older and younger speakers did not show the clearcut and regular
progression which we saw for (aw) on Martha's Vineyard. It was
suspected that the reason for this difficulty was the greater tendency
towards stylistic variation among New Yorkers, and the hetero-
geneity of the population in terms of socioeconomic class and ethnic
membership. Therefore it was necessary to attack the embedding
problem first, before the transition problem.
The variable (oh), as we have defined it in Ch. 3, is a part of the
system of long and ingliding vowels in the r-less vernacular of New
York City. Thus we will code the six variants of (oh) in Jaw, lore,
talk, stork, broad, board, all, etc. (p. 76).

Ch. 3 describes the methods used for isolating a range of well-

defined contextual styles in the speech of individual informants.
Average index scores were determined for each style, and a system-
atic sampling of a large urban population undertaken. The embed-
ding problem was then attacked by correlating the five chief linguis-
tic variables each with each other, and with other elements of the

linguistic system, with the level of stylistic variation in which they

were recorded, and with the independent variables of socioeconomic
class (occupation, education, and income) sex, ethnic group, and age
level.
In our studies of hypercorrection in Ch. 5, we presented (p. 128)
a style stratification diagram for (oh) in which the transitional state

Convenience was apparently a greater factor in the selection of Atlas informants

9.

in New York than on Martha's Vineyard. The great bulk of the New York population
was poorly represented in the sample, including the working class and lower middle
class. The old-family stock used for Atlas interviews represents only a very small
fraction of the ethnic composition of the city, at most 1 or 2 percent.
172 SOCIOLINGUISTIC PATTERNS

of this variable is seen in synchronic section. Fig. 5.2 thus indicates

that the change has not yet affected all social classes: (oh) is not a
significant variable for lower-class speakers, who do not use partic-
ularly high values of this vowel and show no stylistic stratification
at all. Working-class speakers show a recent stage in the raising of
(oh): very high vowels in casual speech, but otherwise very little
stratification in the more formal styles, and little tendency towards
the extreme, hypercorrect (oh-4) and (oh-5). But lower-middle-class
speakers show the most developed state of the sound change, with
high values in casual speech, and extreme stylistic stratification.
Finally, the upper-middle-class group is more moderate in all re-
spects than the lower middle class, still retaining the pattern of
stylistic stratification.
The ethnic group membership of New York City speakers is even
more relevant to their use of (oh) than socioeconomic class. Fig. 72
shows the differences between speakers of Jewish and Italian back-
ground in the treatment of (oh) in casual speech. For all but the upper
10
middle class, the Jewish group uses higher levels of (oh). Table 7.2

10. The black group does not show any significant response to the variable (oh),
and shows a constant index of performance at a low level. As noted above, the lower
class in general is similarly indifferent to (oh). Table 7.2 shows Jewish and Italian
ethnic groups only, with the lower class excluded.

15-i

x 20
0)

° 25-

30-

0-2
T3-5 6-8

Fig. 7.2. Class stratification for (oh) by ethnic group in casual

speech. Solid line, Jews; broken line, Italians. SEC: 0-2, lower
class; 3-5, working class; 6-8, lower middle class; 9, upper mid-
dle class.
On the Mechanism of Linguistic Change 173

TABLE 7.2
AVERAGE (oh) INDEXES BY AGE LEVEL
AND ETHNIC GROUP IN CASUAL SPEECH
Age Jews Italians

8-19 17 18
20-35 18 18
36-49 17 20
50-59 15 20
60- 25 30

shows that both Jewish and Italian speakers have participated in the
raising of (oh) but the increase seems to have reached its maximum
early for the Jewish group, and later for the Italian group. A separate
solution for the transition problem is therefore required for each
ethnic group.
The transition problem for the Italian group can be seen analyzed
in Fig. 7.3. The procession of values is not absolutely regular, since

20 30 40 50 60 70

15

1 |

20 -A k^ _

\
I
x 1
0) ^
"O
25 A A A
.5
i ^ N

i^

30

35 { 1

Fig. 7.3. Distribution of (oh) index scores for Italian subjects by

age. A women; ^ men.
174 SOCIOLINGUISTIC PATTERNS

socioeconomic membership, sex, and other factors affect the values;

nevertheless, there is a steady upward movement from the oldest
speakers on the right to the youngest speakers on the left. Within
the present sample of New York City speakers, this is the finest
resolution of the transition problem which can be obtained. 11
The embedding problem for (oh) requires an intricate set of cor-
relations with other elements in the linguistic system, in addition
to the extralinguistic correlations exemplified above. We find that
(oh) is firmly embedded within the subsystem of long and ingliding
vowels, and also related structurally to other vowel subsystems.
Quantitative studies of these relations fall into five sets:

1. There is a strong correlation between the height of (oh) and the

height of the corresponding front ingliding vowel (eh) in the word
class of bad, ask, dance, etc. This variable originated as a raising
of /aeh/, but early in the evolution of New York City speech it
merged with /eh/, the word class of bare, bared, where, etc. The
relation between (eh) and (oh) is strikingly parallel to that of (ay)
and (aw) on Martha's Vineyard. The front vowel was raised first,
as early as the 1890's in New York, and the back vowel followed.
Like (aw) on Martha's Vineyard, the variable (oh) became spe-
cialized in the usage of a particular ethnic group: to the extent
that the Italian group shows higher use of (eh) in casual speech,
the Jewish group shows higher values of (oh), until the difference
is largely resolved in the youngest age level by merger of (eh) and
/ih/, (oh) and /uh/.
2. The variable (oh) also has close relations with the higher ingliding
vowel /uh/. As we observe higher and higher variants of (oh) in
the casual speech of the younger informants, it becomes apparent
that a merger of (oh) and /uh/ is imminent. This merger has
undoubtedly occurred in the youngest speakers in our sample
from the working class and lower middle class. In fact, we have
many informants who show the merger even in the most formal
styles, in the reading of isolated word lists, and we can conclude
a fortiori that the merger exists in casual speech. Close study of
the variants of their casual speech shows the merger as an ac-
complished fact: though most listeners who are not conscious of

11. Fig. 7.3 includes Italian informants who refused the original interview, and
whose speech patterns were sampled by the television interview, as described in
Appendix D of Labov 1966a.
On the Mechanism of Linguistic Change 175

the overlap will hear sure as higher than shore, it is in fact in-
distinguishable out of context.
3. There is also a close correlation between (oh) and /ah/, the long
tense vowel heard in guard, father, car, etc. The variable (ah)
represents the choice of back or center options for the subclasses
of hot, heart, hod, and hard. High values of (oh) are correlated
with low back positions of heart, hod, and hard (with the last
two generally homonymous); lower values of (oh) are correlated
with low center positions of the vowels in these word classes. This
correlation is independent of socioeconomic class or ethnic group.
Whereas (oh) is firmly embedded in the sociolinguistic structure
of the speech community, /ah/ is not. As a linguistic variable,
(ah) seems to be a function only of the height of (oh): a purely
internal variable. 12
4. (oh) is also related to the variable height of the vowel in hoy, coil,
etc., (oy) in the front upgliding system. The height of the vowels
in coil and call seem
vary directly together in casual speech,
to
but only (oh) is corrected to lower values in more formal styles,
(oh) carries the major burden of social significance, and is the
focus of nonsystematic pressure from above.
5. Finally, we find that (oh) and (oy) are jointly correlated with the
variable (ay), which represents the backing or fronting of the first
element of the diphthong in my, why, side, etc. High values of
(oh) and (oy) are correlated with back values of (ay), and low
values of (oh) and (oy) with low center values of (ay).

Beyond these immediate correlations, there are more indirect,

diffuse relations with such variables as (aw) and /ih/, through which
(oh) is connected with all of the other vowels in the vernacular
system of New York City speech. This is not the place to pursue the
full details of this intricate set of structural correlations within the
linguistic system: however, it should be apparent that a full solution
to the embedding problem will reveal the ways in which the internal
relations of linguistic elements determine the direction of sound
change. We can summarize the most important relations that center
about (oh) in the following notation, which defines the structural
unitson the left hand side of the equations as linguistic variables:

12. The quantitative correlations are given in Labov 1966a: Ch. 12. The relationship
of (oh) and (ah) held even within a single ethnic group. For a spectrographs display
of all these correlations, see Labov, Yaeger, and Steiner 1972.
176 SOCIOLINGUISTIC PATTERNS

(oh) = f x (St, C, E, A, Sx, (eh)) St = style

(ah) = f 2 ((oh)) C = socio-economic class
(oy) = f3 ((oh)) E = ethnic group
(ay) = f4 ((ah)) = f4 (f 2 ((oh))) A= age level
(ay) = f 5 ((oy)) = f5 (f3 ((oh))) Sx = sex

In NewYork City, the evaluation problem was approached more

directly than on Martha's Vineyard. The unconscious subjective
reactions of the informants to each of the variables were determined.
The details of this method have been presented in Ch. 6; in general,
we can say that the reliability of the tests can be measured by the
high degree of uniformity showed by New Yorkers in contrast to the
scattered results from those raised outside of New York City.
The subjective reaction responses to (oh) give us a clear view of
the social significance of the variable, as shown in Table 7.3. The
majority of informants responded to the test in a way consistent with
the stigmatized status of high (oh). 13 Just as the solution to the
embedding problem showed no significant stylistic response to (oh)
for lower-class speakers, here we find that lower-class speakers
showed no significant (oh)-negative response. The other groups
showed (oh)-negative response in proportion to the average height
of (oh) used in their own casual speech, and to the degree of stylistic
stratification in their speech patterns. This result illustrates the
principle which holds quite generally in New York City: that those
who used the highest percentage of a stigmatized form in casual
speech were the most sensitive in stigmatizing it in the speech of
others. Thus the lower-middle-class speakers between the ages of

TABLE 7.3
(oh)-NEGATIVE RESPONSE BY
SOCIOECONOMIC CLASS AND AGE LEVEL
Socioeconomic : class

Age 0-2 3-4 5-8 9

20-39 25% 80% 100% 60%

4 10 11 5
40-59 18 60 62 57 11 15 13 7
60- 33 [00] 6 1 — —

13. The (oh)-negative response shown here consisted of rating three speakers lower
on a scale of job suitability when they pronounced sentences with high, close (oh)
vowels, as compared to sentences with no significant variables. Those making the
ratings were unaware that they were rating the same speakers.
On the Mechanism of Linguistic Change 177

20 and 39, use the highest values of (oh) in their own casual
who
speech, (oh)-negative response. Similarly, we find
show 100 percent
that the percentage of (oh)-negative response among Jewish and
Italian speakers is proportionate to the height of (oh) in casual
speech.
This solution to the evaluation problem can hardly be called
satisfactory. It is not clear why a group of speakers should adopt

more and more extreme forms of a speech sound which they them-
selves stigmatize as bad speech. 14 Some further explanation must
be given.
First of all, it has become clear that very few speakers realize that
they use the stigmatized forms themselves. They hear themselves as
using the prestige forms which occur sporadically in their careful
speech and in their reading of isolated word lists. Secondly, the
subjective responses tapped by our test are only the overt values
those which conform to the value systems of the dominant middle-
class group. There are surely other values, at a deeper level of
consciousness, which reinforce the vernacular speech forms of New
York City. Our early evidence for this conclusion was anecdotal. But
more recent studies of the black English vernacular in New York
City have demonstrated the existence of such covert norms through
alternate scales on subjective reaction tests (see Ch. 8).
In the case of the alternate preference of Jewish and Italian ethnic
groups for (oh) and (eh), we can put forward a reasonable suggestion
based upon the mechanism of hypercorrection. 15 The influence of
the Yiddish substratum leads to a loss of the distinction between
low back rounded and unrounded vowels in first-generation Jewish
speakers of English, so that cup and coffee have the same vowel.
In second-generation speakers of Jewish descent, the reaction against
this tendency leads to a hypercorrect exaggeration of the distinction,
so that (oh) becomes raised, tense and over-rounded. A parallel
argument applies to Italian speakers. This suggestion is all the more
plausible since hypercorrection has been demonstrated to be an
important mechanism of linguistic change in a variety of circum-
stances. 16

14. Many subjects reacted to the test with violent and unrealistic ratings; as, for

example, marking a person who used high vowels for coffee and chocolate as not
even speaking well enough to hold a factory job.
15. I am indebted to Marvin Herzog for this suggestion.

16. As noted in Ch. 5, hypercorrection is used here not to indicate the sporadic

and irregular treatment of a word class, but the movement of an entire word class
178 SOCIOLINGUISTIC PATTERNS

The Mechanism of Sound Change

Solutions to the transition, embedding, and evaluation problems

have been illustrated by two examples, drawn from Martha's Vine-
yard and New York City. It is possible to apply the results of our
work with these and other variables to a provisional answer to the
question: what is the mechanism by which sound change proceeds?
The following outline is based upon analysis of twelve sound
changes: three on rural Martha's Vineyard, and nine in urban New
York City. 17

1. The sound changes usually originated with a restricted subgroup

of the speech community, at a time when the separate identity
of this group had been weakened by internal or external pres-
sures. The linguistic form which began to shift was often a
marker of regional status with an irregular distribution within
the community. At this stage, the form is an undefined linguistic
variable.
2. The changes began as generalizations of the linguistic form to
all members of the subgroup; we may refer to this stage as change

from below, that is, below the level of social awareness. The
variable shows no pattern of stylistic variation in the speech of
those who use it, affecting all items in a given word class. The
linguistic variable is an indicator, defined as a function of group

membership.
3. Succeeding generations of speakers within the same subgroup,
responding to the same social pressures, carried the linguistic
variable further along the process of change, beyond the model
set by their parents. We may refer to this stage as hypercorrection
from below. The variable is now defined as a function of group
membership and age level.
4. To the extent that the values of the original subgroup were
adopted by other groups in the speech community, the sound
change w th its associated value of group membership spread
4

to these adopting groups. The function of group membership is

now redefined in successive stages.

beyond the target point set by the prestige model. This mechanism is evident on
Martha's Vineyard, as well as New York.
17. The stages suggested here are necessarily ordered in approximately the manner

listed, but there are some rearrangements and permutations in the data observed.
On the Mechanism of Linguistic Change 179

5. The limits of the spread of the sound change were the limits of
the speech community, defined as a group with a common set
of normative values in regard to language.
6. As the sound change with its associated values reached the limits
of expansion, the linguistic variable became one of the norms
its

which defined the speech community, and all members of the

speech community reacted in a uniform manner to its use (with-
out necessarily being aware of it). The variable is now a marker,
and begins to show stylistic variation.
7. The movement of the linguistic variable within the linguistic
system always led to readjustments in the distribution of other
elements within phonological space.
8. The structural readjustments led to further sound changes which
were associated with the original change. However, other sub-
groups which entered the speech community in the interim
adopted the older sound change as a part of the community
norms, and treated the newer sound change as stage 1. This
recycling stage appears to be the primary source for the continual
origination of new changes. In the following development, the
second sound change may be carried by the new group beyond
the level of the first change.
[Stages 1-8 dealt with change from below; stages 9-13 con-
cern change from above.]
9. If the group in whichchange originated was not the
the
highest-status group in the speechcommunity, members of the
highest-status group eventually stigmatized the changed form
through their control of various institutions of the communi-
cation network.
10. This stigmatization initiated change from above, a sporadic and
irregular correction of the changed forms towards the model of
—
the highest status group that is, the prestige model. This prestige
model is now the pattern which speakers hear themselves using:
it governs the audio-monitoring of the speech signal. The linguis-

tic variable now shows regular stylistic stratification as well as

social stratification, as the motor-controlled model of casual

speech competes with the audio-monitored model of more care-
ful styles.
11. If model of the highest-status group does not corre-
the prestige
spond form used by the other groups in some word class,
to a
the other groups will show a second type of hypercorrection:
180 SOCIOLINGUISTIC PATTERNS

shifting their careful to a form further from the changed

speech
form than the by the prestige group. We may call this
target set
stage hypercorrection from above.
12. Under extreme stigmatization, a form may become the overt
topic of social comment, and may eventually disappear. It is thus
a stereotype, which may become increasingly divorced from the
forms which are actually used in speech.
13. If the change originated in the highest-status group of the com-

munity, it became a prestige model for all members of the speech

community. The changed form was then adopted in more careful
forms of speech by all other groups in proportion to their contact
with users of the prestige model, and to a lesser extent, in casual
speech. 18

Many of the stages in the mechanism of sound change outlined

here are exemplified in the two detailed examples given above. The
centralization of (aw) on Martha's Vineyard appears to be a stage 4
change from below. It may indeed have reached stages 5 and 6,
but the techniques used on Martha's Vineyard did not provide the
evidence to decide this question. There is no doubt, however, that
the centralization of (aw) is a secondary change, produced by the
recycling process when the centralization of (ay) reached stage 8.

To place the raising of necessary to con-

(oh) in this outline, it is

sider briefly the evolution of the New York City vowel system as
a whole. The first step in the historical record is the raising of (eh).
We have reason to believe that the merger of /aeh/ with /eh/ began
in the last quarter of the 19th century (Babbitt 1896). The upward
movement of the linguistic variable (eh) continued beyond this
merger, leading to the current cumulative merger of /eh/ with /ih/
among most younger New Yorkers. For the entire community, (eh)
is subject to the full force of correction from above: the change has
reached stage 11 so that the linguistic variable is defined by covari-
ation with social class, ethnic membership, age level, and contextual
style. The was the first recycling process which began
raising of (oh)
when (eh) reached stage 8. The major burden of the raising of (oh)
has been carried by the Jewish ethnic group; the extreme upward
social mobility of this group has led to a special sensitivity to (oh)

18. We find some support in these observations for the idea that people do not

borrow much from broadcast media or from other remote sources, but rather from
those who are at the most one or two removes from them in age or social distance.
On the Mechanism of Linguistic Change 181

in the lower middle class. Thus the merger of /oh/ and /uh/ has
gone quite quickly, and (oh) has reached stage 11 for the lower
middle class; yet it has hardly touched stage 1 for the lower class.
The third stage in the recycling process occurred when (oh)
reached stage 8. The structural readjustments which took place were
complex: (oy) and (ah) were closely associated with (oh), and were
defined as linguistic variables only by their covariation with (oh).
Thus the raising of (oy) and the backing of (ah) were determined
by internal, structural factors. Change from above is exerted upon
(oh), but not upon (oy). In careful speech, a New Yorker might say
[its u:l tin fu:-i], It's all tin foil. But the shift of (ah) and (oy) have

in turn led to a shift of (ay), and this process has apparently begun
a third recycling. Indeed, the backing of (ay) has reached stage 8
itself, and produced an associated fourth recycling, the fronting of

(aw). There are indications that (ay) has evolved to stage 9, with the
beginning of overt correction from above, although (aw) has reached
only stage 4 or 5 (Labov 1966a:Ch. 12).
It is evident that the type of structural readjustments that has been

considered here requires a linguistic theory which preserves the

geometry of phonological space. The structural relations found here
are strikingly parallel to those established by Moulton (1962) in his
study of covariation of mid and low vowels in Swiss German dia-
lects. The techniques, the area, the societies studied are quite differ-
ent, and the coincidence of results provides strong empirical evi-
dence for the functional view of phonological structure advanced
by Martinet (1955). Nevertheless, the purely internal equilibria pro-
jected by Martinet do not provide a coherent theory of the mecha-
nism of sound change. In the scheme that has been outlined here,
they are only part of a more comprehensive process, embedded in
the sociolinguistic structure of the community.

Conclusion

This discussion has focused on the theme that internal, structural

pressures and sociolinguistic pressures act in systematic alternation
in the mechanism of linguistic change. It can no longer be seriously
argued that the linguist must limit his explanations of change to the
mutual influences of linguistic elements defined by cognitive func-
tion. Nor can it be argued that a changing linguistic system is au-
tonomous in any serious sense. Here I have attempted to carry the
182 SOCIOLINGUISTIC PATTERNS

argument beyond the mere cataloguing of possibilities by introducing

a large body of evidence on sound changes observed in progress.
On the basis of this evidence, we can make the stronger claim that
it is not possible to complete an analysis of structural relations within

a linguistic system, and then turn to external relations. The recycling

process outlined here suggests the kind of answer we can make to
the basic questions of the inciting causes of linguistic change, and
the adaptive functions of change, as well as the mechanism by which
change proceeds. We can expect that further investigations will
modify the outline given here, but that data from the speech com-
munity will continue to form an essential part in the analysis of
linguistic change.
 The Study of
8 Language in Its
Social Context

THE first seven chapters of this volume have documented an

approach to linguistic research which focuses upon language in use
within the speech community, aiming at a linguistic theory adequate
to account for these data. This type of research has sometimes been
labelled as "sociolinguistics," although it is a somewhat misleading
use of an oddly redundant term. Language is a form of social behav-
ior: statements to this effect can be found in any introductory text.

Children raised in isolation do not use language; it is used by human

beings in a social context, communicating their needs, ideas, and
emotions to one another. The egocentric monologues of children
appear to be secondary developments derived from the social use
of language (Vygotsky 1962:19) and very few people spend much time
talking to themselves. It is questionable whether sentences that
communicate nothing to anyone are a part of language. In what way,
then, can "sociolinguistics" be considered as something apart from
"linguistics"?
One
area of research which has been included in "sociolinguistics"
is perhaps more accurately labelled "the sociology of language." It
deals with large-scale social factors, and their mutual interaction
with languages and dialects. There are many open questions, and
many practical problems associated with the decay and assimilation
of minority languages, the development of stable bilingualism, the
standardization of languages and the planning of language develop-
ment in newly emerging nations. The linguistic input for such studies
is primarily that a given person or group uses language X in a social

context or domain Y. A number of recent reviews have discussed

work in this area (Fishman 1969) and I will not attempt to deal with
these questions and this research here.
183
184 SOCIOLINGUISTIC PATTERNS

There isanother area of study sometimes included in "sociolin-

guistics" that is more concerned with the details of language in actual

—
use the field which Hymes has named "the ethnography of speak-
ing" (1962). There is a great deal to be done in describing and ana-
lyzing the patterns of use of languages and dialects within a specific
culture: the forms of "speech events"; the rules for appropriate
selection of speakers; the interrelations of speaker, addressee, audi-
ence, topic, channel, and setting; and the ways in which the speakers
draw upon the resources of their language to perform certain func-
tions. This functional study is conceived as complementary with the
study of linguistic structure. Current research and the aims of the
field have been well reviewed by Hymes (1966); in our discussion
of methodology, some of the material of this descriptive study will
be involved, but this review will not attempt to cover the ethnogra-
phy of speaking as a whole. A number of readers and reviews of
this larger field of "sociolinguistics" have appeared recently; and the
reader will find a number of excellent and penetrating studies in
Bright 1966; Gumperz and Hymes 1964; Lieberson 1966; Fishman
1968; Ervin-Tripp 1968; and Grimshaw 1968.
This chapter will deal with the study of language structure and
evolution within the social context of the speech community. The
linguistic topics to be considered here cover the area usually named
"general linguistics," dealing with phonology, morphology, syntax,
and semantics. 1 The theoretical questions be raised will also fall
to
into the category of general linguistics. We concerned with
will be
the forms of linguistic rules, their combination into systems, the
coexistence of several systems, and the evolution of these rules and
systems with time. If there were no need to contrast this work with
the study of language out of its social context, I would prefer to say
that this was simply linguistics. It is therefore relevant to ask why
there should be any need for a new approach to linguistics with a
broader social base. It seems natural enough that the basic data for
any form of general linguistics would be language as it is used by
native speakers communicating with each other in everyday life.
Before proceeding, it will be helpful to see just why this has not
been the case.

1. We have also extended these studies into the area of discourse analysis, which

has not been considered a part of general linguistics or seriously investigated in the
past. Sect. 4 of this chapter gives a brief indication of the nature of this work.
The Study of Language in its Social Context 185

The Saussurian Approach to "Langue"

The basic orientation to the structural analysis of language as most

linguists pursue it today departs from the point of view first ex-
pressed by Ferdinand de Saussure at the beginning of this century.
Linguists often begin theoretical discussions with reference to Saus-
sure's concept of iangue, to be distinguished from parole or 'speech'
on the one hand, and langage or 'language as a whole' on the other.
According to Saussure, langue "est la partie sociale du langage . . .

elle n'existe qu'en vertu d'une sorte de contrat passe entre les
membres de la communaute" (1962:31). For this reason, Saussure's
Geneva school is often referred to as the "social" school of linguis-
tics.Saussure conceived of linguistics as one part of "une science
qui etudie la vie des signes au sein de la vie sociale." Yet curiously
enough, the linguists who work within the Saussurian tradition (and
this includes the great majority) do not deal with social life at all:
they work with one or two informants in their offices, or examine
their own knowledge of langue. Furthermore, they insist that ex-
planations of linguistic facts be drawn from other linguistic facts,
not from any "external" data on social behavior. 2
This development depends on a curious paradox. If everyone
possesses a knowledge of language structure, if iangue is "un systeme
grammatical existant virtuellement dans chaque cerveau" (p. 30), one
should be able to obtain the data from the testimony of any one
—
person even oneself. 3 On the other hand, data on parole, or speech,
2. Saussure's contemporary Meillet thought that the 20th century would see the

development of historical explanation based on the examination of language change

embedded in social change (1921). But students of Saussure such as Martinet (1964b]
actively repudiated this notion, and urged that linguistic explanation be confined to
the interrelations of internal, structural factors. In so doing, they were certainly
following the spirit of Saussure's doctrine: for closer study of his writings suggests
that for him, "social" meant no more than "multi-individual," with no suggestion of
the broader implications of social interaction.
3.Saussure himself was a bit hesitant on this point. For after the quotation just
given, he adds "ou plus exactement dans les cerveaux d'un ensemble d'individus: car
la langue n'est complete dans aucun, elle n'existe parfaitement que dans la masse".
But virtueliement became equivalent to for all practical purposes. Saussure himself
did not engage in any detailed synchronic studies, but linguists who did so discarded
his reservation completely.Thus Bloomfield presents a structural analysis of "standard
English, as spoken Chicago" without further identification (1933:90-92); we assume
in
that he is speaking of his own system, though he does not reach a level of detail
where this would become an issue. Benjamin L, Whorf wrote a paper on the "Phonemic
analysis of the English of Eastern Massachusetts" (1943) which was again a report
based on his own way of speaking.
186 SOCIOLINGUISTIC PATTERNS

can only be obtained by examining the behavior of individuals as

they use the language. Thus we have the Saussurian Paradox: the
social aspect of language is studied by observing any one individual,
but the individual aspect only by observing language in its social
context. The science of parole never developed, but this approach
to the science of iangue has been extremely successful over the past
half-century.
of this abstract "language" the knowledge available
The study —
to —
every native speaker has received new impetus from Chomsky,
who has re-emphasized the Saussurian dichotomy, opposing com-
petence, or the abstract knowledge of the rules of language, to
performance, or the selection and execution of these rules (1965). 4
For Chomsky, linguistics is properly the study of competence, and
he makes explicit the practice which followed from the Saussurian
paradox: that the proper object of linguistic study is an abstract,
homogeneous speech community in which everyone speaks alike and
learns the language instantly (1965:3). Furthermore, Chomsky insists
that the data of linguistics is not the utterance by the individual to
be studied, but his intuitions about language primarily his judg- —
ments as to which sentences are grammatical and which are not and —
also judgments on the relatedness of sentences which sentences —
mean "the same." Theories of language are to be constructed to
explain these intuitions.
This theoretical development is based upon two more or less
explicit assumptions:

1. That linguistic structure is closely associated with homogeneity.

(Weinreich, Labov, and Herzog 1968.) Saussure says that "tandis
que le langage est heterogene, la Iangue ainsi delimitee est de
nature homogene" (1962:32). 5 The general view then, is that lin-

4. Although Chomsky criticized Saussure's conception of Iangue as somewhat

between Saussure's Jangue/paroie dichotomy
limited (1964:59-60), he sees no difference
and his own competence/performance terminology. "The generative grammar inter-
nalized by someone who has acquired a language defines what in Saussurian terms
we may call iangue Clearly the description of intrinsic competence provided by
. . .

the grammar is not to be confused with an account of actual performance, as de

Saussure emphasized with such lucidity" (1964:52).
5. In an introductory textbook by John Lyons, representing a viewpoint somewhat

independent of generative grammar, we find: "When we say that two people speak
the same language we are of necessity abstracting from all sorts of differences in their
speech . . . For simplicity of exposition, we shallassume that the language we are
describing is uniform (by 'uniform' is meant 'dialectally and stylistically' undifferenti-
The Study of Language in its Social Context 187

guistic theoriescan be fully developed on the basis of that portion

of language behavior which is uniform and homogeneous; though
language variation may be important from a practical or applied
viewpoint, such data are not required for linguistic theory and —
in fact will be best understood when the theory of competence
is fully developed.

2. Speakers of the language have access to their intuitions about

langue or competence, and can report them.

Linguistics has thus been defined in such a way as to exclude the

study of social behavior or the study of speech. The definition has
been convenient for the formulators, who by disposition preferred
to work from their own knowledge, with individual informants, or
with secondary materials. But it has also been a successful strategy
in our attack on linguistic structure. There is no a priori reason why
one must enter the speech community to search for data. The large
expenditure of time and effort needed would have to be justified,
and the success of abstract linguistic analysis in the past five decades
has plainly precluded such a development. Indeed, the limiting of
our field of inquiry has certainly been helpful in the development
—
of generative grammar the working out of abstract models based
upon our intuitive judgments of sentences. We cannot afford any
backward steps: anyone who would go further in the study of lan-
guage must certainly be able to work at this level of abstraction. At
the same time, it is difficult to avoid the common-sense conclusion
that the object of linguistics must ultimately be the instrument of
communication used by the speech community; and if we are not
talking about that language, there is something trivial in our pro-
ceeding. For a number of reasons, this kind of language has been
the most difficult object for linguistics to focus on. Some of the
reasons for this difficulty will be outlined below.

Problems in Dealing with Speech

Despite the general orientation of the field towards the study of

language in isolation, there have been many situations where lin-
guists have hoped to obtain confirmation from the study of speech.

ated: this of course, an 'idealization' of the facts

is, . and that all native speakers
. .
)

will agreewhether an utterance is acceptable or not." (1968:140-141). It should be noted

that Lyons' textbook is an introduction to "Theoretical Linguistics" and this ideal-
ization does not represent a response to any practical problems.
188 SOCIOLINGUISTIC PATTERNS

There are four distinct difficulties that have been cited, and which
have had profound effects upon linguistic practice.

1. The Ungrammatically of Speech

At one time, linguists of the Bloomfieldian school asserted that
native speakers never made mistakes. But the opposite point of view
prevails today: that speech of ungrammatical forms, since the
is full

difficulties of performance stand in the way of the full display of

the speaker's competence. 6 It is generally believed that a corpus
drawn from spoken language does not form good evidence, since
it will contain many examples of badly formed sentences which the

speakers themselves condemn and change when their attention is

drawn to them.

2. Variation in Speech and in the Speech Community

It is common for have many alternate ways of saying
a language to
"the same" thing. Some words like car and automobile seem to have
the same referents; others have two pronunciations, like working and
workiri. There are syntactic options such as Who is he talking to?
vs. To whom is he talking? or It's easy for him to talk vs. For him
to talk is easy. 7 In each of these cases, we have the problem of
deciding the place of this variation in linguistic structure. Current
formal analysis provides us with only two clear options: (1) the
variants are said to belong to different systems, and the alternation
is an example of "dialect mixture" or "code-switching"; (2) the
variants are said to be in "free variation" within the same system,
and the selection lies below the level of linguistic structure. Both
approaches place the variation outside of the system being studied.
There are of course many cases which fall appropriately under one
or the other of these labels. But to demonstrate that we have a true
case of code-switching, it is necessary to show that the speaker moves

6. Chomsky has asserted that the "degenerate" character of the everyday speech
which the child hears is a strong argument in support of the nativist position: the
child must have an inborn theory of language, since he could not induce rules from
the ungrammatical speech with which he is surrounded (1965:58).
7. It is customary to say that these expressions have the same meaning, which we

define narrowly by some criterion such as 'having the same truth value'. The end
result of our studies of syntactic variation will be to assign a certain meaning or
significance to a transformation, a type of functional load which we may want to
distinguish sharply from representational meaning.
The Study of Language in its Social Context 189

from one consistent set of co-occurring rules to another; to demon-

strate "free variation" one has to show that he has not moved at all.
It is rare for either of these claims to be established empirically. Most
cases are not easily described under either heading; consider for
example an actual example of language in use. 8

An' den like IF YOU MISS ONESIES, de OTHuh person shoot to skelly;
ef he miss, den you go again. An' IF YOU GET IN, YOU SHOOT TO
TWOSIES. An' IF YOU GET IN TWOSIES, YOU GO TO tthreesies. An'
IF YOU MISS tthreesies, THEN THE PERSON THa' miss skelly shoot THE
SKELLIES an' shoot in THE ONESIES: an' IF HE MISS, YOU GO f'om
tthreesies to foursies.

In this extract, a 12-year-old black boy is explaining the game of

Skelly. We
can treat his variations as examples of code-switching;
each time he uses a marked variant, he moves into the "system"
containing that variant. Lower case would then indicate the "Black
English Vernacular" [BEV] and upper case "Standard English" [SE].
But it is an unconvincing effort: there is no obvious motivation for
him to switch 18 times in the course of this short passage. Further-
more, the great majority of the forms are shared by both systems
and are assigned to one or the other code by the accidents of se-
quencing. In line 2, you go again is assigned to BEV only because
it happens to follow den, and YOU GET IN is assigned to SE only

because it follows the marked form IF. But on the other hand, can
we treat the difference between de and THE as "free variation"? Such
a decision would make no sense to either the speaker or the analyst,
who both know that de is a stigmatized form. Without any clear way
of categorizing this behavior, we are forced to speak of "stylistic
variants," and we are then left with no fixed relation at all to our
notion of linguistic structure. What is a style if not a separate sub-
code, and when do we have two of them? We normally think of
language as a means of translating meaning into linear form. Where
and how do stylistic meanings enter into this process? We speak of
the need for communicating meaning as a controlling factor in
linguistic evolution. What kind if any is exerted by the
of control
need communicate "stylistic" messages?
to
An even more puzzling problem arises when we consider a varia-

8. This quotation is from an interview with "Boot," the verbal leader of a pre-
adolescent group of black children in south-central Harlem, New York City (Labov
et al. 1968).
190 SOCIOLINGUISTIC PATTERNS

ble phenomenon such as consonant cluster simplification in BEV —

process which lies on the intersection of grammar and phonology.
A word such as bold is often simplified to boV, but not always. This
is We immediately want to know if past
also the case with rolled.
tense clusters CVC + D
can be treated in the same way as simple
CVCC forms, without danger of losing the past tense information.
Careful investigation shows us that the distinction is never lost the —
past tense forms are simplified less often by everyone. But our theory
has no way of registering this fact formally: both bold and rolled
fall under the same "optional" rule, and our observations have no
theoretical status in the rules of langue.

3. Difficulties of Hearing and Recording

Records of speech observed in actual use are often very poor in
quality. Acoustic phoneticians gather their data in soundproof rooms,
under the best possible conditions. In the field, we find that room
noise, street noise, and other interference reduces the phonetic value
of our data. If the informant is brought to record under ideal condi-
tions, then his speech has the properties of formal, elicited speech
we tried to avoid. 9 The fundamental problem is that most linguistic
signals are supported by a great many redundant signals, and it is
rare that any one of them carries a heavy burden of meaning; it is
not essential to the overall message that listeners receive any one
signal. Yet to record this item in full form, the linguist would like
to hear it at its clearest, as if it were the only means of signaling
that message. It would therefore follow that the elicited forms given
in the laboratory give the clearest indication of the underlying
system.

4. The Rarity of Syntactic Forms

The data based on what speakers actually say may be adequate
for the most common phonological and syntactic forms. For any deep
analysis of the sound pattern of a language, it will be necessary to

elicitsuch rare words as adz (the only English morpheme ending

in a cluster of voiced obstruents). In the study of syntax, the inade-
quacy of the average corpus is even plainer. Any attempt to specify

9. This situation is not so damaging for phonological analysis as for grammatical

research. In phonology, we can wait for the clear, stressed forms to emerge from the
background noise. But many grammatical particles are reduced to minimal consonants
or even features of tenseness or voicing which are difficult to hear in less than the
best conditions, and many are so rare that we cannot afford to let one escape us.
The Study of Language in its Social Context 191

syntactic rules inevitably involves forms which one could not expect
to hear in any limited investigation. For example, an analysis of the
got passive may depend upon whether it is possible to say such
sentences as "He got kicked out of the army by playing the trumpet",
where we are looking for such rare forms as X got Verb + ed . . .

by Verb + ing — Z.
<j>

These difficulties make clear the basic motivation for the concen-
tration on langue or "competence" to the exclusion of other data.
Given the fact that considerable progress has been made in the
abstract study of langue, and given these difficulties of work in a
natural setting, it should not be surprising that linguistics has turned
firmly away from the speech community. But there are also disad-
vantages to the abstract study of language. Some of its limitations
have recently become painfully prominent; the difficulties of devel-
oping linguistic theory with this limited data base are perhaps even
greater than those outlined above for the study of the speech com-
munity.

Problems in the Study of Intuitions

When Chomsky first made the explicit proposal that the subject
matter of linguistics be confined to the intuitive judgments of native
speakers, he hoped that the great majority of these would be clear
judgments (1957:14). It was expected that the marginal cases, which
were doubtful in the mind of the theorist and/or the native speaker,
would be few in number and their grammatical status would be
decided by rules formed from the clear cases. The situation has not
worked out in this way, for it is difficult to find doubtful cases which
have not remained problematical for the theory. It is not the number
of doubtful cases which is at issue here: it is their locations at points
which are crucial in arguing questions of grammatical theory. One
can see examples of this problem at any linguistic meeting, where
paper after paper will cite crucial data as acceptable or unacceptable
without obtaining agreement from the audience. This is not due to
carelessness or lack of linguistic ability on the part of the authors:
on the basis of
in their earnest intent to explore linguistic theory
where their data take
their intuitions, they inevitably reach a point
this form. The two assumptions of the homogeneity and accessibility
of Jangue which led to this situation are seriously brought into
question by this development.
When challenges to data arise on the floor of a linguistic meeting,
192 SOCIOLINGUISTIC PATTERNS

the author usually defends himself by stating that there are many
"dialects" and that the systematic argument he was presenting held
good for his own "dialect." This is an odd use of the term, and it
raises the question as to what the object of linguistic description can
or should be.

The Object of Linguistic Description: "Dialect" and "Idiolect"

The use of the term "dialect" in the discussions of the variability
of judgments is No evidence is given of differences
difficult to justify.
in two systematic used by two groups of speakers; what
sets of rules
we observe are individual differences of opinion on isolated points.
As we will see, individuals are not at all consistent from one judg-
ment to the next. The question arises, what is being described? In
the search for a homogeneous object to conform to the needs and
assumptions of the Saussurian model, linguists have gradually con-
tracted their focus to smaller and smaller segments of language. Thus
Bloch introduced the term "idiolect" to represent the speech of one
person talking on one subject to the same person for a short period
of time (1948). Although this term has been widely adopted, it is
doubtful if anyone has found within such an "idiolect" the homo-
geneous data which Bloch hoped for. But it must be noted that the
very existence of the concept "idiolect" as a proper object of lin-
guistic description represents a defeat of the Saussurian notion of
langue as an object of uniform social understanding.
It was hoped that, by concentrating upon the judgments of the

native speaker rather than his actual speech, much of this variation
could be bypassed. In some ways, this hope is justified: members
of a speech community do share a common set of normative patterns
even when we find highly stratified variation in actual speech (Labov
1966a:4-35ff.) But such uniformity in intuitive judgments is charac-
teristic only of well-developed sociolinguistic variables which have
received overt social correction. Most linguistic rules are well below
the level of social correction, and have no overt social norms associ-
ated with them.
In an earlier version of this chapter I reported difficulties in repro-
ducing the syntactic dialects which Postal (1971) reported for cross-
over phenomena and pronominalization constraints. More recently
we have begun to explore systematically the internal consistency
of "idiolectal" syntactic dialects: that is, syntactic dialects reported
The Study of Language in its Social Context 193

on the basis of intuitive responses to isolated sentences randomly

distributed among subjects without any geographic or social differ-
entiation.
One such area of syntactic variation concerns the quantifier dia-
lects isolated by Carden (1970). Carden has been one of the few
careful investigators of syntactic dialects, emphasizing the impor-
tance of individual interviews over group questionnaires, controlled
technique, and checks for reliability over a period of time. His basic
datum is differential response to the sentences

1 All the boys didn't leave.

2 All the boys didn't leave until six.
3 All the boys didn't leave, did they?

The interpretation of 1 as 'Not all the boys left' is said to be charac-

teristic of the NEG-Q dialect, and is consistent with finding 3 ac-
ceptable and 2 unacceptable since the negative is then the highest
predicate, dictating a positive tag in 3 but stranding until in a lower
level positive sentence in 2. The interpretation of 1 as 'None of the
boys left' is said to be characteristic of the NEG-V dialect and is

consistent with finding 2 acceptable and 3 unacceptable, since the

negative is then a lower predicate, dominated by a higher predicate
with the qualifier all; this permits until in 2 but demands a negative
tag in 3.

My own interest in the matter has proceeded from the generaliza-

tion of the study of negative attraction to the quantifier any (see Sect.
2 below). Together with Mark Baltin, I have developed a number
of instruments to approach such intuitions on quantifier dialects and
contrast these intuitions with the unreflecting use of the linguistic
rules involved. Ten different studies have been carried out with
samples of 15 to 40 subjects, using various techniques to sharpen
intuitive judgments. Subjects were drawn from middle-class and
working-class backgrounds, largely favoring the former, in tests
carried out in Philadelphia, New York, Providence, and Kansas City.
In general, we find that three-quarters or more of any sample will
give initial NEG-Q responses to 1, with a small percentage of NEG-V
and some equally balanced ambiguous speakers. For this isolated
sentence, we find that only about one-quarter of the subjects are
consistently and determinately NEG-Q in responses to 1-3 and refuse
to admit NEG-V and we rarely find such consistent
possibilities;
NEG-V responses. However, we do not find consistent responses to
194 SOCIOLINGUISTIC PATTERNS

the NEG-Q or NEG-V pattern. In one study of 17 informants, we

tested the most consistent kind of response: a preference for sentence
A over sentence B, e.g., 2 over 3. The first test pair opposed 3 and
2. The second pair was
4 a. All the guys won't start work until the whistle blows,
b. All the guys haven't started work, have they?

We also used two pairs showing semantic disambiguation:

5 a. Since the plant's locked, all of them haven't started

work yet.
b. All the guys haven't started work yet; some are still

on their lunch hour.

6 a. All the guys don't like John; some of them can't stand
him.
b. All the guys don't like John; none of them has anything
good to say about him.

If NEG-Q dialect, he will prefer 3 to 2, 4b

the speaker indeed has a
to 4a, 5b5a and 6a to 6b. Sentence 5a gives the context which
to
most strongly favors the NEG-V interpretation.
If the NEG-Q and NEG-V dialects were characteristic of the gram-

mars of our 17 subjects, we should have obtained a number greater

than chance who gave consistent answers to these four pairs. But
only 8 out of 17 gave consistent responses to the two pairs showing
syntactic disambiguation; only 9 out of 17 gave consistent responses
to the two pairs showing semantic disambiguation; and only 3 out
of 17 gave consistent responses to both. Furthermore, only 1 of these
3 gave the same consistent response to both sets — that is, was con-
sistently NEG-Q or NEG-V throughout.
This result is consistent with other data which show that almost
all subjects will in fact respond either NEG-Q or NEG-V when we
control the context effectively. That is, all are ambiguous. We con-
structed a test which looks "through" the grammar to extralinguistic
states of affairs. We show the subjects pairs of diagrams such as Figs.
8.1a and 8.1b and ask which one is designated by

7 All the circles don't have dots in them.

Faced with this problem, most subjects now switch to NEG-V inter-
pretations and select 8.1b. In our first approach, we submitted Fig.
8.1 before any sentence tests such as 1-3 and we obtained 100 percent
The Study of Language in its Social Context 195

(b)
Fig. 8.1. "All the circles don't have dots in them.
196 SOCIOLINGUISTIC PATTERNS

NEG-V. When we placed the diagram after the sentence tests, the
majority switched to NEG-V from NEG-Q. In one recent series,
still

39 of 58 subjects responded NEG-V to the diagrams, and of these

only 17 could be led to see NEG-Q as an alternate possibility. Fur-
thermore, we were able to control this response by using variant
forms of Fig. 8.1a with one out of eight circles having dots, seven
out of eight, or four out of eight. Table 8.1 shows how responses shift
systematically from the most favoring situation (1 out of 8) to the
least (4 out of 8). When there is only one dot as in Fig. 8.1a, the great
majority see only the NEG-V meaning of 7 referring to 8.1b. When
the diagram has four dotted circles about half the subjects now
first

see both NEG-Q and NEG-V interpretations. Moreover, there is no

correlation between these responses and responses to isolated sen-
tences. Table 8.1 also shows for each test the distribution by response
to sentences 1-3. In this, as in other tests we find that there are more
subjects who switch responses from sentences to diagrams than those
who maintain a consistent position. Table 8.2 shows an even more
dramatic inconsistency in another series using a test sentence

8 Everybody doesn't know how to play bridge.

Interpretations of sentence 8 in isolation favor NEG-Q less heavily

than sentence 7. In Table 8.2 we see that only 3 out of 19 subjects

TABLE 8.1

EFFECT OF VARIANT FORMS OF FIG. 8.1

ON NEG-V AND NEG-Q RESPONSES

Dots per 8 circles (in Fig. 8.1)

Vs % %
Fig. 8.1 Q* V Q V Q v

V only 5 7 1 3 2 5
V>Q 1 1 1 1 1

V = Q 1 3 3 1 —
Q>v 1 3 1

Q only 1

7 7 7 7 7 7

NEG-Q sentence: All the guys didn't leave, but some did.
NEG-V sentence: All the girls didn't arrive until eight, so the
room was empty at seven thirty.
*Subject's preference for NEG-Q or NEG-V sentences.
The Study of Language in its Social Context 197

TABLE 8.2
INCONSISTENT NEG-Q AND NEG-V RESPONSES
TO DIAGRAMS AND SENTENCES
Interpretations of sentence 8
Reactions to
Fig. 8.1 (7 dot) Q only Q> V V >Q V only

Q only
Q>v
V>Q 1

V only 4

were consistent between sentence and diagram responses a result —

that matches closely the internal consistency pattern for sentences
2-6 noted above. We can obtain an even heavier NEG-V reaction
by adding a restrictive modifier to 1, producing

9 All of us here don't play bridge.

The quantifier all is now seen by the great majority of speakers as

applying to the entire quantity we who are here, and this is inter-
preted as 'None of us here plays bridge.' In test after test, 90 to 95
percent of the subjects select NEG-V interpretations of 9, and the
great majority do not accept NEG-Q readings.
To illustrate the force of this shift in construction, we note that
in a recent test, 21 of 26 subjects gave a NEG-Q response to 1. When
this was followed immediately with sentence 9, the majority switched
to NEG-V and only eleven remained NEG-Q.
These responses indicate that the eliciting context can be con-
trolled to produce NEG-Q and NEG-V "dialects" at will. Even more
convincing evidence can be derived by observing the use of language
in everyday life. When I first introduced Fig. 8.1 to my family, there
was a heated argument. My wife gave most determined NEG-Q
answers; after thirty minutes of argument between my children and
her, she still maintained that the NEG-V interpretation was so remote
that it could not be seriously considered. As I was lying in bed the
next morning, I heard my wife calling to my son from the hallway:

10 Simon, get up! Everybody's not helping!

The "Quantifier have played an important role in recent

dialects"
have been claimed for inter-
linguistic theory. Dialect differences
pretations of sentences 11a and lib.
198 SOCIOLINGUISTIC PATTERNS

11 a. Half of the people in this room speak two languages,

b. Two languages are spoken by half of the people in this
room.

These sentences bear crucially on the problem of whether or not

transformations preserve referential meaning. A linguist who finds
that others disagree with him about whether or not 11a means the
same as lib will defend his own reaction as "my dialect," and claim
that the theory must account at least for the "facts" of his dialect.
Grinder and Postal (1971) developed some important arguments
on identity of sense anaphora which they hoped would decide the
conflict between generative semantics and interpretative theory.
Most of the sentence types put forward produce a mixed response;
those who favor generative semantics generally view the data as the
authors do, rejecting such sentences as

12 *Max wasn't imprisoned but Joe was imprisoned and he

is still there.
13 *Max is off his rocker and so is his mother.

But those who do not favor their position usually do not reject 12
and 13. In their conclusion, Grinder and Postal run into a head-on
conflict with Chomsky in the interpretation of sentences such as

14 John hasn't been here for a month, but Bill has.

The issue here is that generative semantics will expand the ellipsis
to an underlying form

14' John hasn't been here for a month, but Bill has been here
for a month.

which allows only a durative interpretation; thus a generative se-

mantic position rules out a punctual interpretation for 14. But
Chomsky allows a punctual reading by an interpretive rule, since
he does not derive 14 from a deep structure with 14'. Postal and
Grinder call Chomsky's argument "misleading and distorting" since
they find that 9 out of 10 subjects they asked agreed with them. A
student from the University of California, Los Angeles, has written
to me that the great majority of linguistic students there agree with
Chomsky's intuitions. It should be noted that the linguistics depart-
ment at U.C.L.A. favors interpretive theory. Postal and Grinder argue
that the best theory is one which can account for the facts of dialect
The Study of Language in its Social Context 199

variation; but these dialects seem to be artifacts of a theoretical

position. As linguists become more deeply involved in such theoret-
ical issues, it is likely that their intuitions will drift further and
further from those of ordinary people and the reality of language
as itused in everyday life. 10 We return to the painfully obvious
is

—
conclusion obvious at least to those outside linguistics that lin- —
guists cannot continue to produce theory and data at the same time.
Again, one must pay tribute to the difficulty and subtlety of the
questions posed. Obviously techniques for investigation must be
developed further. Yet there is no evidence that consistent and
homogeneous judgments can be obtained from native speakers on
such crucial matters. Variation in syntactic judgments can be studied
with profit and the implicational series within them analyzed to
decide the form of the rules (Elliott, Legum, and Thomson 1969). But
it is now evident that the search for homogeneity in intuitive judg-
ments is a failure. Once motiva-
this result is accepted, the strongest
tion for confining linguistic analysis tosuch judgments disappears.
In many ways, intuition is less regular and more difficult to interpret
than speech. If we are to make good use of speakers' statements about
language, we must interpret them in the light of unconscious, un-
reflecting productions. Without such control, one is left with very
—
dubious data indeed with no clear relation to the communicative
process we recognize as language itself.

Problems in the Relation of Theory to Data

The procedures of generative grammar, working with intuitions

about language, have enabled us to elaborate elegant and insightful
models of linguistic structure. We have unearthed a great fund of
problems which had never been touched on or discussed before. It

10. A drift of linguists' intuitions away from the general understanding was in-
dicated in recent unpublished research of N. Spencer of Pennsylvania State University.
She took 150 sentences from generative papers of some importance to generative theory
by Perlmutter, C. Smith, Postal, Ross, Rosenbaum, and R. Lakoff. These were submit-
ted for judgments of acceptability to 60 subjects: 20 graduate students in linguistics,
20 graduate students not in linguistics, and 20 subjects outside the university. There
was more than a little disagreement with the original judgment: in 44 of the 150 cases
the authors failed to get a majority of the subjects to agree with them, whether the
judges were naive or not. But when there was disagreement among the judges, the
non-linguist graduate students regularly sided with the naive judges from town, and
the linguistic students were left by themselves.
200 SOCIOLINGUISTIC PATTERNS

is now commonplace to assert that generative grammar is the best

discovery procedure we have. The study of intuitive judgments
focuses our attention on important relations between sentences and
the deeper structures which underlie them. But as a theory of lan-
guage approach is seriously defective, since it offers us no means
this
of discovering whether our model is right or wrong. Originally, the
generative grammar was constructed to produce all the acceptable
sentences of the language and none of the unacceptable ones. But
if we now compare the model with what speakers say, we cannot

draw any decisive conclusions from the way it matches or fails to

match the data.
1. If someone uses a sentence structure that is not generated by the

grammar, there is nothing to prevent us from setting it aside as

a mistake or a dialect difference.
2. If no one ever uses a sentence structure which is predicted by
the grammar, this fact can be discounted because most complex
syntactic forms are known to be very rare — the occasion simply
has not arisen.

This second situation can be extremely embarrassing when the

syntactic forms concerned are at the very center of the theoretical
argument. Chomsky's original argument against finite state grammars
(1957) depended on the existence of self-embedded structures in
natural language. Everyone seems to accept such sentences as

15 The man (that) the girl (that) I used to go with married

just got drafted.

as grammatical (in competence) though a bit difficult to follow (in

performance). But now that Peter Reich has challenged the gram-
matical status of this pattern and reasserted the finiteness of natural
language (1969) one looks in vain for empirical evidence of the use
of such doubly embedded forms. 11 In the interviews and conversa-
tions we have recorded, no such example from unreflecting, natural
speech has yet emerged. Current explorations with sociolinguistic
techniques of enriching the data suggest that we can establish the
grammaticality of 15, but until we have strong evidence of its use
or comprehension in natural conversation, the basic argument for

11. A finite state grammar may produce single embeddings such as "The girl I used
to go with just got married." The problem only arises when the grammar must be
in a state to "remember" that it has two subjects stored for which it must produce
predicates.
The Study of Language in its Social Context 201

hierarchical phrase structure will not rest on a solid foundation.

The problems we as linguists face in dealing directly with the data
of language are not peculiar to our discipline. This is a general
problem for all the social sciences. Garfinckel (1967) has demon-
strated that there exists in every field of research an inevitable gap
between the raw data as it occurs and the protocols in which the
data are recorded as input to the theoretical pursuit. In the socio-
we find many kinds of data used to provide
linguistic literature cited,
information about language in actual use: census data; ques-
tionnaires; extracts from plays and novels; psychological tests; eth-
nographic reports of community norms. No matter how insightful
or productive these studies may be, they do not bring us much closer
to the fundamental data of language in use than we were before.
There are many open questions we simply cannot answer. What is
the relation between the novelist's stereotype and the language
behavior of the people in question? what is the connection between
word association tests and the semantics of natural language? how
do we discover when a speaker uses tu when all we have is his
self-report, or discover when he speaks French by asking him when
he does so? what is the relation between the norms which the
anthropologist reports and the practice of members in conforming
to those norms? There are many acts of perceiving, remembering,
selecting, interpreting, and translating that lie between the data and
the linguist's report, and these are almost all implicit in such papers.
As Garfinckel has pointed out, every coding and reporting procedure
that transforms the data will show an ineradicable residue of
common-sense operation which cannot be reduced to rule. To come
to grips with language, we must look as closely and directly at the
data of everyday speech as possible, and characterize its relationship
to our grammatical theories as accurately as we can, amending and
adjusting the theory so that it fits the object in view. We can then
turn again and re-examine the methods we have used, an inquiry
which will greatly increase our understanding of the object we are
studying.

The Direct Study of the Linguistic Data

The critique of the conventional linguistic methods just given must

not be taken as a suggestion that they be abandoned. The formal
elicitation of paradigms, the exploration of intuitive judgments, the
study of literary texts, experimentation in the laboratory, and ques-
202 SOCIOLINGUISTIC PATTERNS

all important and valuable modes

tionnaires on linguistic usage are
of investigation. The
two procedures must be mastered by
first

anyone who hopes to do significant linguistic analysis. The tech-

niques to be discussed below for the direct observation of language
in use presuppose that the outlines of the grammar have been
—
sketched in that the main possibilities are known. Thus the pho-
netic transcription of an unknown language
even unknown(or
words) is quite beyond The ear is a very poor instru-
our capacity.
ment for judging the absolute quality of isolated sounds. But given
an understanding of the syntax and the morphemes intended, the
ear is a superb instrument for judging which of several possibilities
are realized. 12 In syntax, our first analyses of a given form are
relatively superficial; but when many relationships with other sen-
tence forms are noted, a rich field of possible underlying structures
begins to emerge. There is a second Cumulative Paradox involved
here: the more that is known about a language, the more we can
find out about it.
The limitations placed upon the input data by Chomsky have led
him to the conviction that the theory is underdetermined by the data
(1966) —
that there will always be many possible analyses for each
body of data, and we will need internal evaluation measures to
choose among them. We take the opposite view. Through the direct
study of language in its social context, the amount of available data
expands enormously, and offers us ways and means for deciding
which of the many possible analyses is right. In our preliminary
operations upon the initial data, considerations of simplicity will
always find a place; but given the correct line of attack, 13 it is possible
toprove whether the simple hypothesis constructed is the correct
one.The studies of covariation and change in progress discussed
below will provide considerable support for this claim.

12. Our own work in tracing sound changes in progress through spectrographs

measurements confirms the remarkable accuracy of impressionistic phonetics used

to compare two sounds. See Chs. 1 and 2.
13. Watson's discovery of the structure of DNA is one of the most striking cases

of the role of simplicity in scientific research. Watson was convinced that the solution
must be a simple one, and this conviction motivated his persistent attempts at
model-building (1969). But simplicity merely suggested the best approach: the validity
of his model was established by the convergence of many quantitative measures.
Hafner and Presswood (1965) cite another case in the theory of weak interactions
where considerations of simplicity led to a new theoretical attack; but again, as in
all other cases I know, the acceptance of the theory as correct depended upon new

quantitative data.
The Study of Language in its Social Context 203

Resolution of Problems in the Study of Everyday Language

Among the motivations discussed for the restriction of linguistic

data to intuitions were difficulties in working with everyday speech.
Fortunately for our studies, many of these problems turned out to
be illusory, or greatly exaggerated.

1. The ungrammaticality of everyday speech appears to be a myth

with no basis in actual fact. In the various empirical studies that
we have conducted, the great majority of utterances about 75 —
—
percent are well-formed sentences by any criterion. When rules
of ellipsis are applied, and certain universal editing rules to take
care of stammering and false starts, the proportion of truly un-
grammatical and ill-formed sentences falls to less than two per-
cent (Labov 1966b). When nonacademic speakers are talking about
—
subjects they know well narratives of personal experience the —
proportion of sentences that need any editing at all in order to
be well-formed drops to about ten percent. I have received con-
firmation of this general view from a great many other linguists
who have worked with ordinary conversation. The myth of the
ungrammaticality of spoken language seems to have two sources:
data taken from transcripts of learned conferences, where highly
educated speakers are trying to express complex ideas for the first
time; and the usual tendency to accept ideas that fit into our frame
of reference without noticing the data with which we are sur-
rounded.
2. The existence of variation and heterogeneous structures in the
speech communities investigated is certainly well-established in
fact. It is the existence of any other type of speech community

that may be placed in doubt. There is a kind of folk-myth deeply

embedded among linguists that before they themselves arrived
on the scene there existed a homogeneous, single-style group who
really "spoke the language." Each investigator feels that his own
community has been corrupted from this normal model in some
—
way by contact with other languages, by the effects of education
and pressure of the standard language, or by taboos and the
admixture of specialized dialects or jargons. But we have come
to the realization in recent years that this is the normal situation
that heterogeneity is not only common, it is the natural result of
basic linguistic factors. We argue that it is the absence of style-
shiftingand multilayered communication systems which would
be dysfunctional (Weinreich, Labov, and Herzog 1968:101).
204 SOCIOLINGUISTIC PATTERNS

Once we assumed association between structure

dissolve the
and homogeneity, we are free to develop the formal tools
needed
to deal with inherent variation within the speech community.
Again, we find ourselves fortunate in that the patterning within
this variation is by no means obscure: it does not require the
statistical analysis of hundreds of speakers' records as linguists
traditionally feared (Hockett 1958:444). On the contrary, we find
that the basic patterns of class stratification, for example, emerge
from samples as small as 25 speakers. 14 As we saw in Ch. 2, regular
arrays of stylistic and social stratification emerge even when our
individual cells contain as few as five speakers and we have no
more than five or ten instances of the given variable for each
speaker. With this regular and reproducible data, we are in a
position to specify what we mean by the "stylistic" or "social"
meaning which seems so elusive when language is studied out
of context.
3. The problem of recording speech in natural settings is a technical
one: and the development of professional battery-operated tape
recorders has made it possible to obtain excellent results in the
field.Given a good microphone, a field worker can obtain excel-
lent recordings under noisy conditions by minimizing mouth to
microphone distance for each speaker. 15 In general, it may be said
that the current problem lies primarily in the failure of linguists
to respond to the invention of the magnetic tape recorder in
Germany in the 1930's. There is no tradition in linguistics of
solving technical or experimental problems, of assessing techno-
logical developments and responding to them; otherwise the
nature of linguistic operations would have been transformed 30
years ago.
4. The fourth problem to be resolved is the rarity of crucial gram-
matical forms needed for evidence. There is no immediate solu-
tion on hand but the direction of the answer is beginning to
appear. A deeper understanding of the communicative function
of grammatical forms will enable us to enrich the data of ordinary

14. This conclusion is supported throughout Labov 1966a and Labov et al. 1968,
but most strikingly in Shuy, Wolfram, and Riley 1967. From a very large sample of
over 700 interviews, 25 were selected for analysis, and extremely regular patterns of
social stratification emerged for a number of linguistic variables. See also Kucera
1961:97-98, where 19 subjects are stratified into at least four classes.
15. See Sect.1 below for some difficult consequences of this advance.
The Study of Language in its Social Context 205

conversation. The ideal mode of operation is for the linguist to

engage in a normal conversation with an informant, and be able
to elicit the informant's natural use of a given form without using
it himself. Obviously there is feed-back here between abstract

analysis and field methods: the ability to control the production

of a given form confirms our analysis and provides contextual
data on its use. We have had some success in eliciting and con-
trolling items such as the English passive and present perfect
forms in this way. Eventually, we will be in a position to assert
that a speaker does not have a given form in his system because
of his consistent failure to use it in a context where other members
of the community do so regularly.

Sources for the Study of Language in its Social Context

There are now in print a number of empirical studies which

demonstrate convincingly that the direct study of language is a
practical and fruitful procedure. The research to be discussed in this
chapter and the next is relatively recent: the work of ten investi-
gators or groups of investigators who use as their primary data
accurate records of language in its normal social context. The first
two are brief studies incidental to other research; the others are
large-scale undertakings, specifically designed for the study of the
speech community.

1. John L. Fischer's brief study of the -ing suffix used by children

in a New England community (1958).
2. Henry Kucera's observations of the use of Common Czech and
LiteraryCzech variables in the speech of 19 exiles on French
radio stations (1961).
3. John Gumperz' investigations of dialect stratification and code
switching in Khalapur, India and Hemnes, Norway (1964, 1967),
and his study of Marathi-Kannada bilingualism in Kupwar, India
(Gumperz et al. 1971).
4. Lewis Levine and Harry Crockett's report on the use of post-
vocalic r in their sociolinguistic survey of Hillsboro, N.C. (1966);
Frank Anshen's study of four phonological variables in the black
population of that city (1969).
5. Investigations of Spanish-English bilingualism in the Puerto
Rican community of New York City and Jersey City by Joshua
206 SOCIOUNGUISTIC PATTERNS

Fishman, John Gumperz, and Roxana Ma (Fishman et al. 1968)

and particularly Ma and Herasimchuk's study of Spanish and
English variables.
6. Roger Shuy, Walt Wolfram, and Ralph Fasold's study of the social
phonological and grammatical variables in De-
stratification of
troit English (Shuy, Wolfram, and Riley 1967) and Wolfram's
analysis of black speech within that study (1969).
7. The sociolinguistic investigations of the French of Montreal by
Gillian Sankoff, Henrietta Cedergren, and their associates
(Sankoff and Cedergren 1971; Sankoff, Sarrasin, and Cedergren
1971; Sankoff 1972).
8. The investigation of the social stratification of English in Nor-
wich, England, by Peter Trudgill (1971).
9. The study of the Spanish of Panama City by Henrietta Cedergren
(1970).
10. My own study of centralization on Martha's Vineyard (Ch. 1);

of the stratification of English in New

York City (Chs. 3-6); and
with Paul Cohen, Clarence Robins and John Lewis, of the struc-
ture and use of the Black English Vernacular (Labov 1972a; Labov
et al. 1968).

In addition, I will be drawing on studies of social attitudes towards

languages and dialects by Lambert and his colleagues (1967). Though
these are based entirely on test reactions, they fit in and help to
explain the other data cited above. There are also smaller studies
carried out by students, and a number of major studies now in
progress which add to our understanding of the principles involved.
Wecan best understand the value of this empirical research if we
apply it to the kind of specific theoretical problems of linguistic

structure which concern all linguists. The study of language in its

social context takes up the same range of linguistic problems as other
approaches to linguistic theory. We can isolate five general questions:

1. What is the form of the linguistic rule? and what constraints may
be placed upon it?

2. What are the underlying forms upon which rules operate, and
how can they be determined accurately in any given case?
3. How are rules combined into systems? and how are they ordered
within these systems?
4. How are systems related to each other in bilingual and poly-
systemic situations?
The Study of Language in its Social Context 207

5. How do rules and rule systems change? What is the mechanism

of the fundamental processes of language acquisition, or how do
rules change in the larger course of linguistic evolution?

Sect. 1 will present methods for gathering reliable data within the
speech community; Sect. 2 will deal with the methods used for
analyzing these data and show the kind of solutions to internal lin-
guistic problems that are possible; Sect. 3 will deal with the broader
sociolinguistic structures and the interaction of social and linguistic
factors. The and the formal approach is origi-
theoretical analysis
nally my own, based to a large extent on the studies listed under
10, above, but the convergence of findings and principles in the field
is very striking indeed. Most recently, important modifications have

been made to the formalization of variable rules which are reflected

in the version given in the following pages. Cedergren and D. Sankoff
(1972) have quantified the variable constraints as underlying proba-
bilities, and thus raised the entire discussion to a higher level of

accountability.
In all of these discussions we will make use of the facts of inherent
variation to resolve abstract questions which would otherwise re-
main as undecided, moot possibilities. The aim here is not neces-
sarily to provide linguistics with a new theory of language, but rather
to provide a new method of work.

1. Methodology

In any academic course that deals with research in the speech

community, there is always a great deal of interest in the first steps
to be taken: "What do you say to people?" This is not a trivial
question. The elementary steps of locating and contacting inform-
ants, and getting them to talk freely in a recorded interview, are
formidable problems for students. It is an error for anyone to pass
over these questions, for in the practices and techniques that have
been worked out are embodied many important principles of lin-
guistic and social behavior. Close examination of these methodo-
logical assumptions and findings will tell us a great deal about the
nature of discourse and the functions of language.
The fundamental sociolinguistic question is posed by the need to
understand why anyone says anything. There are methodological
questions of sampling and recording which merely set the stage for
208 SOCIOLINGUISTIC PATTERNS

the basic problems. It was noted above that good data require good
recording, especially for the grammatical analysis of natural speech.
After the crucial variables have been defined and isolated, a great
deal can be done with handwritten notes. But our initial approach
to the speech community is governed by the need to obtain large
volumes of well-recorded natural speech.
We can isolate five methodological axioms supported by the find-
ings of the field research projects cited which lead to a methodologi-
cal paradox; the solution to this paradox is the central methodologi-
cal problem.

1. Style Shifting. As far as we can see, there are no single-style

speakers. Some informants show much wider range of style
a
shifting than others, but every speaker we have encountered
shows a shift of some linguistic variables as the social context
and topic change (Ch. 3). Some of these shifts can be detected
qualitatively in the minor self-corrections of the speaker, which
are almost always in a uniform direction.
2. Attention. There are a great many styles and stylistic dimensions
that can be isolated by an analyst. But we find that styles can
be ranged along a single dimension, measured by the amount of
attention paid to speech. The most important way in which this
attention is exerted is in audio-monitoring one's own speech,
though other forms of monitoring also take place. 16 This axiom
(really an hypothesis) receives strong support from the fact that
speakers show the same level for many important linguistic vari-
ables in casual speech, when they are least involved, and excited
speech, when they are deeply involved emotionally. The common
factor for both styles is that the minimum attention is available
for monitoring one's own speech.
3. The Vernacular. Not every style or point on the stylistic contin-
uum is of equal interest to linguists. Some styles show irregular

phonological and grammatical patterns, with a great deal of

"hypercorrection." In other styles, we find more systematic
speech, where the fundamental relations which determine the
course of linguistic evolution can be seen most clearly. This is
—
the "vernacular" the style in which the minimum attention is
given to the monitoring of speech. Observation of the vernacular
gives us the most systematic data for our analysis of linguistic
structure.

16. See experiments of Mahl with white noise in Ch. 3.

The Study of Language in its Social Context 209

4. Formality. Any
systematic observation of a speaker defines a
formal context which more than the minimum attention is paid
in
to speech. In the main body of an interview, where information
is requested and supplied, we would not expect to find the ver-

nacular used. No matter how casual or friendly the speaker may

appear to us, we can always assume that he has a more casual
speech, another style in which he jokes with his friends and argues
with his wife.
5. Good Data. No matter what other methods may be used to obtain
samples of speech (group sessions, anonymous observation), the
only way to obtain sufficient good data on the speech of any one
person is through an individual, tape-recorded interview: that is
through the most obvious kind of systematic observation. 17

We are then left with the Observer's Paradox: the aim of linguistic
research in the to find out how people talk when
community must be
they are not being systematically observed; yet we can only obtain
these data by systematic observation. The problem is of course not
insoluble: we must either find ways of supplementing the formal
interviews with other data, or change the structure of the interview
situationby one means or another. Of the various research projects
mentioned above, not all have been successful in overcoming this
paradox. Many investigators have completed their work with only
a limited range of stylistic data, concentrated in the more formal
ends of the spectrum. Systematic study of the vernacular has been
accomplished primarily in Gumperz' work, in our own work in New
York City and in urban ghetto areas, and in the Fishman-
Gumperz-Ma project in Jersey City.
One way of overcoming the paradox is to break through the
constraints of the interview situation by various devices which divert
attention away from
speech, and allow the vernacular to emerge.
This can be done in various intervals and breaks which are so
defined that the subject unconsciously assumes that he is not at that
moment being interviewed (Ch. 3). We can also involve the subject
in questions and topics which recreate strong emotions he has felt
in the past, or involve him in other contexts. One of the most suc-
cessful questions of this type is one dealing with the "Danger of
Death": "Have you ever been in a situation where you were in serious

17. There are some situations where candid recording is possible and permissable,
but the quality of the sound is so poor that such recordings are of confirmatory value

at best.
210 SOCIOUNGUISTIC PATTERNS

danger of being killed?" (Ch. 3, p. 92). Narratives given in answer

to this question almost always show a shift of style away, from careful
speech towards the vernacular. 18
One cannot expect that such devices will always be successful in
obtaining a radical shift of style. A more systematic approach uses
the normal interaction of the peer-group to control speech instead
of the one-to-one confrontation of subject and interviewer. In Gum-
perz' work in Hemnes (1964), the fundamental data was obtained
through recorded sessions with natural groups. In our work in
south-central Harlem, (Labov et al. 1968) we studied adolescent peer
groups through long-term participant observation. Individual in-
terviews were carried out with all members of the group, yielding
the individual data we needed on each individual. A series of group
sessions was held in which the speech of each member (picked up
from a lavaliere microphone) was recorded on a separate track.
There was no obvious constraint in these group sessions; the adoles-
cents behaved much as usual, and most of the interaction physical —
—
and verbal took place between the members. As a result, the effect
of systematic observation was reduced to a minimum. This is at
present the only quantitative study of a self-selected, naturally
formed peer group in the literature. It is hoped that other such studies
will be carried out in the near future.

Rapid and Anonymous Interviews

In the methods just described, the identity and demographic posi-
tion of each subjectwell known. One can also carry out systematic
is

observation anonymously, in conversations which are not defined

as interviews. In certain strategic locations, a great many subjects
can be studied in a short period of time, and if their social identity
is well defined by the objective situation, the findings can be very
rich. The department-store survey (Ch. 2) has provided a model for
such rapid and anonymous work and a number of similar projects
have confirmed the reliability and practicality of this approach (for
details and other possibilities see Ch. 2).

18. One of the most interesting aspects of this question is that it involves a yes-no
answer, which we
normally avoid. The mechanism seems to be that the informant
is willing to commit himself to the fact of having been in such a situation, though

he may be unwilling to volunteer an account. But having so committed himself, he

finds it very difficult to avoid giving a full account when the interviewer asks, after
some delay, "What happened?" Otherwise, he would appear to have made a false
claim.
The Study of Language in its Social Context 211

Unsystematic Observations
The crucial question to be asked in any of these studies is whether
one has indeed obtained data on the fundamental, systematic ver-
nacular form of the language. Unsystematic and candid observation
in speech at various strategic points can tell us a great deal about
our success in this regard. One can record a number of constant and
variable features from large numbers of people in public places such
as trains, buses, lunch counters, ticket lines, zoos —
wherever enough
members of the speech community are gathered together so that their
speech is naturally and easily heard by others. There are many biases
built into such observations —
loud and less educated talkers, for
example, are strongly selected. But as a corrective to the bias of the
interview situation, such data can be very valuable.

Mass Media
It is also possible to obtain some systematic data from radio and
television broadcasts, although here the selection and the stylistic
constraints are usually very strong. In recent years, we have had a
great many direct interviews at the scene of disasters, where the
speakers are too strongly under the immediate influence of the event
to monitor their own speech. Conversation programs and speeches
at public events can give us a good cross-section of a population, but
here the style is even more formal than that we would obtain in a
face-to-face interview.

The Formal End of the Stylistic Range

It is relatively easy to extend the range of styles used by the

speaker towards the formal end of the spectrum, where more atten-
tion is given to speech. There are many questions which naturally
evoke more careful speech (such as questions about speech itself).
In most of the urban studies carried out so far, reading texts were
used to study phonological variations. In general, linguistic variables
show a marked shift from the most formal elicitation to the least
formal reading. Traditional texts such as "Grip the Rat" are relatively
artificial, designed to include as many dialect-sensitive words as
possible. A text that reads well, focusing on vernacular or adolescent
themes, will yield much less formal speech than these or lists of
isolated words. Minimal pairs can then be embedded in such a text
so that the speaker is not aware of the contrast (Ch. 3). The work
of Shuy, Wolfram, and Riley (1967), Wolfram (1969), and Trudgill
212 SOCIOLINGUISTIC PATTERNS

(1971) utilized such texts, although the construction of special read-

ings for particular variables has not been advanced as sharply as
the work deserves.
Levine and Crockett (1966) and Anshen (1969) used another method
to extend the stylistic range of readings. Sentences were constructed
in which the variables were embedded, and at other points in the
same sentences blanks were inserted for the subject to fill in lexical
items as he read, diverting his attention from the variables. The
pronunciation of the phonological variable (r) in this context showed
than in the reading of isolated words.
less (r-1)
A number
of formal tests do not require any reading on the part
of the subjects. Perception tests of the ABX form provide useful
information: in the case of total merger of a phonological distinction,
speakers cannot hear whether X is closer to A or B; but where
variable rules are operating, and the merger is not complete, they
will show partial success. A surprising amount of grammatical in-
formation can be obtained by repetition tests with older subjects.
Psycholinguists have long used such repetition tests with children
2 to 5 years old, but we found to our surprise that with speakers
of nonstandard dialects the underlying grammatical rules of much
older subjects, 10 to 17 years old, controlled the form of their repeti-
tions. Speakers of the Black English Vernacular had no difficulty in
repeating accurately long sentences within their own grammatical
system, but many sentences in standard English were given back
instantly in vernacular form (Labov et al.: 1968, 3.9). 19
A number of formal tests have been developed to isolate social
attitudes towards language, and the social information carried by
dialect forms. One can play family background tests —
taped sections
—
of "typical" speakers and ask subjects to identify their ethnic
background, race, social class (Labov et al. 1968: 4.4; Brown 1969).
This tells us whether or not the listeners can obtain this social
information from speech, but not where the information is located
in the speaker's grammar, phonology, intonation, or voice qualifiers.
Subjective reaction tests allow us to separate the linguistic variables
from personal factors. The "matched guise" technique used by
Lambert and his students (Lambert 1967) presents for the subject a

19. These observations have since been confirmed by larger-scale tests carried out
with school populations, where the subjects' relation to the vernacular was not as
well known (Baratz 1969; Garvey and McFarlane 1968).
The Study of Language in its Social Context 213

series of tape-recorded sections in which voices of the same speakers

are heard using different languages or dialects. The subjects are
asked to make judgments of the speakers' personalities. As long as
they cannot know how they have rated the same speakers before,
they unconsciously translate their social attitudes towards language
into differential judgments of the speaker's honesty, reliability, in-
telligence, etc. In our own subjective reaction tests (Ch. 6) the same
speakers are heard reading sentences which differ principally by
their treatment of the linguistic variable being studied. The subjects'
evaluation of the social significance of this variable is registered by
their differential responses to the matched sentences, on such scales
as "What is the highest job the speaker could hold, talking as he
does?" or "If the speaker was in a street fight, how likely would he
be to come out on top?"
Speakers' attitudes towards well-established linguistic variables
will also be shown in self-evaluation tests. When asked which of
several forms are characteristic of their own speech, their answers
reflect the form which they believe has prestige or is "correct," rather
than the form they actually use. Here again, this kind of test data
cannot be interpreted without data on the subjects' actual speech
patterns. (See Labov 1966a, Ch. 12; Trudgill 1971.)
We can investigate speakers' awareness of stigmatized well-
marked social variants by ciassroom correction tests, asking them
to correct sentences which depart from school or classroom models
(Labov et al. 1968:4.4). But it is almost impossible to obtain inter-
pretable results on the reverse type of vernacular correction tests,
in which the subject is asked to correct standard prestige forms to
the nonstandard vernacular. The influence of the formal test situation
is such that the subject cannot perceive accurately the nonstandard

rules. There is some evidence that the audio-monitoring norm which

governed production of the nonstandard form in childhood is re-
placed by the prestige norm, so that it is not possible in general for
most speakers to direct their attention accurately to nonstandard
rules. This result reflects an important axiom of vernacular shifting:
whenever a subordinate dialect is in contact with a superordinate
dialect, answers given in any formal test situation will shift from
the subordinate towards the superordinate in an irregular and
unsystematic manner. The terms "superordinate" and "subordinate"
here refer to any hierarchical social dimension equivalent to "pres-
tige" and "stigmatized." Some linguists hope that by "educating" the
214 SOCIOLINGUISTIC PATTERNS

informant in the goals of the analysis, it will be possible to diminish

this effect, and gradually obtain answers characteristic of the pure
vernacular. But this is an illusion. Instead, the subject may use his
knowledge of the prestige dialect to avoid giving any vernacular form
which is identical or similar to the standard, and so produce stereo-
typed forms which are simply a collection of the "most different"
or "worst" sentence types. Speakers who have had extensive contact
with the superordinate form no longer have clear intuitions about
their vernacular available for inspection. 20
There is further reason to regard as suspect data on a nonstandard
vernacular gathered from an "educated" informant. Usually the
investigator speaks the standard superordinate dialect which is
dominant in this face-to-face interviewing situation. The informant's
capacity to learn languages is operating at all times, and there is
evidence that his grammatical rules will be heavily influenced by
the standard during this period of elicitation. 21
Once in a great while we encounter an informant who seems
—
almost immune to "correction" of this sort who seems to have direct
access to his intuitions, despite his knowledge of the standard dialect.
An important task for psycholinguists is to identify other traits which
accompany or determine this behavior, so that we will be able to
search a given population for "ideal" informants. But it will always
be necessary to calibrate the informant's responses against other data
of the vernacular to see if he does indeed have access to his original
rules. To evaluate this data, we must already know the rules of the
vernacular from the direct observation of casual speech. But the
procedure is not entirely circular; for if we have confidence in the
"immune" informants, we may obtain crucial data
introspections of
on forms which are too rare to find in any body of casual speech.
Whether or not we are safe in extrapolating from observed stability
on common forms to unobserved stability on rare forms is an open
question.
These considerations do not necessarily apply to linguists studying

20. This is obviously true in the case of children. One cannot ask young children

whether a nonstandard sentence of theirs is well-formed, nor ask adults to reconstruct

their childhood grammars. It is true in general that learning one series of rules closely
related to the older series makes it impossible to reconstruct the earlier situation.
21. Our own worker in south-central Harlem, John Lewis, showed a strong
field
shift of the nonstandard variables we were investigating from the time that he was
first interviewed (1965) to the time that he finished interviewing others (1967).
The Study of Language in its Social Context 215

languages through an intermediate language which is not marked

22
socially with regard to their object language. It is normal for a

linguist who approaches a language for the first time to work with
bilingual informants, who may not even be good speakers of the
object language. Such preliminary steps in formal elicitation are of
course necessary prerequisites to the accurate study of language in
its social context. Good linguists can go further than this, and draw

their best data from recordings of native speakers talking to each

—
other parallel to the group sessions mentioned above. The study
of language in its social context can only be done when the language
is "known" in the sense that the investigator can understand rapid

conversation. When an anthropological linguist enters into this more

advanced study, then the axiom of vernacular shifting will apply,
for there will inevitably be stylistic levels which he will want to
distinguish.
Although one can achieve a certain amount of insight working with
bilingual informants, it is doubtful if as much can be said for "bi-

dialectal" informants, if indeed such speakers exist. We have not

encountered any nonstandard speakers who gained good control of
a standard language, and still retained control of the nonstandard
vernacular. Dialect differences depend upon low-level rules which
appear as minor adjustments and extensions of contextual condi-
tions, etc. It appears that such conditions inevitably interact, and
although the speaker may indeed appear to be speaking the vernac-
ular, close examination of his speech shows that his grammar has
been heavily influenced by the standard. He may succeed in con-
vincing his listeners that he is speaking the vernacular, but this
impression seems to depend upon a number of unsystematic and
heavily marked signals. 23
There are speakers in every community who are more aware than
others of the prestige standard of speech, and whose behavior is more
influenced by exterior standards of excellence. They will show

22. In his first approach Lolo-Burmese language of Thailand and Burma,

to Lahu, a

J.
It is his opinion that if he had
Matisoff used an English-Lahu bilingual speaker.
used a more closely related language such as Thai, the distortion of the data would
have been much greater.
23. The ways in which such speakers convince their listeners that they are speaking
the vernacular is an important problem for sociolinguistic study. Educated leaders
of the black community in the United States provide many examples of this phe-
nomenon.
216 SOCIOLINCUISTIC PATTERNS

greater style shifting than thosewho do not recognize such a stand-

ard. This can be measured by linguistic insecurity tests. For a
trait
selected list of socially marked variants, the subject is asked which
of two forms is correct; and then which he actually uses himself.
The total number of items where the two responses differ forms a
sensitive index of linguistic insecurity (Chs. 4 and 5).

2. Resolving Problems of Linguistic Structure

This section will present three distinct problems of linguistic

structure which have arisen in the study of the black English ver-
nacular (BEV): problems concerning the internal rules and the un-
derlying elements upon which the rules operate. Within the abstract
study of linguistic possibilities, these problems are only partly de-
cidable. Data from the study of speech in its social context, obtained
by the methods outlined above, will be used to provide what seems
to us decisive solutions for each of these problems.
Each of these solutions is reinforced by the convergence of data
drawn from many sources. Within our own studies, we have parallel
data for six different adolescent BEV peer groups, several adult
populations, and many exploratory samples in other cities. On each
of these problems, independent confirmation is provided by the work
of Wolfram in Detroit (1969) with a completely different population
and different analysts. The regular convergence of data drawn from
completely different studies provides the kind of strong evidence that
leads us to assert that these are correct solutions.

1. Consonant Cluster Simplification and the Past Tense Suffix

As noted above, BEV shows a marked pattern of consonant cluster

simplification at the ends of words. Words which show in standard
English [SE] consonant clusters ending in -t,d, frequently appear in
BEV with only the first consonant. Thus bold, find, and fist are
frequently pronounced boJ', fin' and /is'. The question arises, is this
indeed a case of cluster simplification, or are these final consonants
simply absent in BEV? The existence of plurals such as iisses for
lists suggests that some such words have the underlying forms with-

out t. We can put this question sharply only after a series of prelimi-
nary investigations which enable us to define the variable as we have
done here. The argument presented here outlines the solution given
in detail in Labov et al. 1968:3.2. Given a proper definition of the
The Study of Language in its Social Context 217

variable, any small body of speech from any BEV group or individual
will provide the following evidence:

a. There are no speakers who never have these clusters: nor are there
any who always preserve them: it is a case of inherent variation
in BEV.
b. For every speaker and every group, the second consonant is absent
more often when the following word begins with a consonant than
when it begins with a vowel. This regular effect of a following
vowel is a characteristic feature of other phonological rules: it
also constrains the vocalization of final r, 1 or nasals in many
dialects.
c. There is little or no hypercorrection: that is, final -t or -d are not
supplied for the wrong word class, giving us mold for mole or
iipt for Jip.

These facts show that the full cluster is present in the underlying
form of act, bold, or find, and that a variable rule deletes the second
consonant. In our formal representation of this process, fact c can
easily be shown by supplying the correct underlying forms in the
dictionary. Fact a can be shown by making the deletion rule optional.
But in conventional generative terms, there is no way to show for-
mally fact b. If we wrote:

15 pt [-cont]-^0/[ + cons] ## [-syl]

we would be stating that a stop is optionally deleted after a con-
sonantal segment (liquid or obstruent), at the end of a word, if the
next word does not begin with a vowel. 24
This rule provides a reasonable description of the colloquial
system of many middle-class speakers, who often say firs' thing and
las' month, but not firs' of all or las' October. But 15 is not at all
adequate for most other dialects where some clusters are simplified
a good proportion of the time when the next word begins with a

24. Wolfram argues that only those clusters which show homogeneous voicing
should be included in the rule, thus excluding jump, belt, bent, etc. Certainly the first
has to be excluded, but we do so by restricting the rule to apical stops t,d. Clusters
-sp and -sk are affected by a preceding rule, along with -st which operates at a much
higher frequency and with special constraints such as the fact that -sps and -sts are
categorically simplified in many dialects. But we find that there is a high percentage
of simplification of -It, -nt, etc., in many dialects, so that if the rule is restricted to
homogeneous voicing, a special rule will have to be written for these types.
218 SOCIOLINGUISTIC PATTERNS

vowel. For these dialects we would have to remove — syl] from the
[

environment:

16 pt [-cont]^0/[ + cons] ##
This rule states that any final -t,d cluster can be simplified and that
it makes no difference whether or not a vowel follows. But this is
so patently false to the facts of the situation thatit offers a very poor

description of the language. For speakers of English, rule 16

all

operates more often when the next word does not begin with a
vowel. The existence of such phonological conditioning, as we noted
above, is the most important indication of the speaker's knowledge
of the underlying forms. If we are to represent these facts fairly we
must somehow capture in our formal representation the existence
of variable phonological conditioning.
The first step in developing such a formal notation is to generalize
the notion of optional rule to that of a variable rule. We do so by
assigning to every rule a quantity y representing the proportion of
cases in which the rule applies out of all those cases in which it
might do so. For a categorical, invariant rule, <p = 1, and in a variable
rule, < <
(jp 1. Such a variable output is indicated by angled
brackets around the element to the left of the arrow. If <p is affected
by the presence or absence of some feature in the environment, that
element acts as a variable constraint and is placed in angled brackets
in the environment to the right of the slash. Thus a following obstru-
ent or pause in BEV favors the operation of t,d deletion, and con-
strains what would otherwise be free variation.

17 [-cont]-> <0> / [ + cons] ## <-syl>

Informally, 17 states that a stop is variably deleted after a conso-
nantal segment at the end of a word, more often if a vowel does
not follow than if a vowel does follow.
This form of the rule is quite satisfactory for many nonstandard
dialects,whose speakers do occasionally say firs' of all. But it applies
only to clusters of the form CC without a morpheme boundary
between the two consonants; if does not apply to clusters in passed
form
[paest] or rolled [rold] of the abstract C#C where the second
consonant represents the past tense. In most nonstandard dialects,
—
such clusters are occasionally deleted in the southern United States
more often than in the North, and especially in BEV. We might allow
this by inserting the boundary optionally in our rule:
The Study of Language in its Social Context 219

18 [-cont] -^ <0> / [ + cons] (#) ## <-syl>

This is an odd rule, for if a consonant after # is deleted, then the

25
entire signal of the past tense disappears for regular verbs. The
question arises whether the boundary # is indeed present in BEV:
do speakers "know" in any linguistic sense that the [-st] cluster in
passed represents the past tense? Grammatical searching of our
group sessions, individual interviews, shows that for every individual
and for every group, the following facts hold:

a. There are no speakers who always delete the -ed in these clusters,
and no speakers who never do.
b. There is phonological conditioning for these clusters as well: a
following vowel has a strong effect in preserving them.
c. In each phonological environment, past tense clusters are deleted
less often than monomorphemic clusters.
d. There is no hypercorrection: the -ed ending is not supplied
wrongly where the present tense would be expected.

For any samples of speech of even moderate length, we then find

the relations holding that are expressed in Fig. 8.2.

-V

Monomorphemic Past tense

Fig. 8.2. Four relations governing — t,d deletion.

25. Whenever the consonant deleted represents a whole morpheme, then the effect
of a following vowel will not allow native speakers to reconstruct the underlying form.
For even if the [t] is preserved in I passed Edith, this does not tell us anything about
whether the -ed signal is present in I passed Mike. For monomorphemic clusters in
fist, etc., such an alternation does tell us what the underlying form of the word is.
220 SOCIOLINGUISTIC PATTERNS

If we divide -t,d clusters into these four classes, we find in every

case that past tense clusters are simplified less often, and clusters
before vowels less often than other clusters. These relations are
remarkably uniform: they hold for every individual and every group.
The constraints upon the rule then appear as

19 [-cont] -> <0> / [ + cons] <0> ## <-syl>

Here it is boundary which favors the rule. We can
the absence of a
restate 19 informally by saying that a stop is variably deleted after
a consonantal segment at the end of a word, more often if it is not
a separate inflectional morpheme, and more often if it is not followed
by a vowel.
The rule 19 with two variable constraints will now apply generally
to a wide variety of dialects. Wherever variability exists, these
constraints are binding on all speakers of English. For a particular
dialect, such as BEV, we may want to establish more specific rela-
tions of order among the constraints, and weight one more heavily
than the other. We can do so by the use of Greek letters to the upper
left of the first angled bracket which indicates the relations of more

or less. Thus 20 indicates that the phonological constraint is weighted

more heavily than the grammatical constraint:

20 [-cont]-> <0> / [ + cons] ^<0> ## *<-syl>

This representation corresponds to Fig. 8.2, where the major effect
is the grammatical constraint. This is the situation for BEV among

the adolescent peer groups we have studied. But as the speaker gets
older, or as he talks more formally, the grammatical environment
has a stronger effect, until the positions of a and /? are reversed. This
alternation in the ordering of constraints upon rules represents one
of the elementary forms of linguistic change, genetically or dia-
chronically. It is one of the most important motivations for incorpo-
rating variable constraints into our representation of rules, for oth-
erwise we have no formal way of registering this fundamental aspect
of language development.
The existence of variation does not itself tell us that the variable
element is actually present in our underlying grammar. For example,
third singular -s also appears variably in BEV clusters, as in He works
vs. He work. But in contrast with the -t,d situation, our grammatical
searching establishes the following facts:
The Study of Language in its Social Context 221

a. There are some speakers who show no third singular -s at all even
in careful speech, and other individuals vary widely in the amount
of -s they use.
b. There is no general phonological process operating on clusters
ending in -s or -z, for the plural is almost completely intact in
BEV.
c. A following vowel does not act to preserve third singular -s. If
anything, this -s is present less often when a vowel follows.
d. There is a great deal of hypercorrection: the -s appears unpre-
dictably in other persons and numbers (We works there) or even
in nonfinite positions (He can gets hurt).

Evidence from formal tests confirms in many ways the analysis

given here. Repetition tests, perception tests, and comprehension
tests all show that the -ed suffix is easily supplied by BEV speakers
of all ages, like the plural -s suffix, but that third-singular -s is very
difficult for them to perceive, produce reliably, or comprehend. 26
The independent investigation of Wolfram in Detroit provides
precisely the same set of facts. Table 8.3 shows the percentages of
-t,d environments for three adult social
clusters simplified in the four
classes and four peer groups in New York City, and four social classes
in the Detroit study. 27 Here we see that for every group, the relations
given in Fig. 8.2 hold. Column 1 is greater than column 2, and column
3 is greater than column 4, showing the effects of a following vowel.
Column 1 is greater than column 3, and column 2 is greater than
column 4, showing the effect of the past tense boundary. As with
all the other sociolinguistic data to be presented here, it is obvious
to the sophisticated statistician that tests of significance are irrele-
vant. The original data are given in a form such that statistical
analysis can be carried out if desired, but it is plain that even if a

26. Recent work of Jane Torrey (1969) with younger children in south-central Har-

lem shows that the third singular -s can be used as a signal of the present tense, but
not to distinguish singular from plural, as in The cat splashes vs. The cats splash.
27. This data is given for the individual interviews in New York City and in Detroit.
The relations of Fig. 8.2 are preserved in the group sessions in New York, with the
grammatical constraint operating at a lower level. The Detroit figures are not exactly
comparable to the New York City figures for two reasons: (1) the Detroit interviews
do not distinguish casual and careful speech for any age group, while in New York
the adult interviews show casual and careful styles distinct [Table 8.3 giving the latter
values]; (2) final positions are included in the —
K environment in the New York City
studies but in —
V (that is, nonconsonantal) for Detroit.
222 SOCIOLINGUISTIC PATTERNS

TABLE 8.3.

CONSTRAINTS ON -1,6
DELETION IN THE BLACK SPEECH COMMUNITY
°/
/o simpli fied

Clusters in single Past tense

morphemes clusters
Speech
community K V K V

New York City

Adults
Middle class 60% 28% 19% 04%
Upper working class 90 40 19 09
Lower working class 89 40 47 32
Adolescents
Thunderbirds (10-13) 91 59 74 24
Aces (10-13) 98 64 85 43
Cobras (12-17) 97 76 73 15
Jets (12-17) 90 49 44 09
Detroit
Upper middle class 79 23 49 07
Lower middle class 87 43 62 13
Upper working class 94 65 73 24
Lower working class 97 72 76 34

Source: Labov et al. 1968: Table 3.9, and Wolfram 1969: Figs. 7 and 9.

particular case were below the level of significance, the convergence

of somany independent events carries us to a level of confidence
which is unknown in most social or psychological research.
The confirmation of the New York study by the Detroit study of
-t,d deletion leaves no doubt as to the existence of the underlying
-ed form
in BEV, and the nature of the rule which is operating. The
contrast between -ed and third singular -s is also delineated clearly
in the Detroit data (Wolfram 1969:161 ff.). Since Wolfram's data do
not include group sessions or casual speech, we do not find the
lowest levels of third singular -s, as noted under a above. But we
do find b, the absence of any general phonological process affecting
-s,z:not only are other inflections preserved, but the third singular
-s isabsent just as often after verbs ending in vowels as verbs ending
in consonants. Again, c, there is no clear effect of a following vowel
in preserving the -s: on the contrary, it is present in 38 percent of
the cases before a consonant, and only 33 percent before a vowel.
The Study of Language in its Social Context 223

Finally, d, Wolfram notes frequent but irregular hypercorrection,

with the -s appearing in other persons and in nonfinite positions.
These examples give only some indication of the precision and detail
of the convergence between the New York and Detroit studies, which
allows us to write both variable and categorical rules with con-
fidence.

Is Variation Inherent in the System?

Before proceeding, we must take into account the possibility that

we have not really succeeded in isolating the basic vernacular. Even
the wild and uncontrolled group sessions with adolescents may show
the effect of observation, which may be responsible for some un-
simplified clusters. Or perhaps children at the age of 10 have already
begun to show dialect mixture with standard English in their most
casual speech, and to find the pure vernacular we must look to even
younger children. Two findings make this possibility seem quite
unlikely. First, our work with younger children does not show us
any greater homogeneity in their speech. 28 Secondly, we have already
noted that the systematic character of -t,d deletion makes it most
probable that these final consonants are present in the underlying
representation. But there is a more important theoretical reason why
we most accurate characterization
posit the variable rule 20 as the
of the basic vernacular. If the underlying rule in some "pure" system
was a categorical rule which went to completion, eliminating all final
-t,d clusters, then it would eliminate itself. There would be no basis
for positing the form hold in the dictionary of this dialect; instead
children would infer that the word was hole and the rule would have
nothing to operate on; the same would be said for rolled. A categor-
ical rule would eliminate the regular past tense inflection. 29
It is important to note that in the course of language evolution,

change does go to completion, and variable rules have become

invariant. When this happens, there is inevitably some other struc-
tural change to compensate for the loss of information involved. I

28. Work done with children 4 to 7 years old by Jane Torrey in south-central Harlem

shows the same patterns of variation in the -ed suffix and -t, d clusters as with older
children, and similar patterns with the copula and negative concord to be discussed
below.
29. This would be the case even if was context-restricted so
the categorical rule
that it did not operate before vowels. As noted in fn. 25 above, there is no way to
reconstruct the underlying form if the morpheme disappears entirely in most positions.
224 SOCIOLINGUISTIC PATTERNS

would like to cite three examples of such dramatic structural change

to support this claim.

a. In a number of English-based Creoles, phonological and gram-

matical simplification has effectively reduced clusters so that final
inflections are typically eliminated altogether. Whereas BEV
preserves the past tense accurately with irregular verbs such as
give — — —
gave, keep kep', tell tol', and the -ed ending in rolled
remains embedded in a variable rule, Trinidad English and Ja-
maican Creole use the invariant simple forms roll, give, keep, tell
for the simple past. How is the past then distinguished from the
present? In Trinidad, the auxiliary do is used, so that now the
present tense becomes the marked form, He does give as opposed
to the past He give (Solomon 1966).
b. In the English dialects of Scotland, the simplification of -t,d
clusters after voiceless stops has become categorical, giving a
uniform ac' for act, and ap' for apt. The same rule would have
eliminated the past tense in liked and stopped. But a generalized
rule of epenthesis preserves these past tense morphemes so that
they do not form clusters, and we have [laikit] and [stapit] for
liked and stopped (Grant and Dixon 1921).
c. In the evolution of the French language, a sound change moving
down from the north eliminated final -p, -t, -k and -s. When final
-s was lost for most dialects the normal way of distinguishing

singular from plural in articles, adjectives, and nouns was lost

as well. Thus the singular article la could no longer be opposed
to the plural article las, except when a vowel followed. In most
cases, radical changes in the plural forms preserved the singular-
plural distinction. The Atlas Linguistique de France shows that
in one area of south-central France, near the southern limit for
the loss of final -s, another sound change was taking place: un-
stressed a was changing to o. Normally, this change took place
in both singular and plural forms. But in a sizeable subregion to
the northward, where -s was lost, this sound change was differen-
tiated so that a changed to o only in the singular, opposing lo
(singular) to la (plural).
This recent finding of Penelope Eckert (1969) is of the greatest
importance in showing how grammatical functions can directly
modify a sound change for communicative needs. It is also relevant
for our discussion here that this compensating differentiation of the
The Study of Language in its Social Context 225

a-^> o change only took place where the loss of -s had become a
uniform and categorical rule. In the area of "-s disturbance," there
is no compensating effect on the a—> o rule. This implies that if a

variable rule is regular enough, it provides language learners with

enough information to preserve the basic distinctions and the under-
lying forms.
If a sociolinguistic study of -s were carried out in that area, we

would no doubt have found a variable rule similar to 20. The deletion
of final -s in France was not recorded in any detail by the dialect
geographers. But a similar variable rule in Puerto Rican Spanish has
been studied by Roxana Ma and Eleanor Herasimchuk (Fishman et
al.1968: 689-703). They found regular patterns of variable constraints
similar to those just given for -t,d deletion: (1) a complex pattern
—
of grammatical constraints the plural -s on articles is retained most
often, verbal -s next, and plural -s on nouns the least; (2) a phono-
which deletion occurs
logical constraint in less often before a fol-
lowing vowel than a following consonant; (3) a regular pattern of
style shiftingwhere increased formality disfavors the operation of
the rule. There is an intermediate form [h] which complicates the
rule, but the form of the Puerto Rican s —> h —> <£ rule is probably
quite similar to the French s —» 4>.

We
conclude that variation such as is shown in the deletion of
-t,d clusters is not the product of irregular dialect mixture, but an

inherent and regular property of the system. The status of variable

rules in a grammar may be questioned on another ground: they
involve a fundamental asymmetry between production and percep-
tion. We can argue that the speaker displays his knowledge of the
past tense suffix by simplifying fewer -ed clusters than others (and
presumably preserving -ed more often when it carries a heavy se-
mantic load). But all that the listener has to know is whether or not
the -ed is optional, since he interprets each past signal as it comes.
(There is evidence that listeners can and do react to frequency, but
it would be difficult to connect this capacity with the grammatical

function of the 'past'; such reactions have to do with overall charac-

terization of the speaker.) In any case, it is clear that variable rules
are rules of production. The issue is whether or not the symmetry
of production and perception is a well-founded assumption about
linguistic structure, or even an attainable goal of theory construction.
As attractive as this might seem, there is now solid evidence that
it is an invalid assumption. In repetition tests with BEV speakers
226 SOCIOLINGUISTIC PATTERNS

14 to 17 years old, we find that sentences such as

I asked him if he

did it axed him did he do it. The meaning

are repeated instantly as I

of the sentence is grasped perfectly, but it is produced automatically

—
by BEV rules there is evidently a deep asymmetry between per-
ception (SE and/or BEV) and production (BEV only).
Finally, one may set aside variable rules on the ground that they
are rules of performance. The less said about this "wastebasket" use
of the performance concept the better. For it must be noted that the
great majority of our transformational and phonological rules may
also be characterized as "performance" rules. Extraposition, wh-
attraction, adverbial postposing, etc., are all means of facilitating the
linearization of the phrase structure input, eliminating discon-
tinuitiesand left-hand embedding, coordinating and assimilating
elements to one another so as to make the "performance" of the
sentence that much easier.
The ability ofhuman beings to accept, preserve, and interpret rules
with variable constraints is clearly an important aspect of their
linguistic competence or icmgue. But no one is aware of this com-
petence, and there are no intuitive judgments accessible to reveal
it to us. Instead, naive perception of our own and others' behavior

is usually categorical, 30 and only careful study of language in use

will demonstrate the existence of this capacity to operate with
variable rules.

2. The Deletion of the Copula in BEV

We now turn to a much more complex problem, concerning the
variable appearance of the present copula and auxiliary is BEV.
in
We frequently hear in this dialect such sentences as He wild, and

30. Given a continuous range of frequency in the application of a rule, such as

"dropping the g" in -ing, we observe listeners reacting in a discrete way. Up to a certain
point they do not perceive the speaker "dropping his g's" at all; beyond a certain
point, they perceive him as always doing so. This is equally true with the (th) and
(dh) rules discussed below, and any other well-developed linguistic variable. The same
categorical judgments appear in the perception of others' eating habits ("She eats
like a bird; he never knows when to stop") or housekeeping ("She cleans night and
day; the dust could be that thick . . ."). Whenever there are strong social values
associated with standards of role performance, we tend to get such categorical
perception. But note that even this sharp alteration of judgments requires the observer
to be (unconsciously) sensitive to frequency. We can speculate that each occurrence
of the marked form sets up an unconscious expectation, which becomes overt if
reinforced within a given period of time, but is otherwise extinguished without effect.
The Study of Language in its Social Context 227

She out the game, and He gon' try to get up corresponding to SE

He is wild, She is out of the game, and He is going to try to get up.

These copula-less sentences are similar to those of many languages

like Russian, Hebrew, or Hungarian, which have no present copula;
to Jamaican English or Creole which has the copula only before noun
phrases (and locatives) or to the speech of young children who say
That a lamh, and Mommy busy. The question arises whether or not
a copula is present in the deep structure or higher-level structure
of BEV; and if so, whether it is then deleted as a whole on the
morphological level or by lower-level phonological rules. These are
important questions for both theoretical and applied linguistics, for
they bear on the issue of how dialects of a language differ, and how
they are to be taught. The brief outline of the argument as presented
here is abstracted from a detailed presentation in Labov 1972a: Ch.
3 and Labov et al. 1968:3.4.

a. We first find that there

are no speakers of BEV who always delete
the copula, and none who never do so. Everyone shows some full
forms, some contracted forms, and some zero forms. The regular-
ity of this behavior, and the pattern of the variable constraints
discussed below, show that we are dealing with a variable rule
within the BEV system.
b. There are syntactic positions where deletion never takes place:
in elliptical forms (He is too), after wh-attraction (That's what
he is). In general, we find that wherever standard English can
contract, BEV can delete the copula, and wherever SE cannot
contract, BEV cannot delete.
c. This connection between contraction and deletion makes it nec-
essary to explore the general conditions for English contraction,
analyzing the evidence of our own intuitions within the generative
model. We find that contraction of am, is, are, will, has, have,
had is the removal of a lone initial schwa before a single conso-
nant in a word which contains the tense marker. This process,
which yields He's here, I'm coming, You're there, I'll go, He's got
it, dependent upon rules which delete initial glides, and
etc., is

upon the vowel reduction rule which reduces unstressed lax

vowels to schwa. The vowel reduction rule is in turn dependent
upon the stress assignment rules which are determined by the
syntax of the surface structure as developed by Chomsky and
Halle (1968). The rules for contraction fit in with and confirm by
228 SOCIOLINGUISTIC PATTERNS

independent evidence Chomsky and Halle's construction of the

transformational cycle in English phonology. 31
d. That deletion of the copula is related to contraction is also indi-
cated by our finding that BEV does not delete forms with tense
vowels which cannot be reduced to schwa: be, ain't, can't are
preserved. That deletion is a phonological process is also shown
by the fact that the m
of I'm is rarely deleted: in general, final
nasals are not deleted in BEV.
e. The variable rules which control contraction and deletion in BEV
show a series of variable constraints according to grammatical
environment. The rule is favored if the preceding noun phrase
is a pronoun. The following grammatical environment constrains

the rules in the order (from least favorable to most) predicate noun
phrase; adjectives and locatives; verbs; and the auxiliary gonna
before a verb. If we take the view that contraction operates first,
and deletion removes the lone consonant which remains after
contraction, then we see that these constraints operate in the same
way on both rules, and contraction in BEV will show a pattern
similar to contraction in other English dialects. The fact that there
are two separate rules is indicated by the fact that the quantitative
effects of these grammatical environments are intensified with
deletion: the constraints have applied twice.
f. Although the same grammatical constraints operate upon con-
traction and deletion in BEV, the phonological effect of a pre-
ceding vowel or consonant is reversed. For contraction, the rule
is favored if the subject ends in a vowel; for deletion, if it ends

in a consonant. This reversal matches the phonological difference

between contraction and deletion, for the former is the removal
of a vowel and the latter of a consonant, and in each case the
favored context leads to a CVC structure.
We thus conclude that the basic form of the contraction and
deletion rules for is is:

31. An Zwicky (1970) argues that there are cases

analysis of auxiliary reduction by
of uncontractible is which cannot be accounted for by stress rules. In the critical
or has
cases, the h- of has is said to be deletable and the vowel reduced but not contracted:
Gerda has been to North Dakota as often as Trudi (h)as. It is possible that these
intuitive (and admittedly disputed) data are the product of contrastive stress on the
second subject Trudi. Dependence on stress would be shown by the unreducibility
of has in I don't think anyone should go to North Dakota as often as Trudi has,
where there is no contrastive stress on Trudi.
The Study of Language in its Social Context 229

21 Contraction

a - <»> I <-co°n^> ##[ +1=] z **

<^+£y

22 Deletion

<0>/^i™£>##
V H-cons/ —##
A more general form of the contraction rule shows it operating before
zero or one consonant, CJ; this consonant then generally elimi-
is

nated by 22. But here we are dealing specifically with the contraction
and deletion of is, where the variable constraints are best known.
The preceding constraints show that the rule is favored by a pronoun
subject (which generally ends in a vowel) or by some other noun
phrase ending in a vowel or a glide. This constraint is reversed for
deletion, which is favored by < + cons>. The following constraints
shows that the rule is favored by a following verb, especially if this
verb is a future form (gonna, gon'); if the next element is not a verb,
then the rule is favored if it is not a noun phrase that is, it is a —
locative or adjective. The vertical ordering of the variable constraints
within the angled brackets reflects their weighting, but there is no
relative weighting indicated here between the preceding and fol-
lowing constraints.
The full data to support 21 and 22 are given in Ch. 3 of Labov
1972a. There are some unresolved issues in the contraction rule in
regard to the effect of a following noun phrase when a noun phrase
precedes; in this situation, some BEV groups seem to differ from
others. Otherwise, the constraints are fully independent and are
replicated in each of the vernacular groups that we have studied.
Table 8.4 shows a small part of the pattern: the effect of the follow-
ing grammatical environment on the deletion of is with pronoun
subjects. For each of the three New York peer groups, the frequency
of deletion rises regularly across these four environments. Below
these are figures derived from Wolfram's studies of black speakers
in Detroit: in this case, the most comparable working-class popula-
tion based on individual interviews with adults as well as adoles-
cents. In general, these Detroit speakers follow rules 21 and 22 in
230 SOCIOLINGUISTIC PATTERNS

TABLE 8.4.

THE EFFECT OF FOLLOWING GRAMMATICAL ENVIRONMENT ON DELETION OF

IS IN BEV WITH PRONOUN SUBJECTS

% deletion before
Speech
community Noun phrase Adj, Loc Vb gonna

New York City

Thunderbirds 35% 51% 74% 91%
Cobras 53 77 80 100
Jets 63 72 78 95
Detroit
Working class 37 46 50 79

Source: Labov et al. 1968:3.4, and Wolfram 1969:211.

their treatment of is. Table 8.4 shows that they follow the same
pattern for the four grammatical environments as the New York City
groups.
The contraction and deletion rules 21 and 22 are not simply sum-
maries of the performance of particular groups. They are general
constraints reflecting the linguistic system of BEV speakers in many
areas. Not only do the New York and Detroit samples coincide, but
we also find the same pattern in Washington, as analysis of con-
versations by Loman (1967) shows; in San Francisco, as Mitchell-
Kernan (1969) shows in her detailed study of two adult speakers; in
Los Angeles, as shown in less detail by studies of younger black
children (Legum et al. 1971).
We are now in a position to look more closely at the formal
character of such variable rules. In so doing, I will be following the
formal interpretation developed by Cedergren and Sankoff (1972)
accepting their modifications of the original semi-quantitative inter-
pretation of Labov 1969. 32 We must first re-examine the significance
of the quantity cp, which represents the proportion of cases in which
a rule applies out of all those cases in which it might apply. In earlier

32. The original model presented in Labov 1969 had two related defects. Since it

was additive, a postulate of geometric ordering was necessary to prevent the contri-
butions of individual constraints from totalling more than 1. This type of ordering
was not always found, so that there were cases where the two most important
constraints had roughly the same effect on the rule. It would then follow that an
increase in the effect of one constraint would necessarily involve a decrease in the
effect of another, so that they would not be independent in their action.
The Study of Language in its Social Context 231

discussion, has been used for the output frequency of a rule for
cp

a given sample. But such frequencies cannot reasonably be seen as

an aspect of the rule itself. Rather <p must be seen as the probability
of the rule applying for any given sample with any given configura-
tion of relevant environments. The hypothesis made explicit in
Cedergren and Sankoff's treatment is that each of the variable con-
straints makes an independent contribution to this probability: this
hypothesis is of the greatest importance to linguistic theory, for it
provides the first strong justification for the linguist's assembly of
individual rules into rule schema.
There is a good empirical basis from the study of linguistic change
to see most rules as tending to apply maximally to be generalized —
to all environments and to go to completion in a given environment
(Labov 1972a, Wang 1969, Chen and Hsieh 1971) and to be reordered
to apply to the maximum number of cases (Kiparsky 1968). Within
a given environment, it will be normal for <p = 1; in a variable rule
there will be some factor preventing the rule from applying, so that
<p = 1 — k . If this limiting factor k =
then the rule will not apply 1,

at all. In the cases we are considering k is limited or diminished

by a series of factors which favor the rule: an input probability p ,

and variable constraints such as those in 21 or 22. The effect of such

constraints in limiting k may be symbolized as v 1? v 2 v 3 vn , . . . .

Thus the general formula for the probability of a given variable rule
applying is

23 9 = 1 - k
24 k = (1 - Po ) (1 - vj (1 - v 2 ) ... (1 - v n ).

If a given variable constraint < + fea >

4
is not present then Vj = 0, and
the probability of the rule applying is unaffected. If the feature is

present, then the factor —

v^ operates to diminish the limiting
(1
factor k and so increases the probability of the rule applying in
,

that environment.
This model of Cedergren and Sankoff is founded on the hypothesis
that the variable constraints are independent, and contribute the
same element v to the probability of the rule irrespective of what-
i

ever other constraints are present. They have applied this model to
establish underlying probabilities for the variable constraints on the
contraction of is reflected in Table 6 of Labov 1969 (Labov 1972a,
Ch. 3), using a slightly different form of the contraction rule from
232 SOCIOLINGUISTIC PATTERNS

21. 33Through the statistical method of maximum likelihood they

produced the estimates of 0.27 for the input probability p and five
other estimates for v ranging from 0.13 to 0.90. They are then able
i

to predict the original table with twelve cells (four following and
three preceding environments) with only six parameters. The results
are quite close: in eight cells, the formula predicts the original
numbers of cases in the table: in four cells the prediction differs by
one case.
Cedergren and Sankoff have applied their model to the quanti-
tative investigation of a number of other variable rules in Panama-
nian Spanish and Montreal French. Current work of G. Sankoff,
Cedergren, and their associates in Montreal is carrying the investi-
gation of variable rules to a higher level of accountability. By the
use of estimates of maximum likelihood it is possible to confirm or
reject the hypothesis of independence of variable constraints in any
particular instance, and thus provide crucial evidence for the validity
of the basic linguistic operation of assembling rule schema.
Given a series of linguistic rules

25 a. X^Y/A C
b. X^Y/A D
c. X^Y/B C
d. X-+Y/B D
e. X->Y/ C
f. X^Y/ D
linguists believe that it is economical and revealing to represent all
these possibilities by a single rule schema such as:

26 X^Y/({A))_(C
The general argument of Chomsky and Halle is that the abbreviatory
notations of 26 or similar ones which lead to the maximum economy
give us substantive information on the formal structures available
to the language learner (1968:331). However, arguments based on

The Cedergren and Sankoff demonstration took as variable constraints the

33.

categories under which the data were tabulated. But in terms of underlying features,
several of these overlap. Thus a preceding + Pro] ends with a vowel, so that it is
[

also + V] in the same way as [—Pro] noun phrases which end with a vowel like
[

Joe. The independence of these underlying features should follow from the method,
if indeed they represent the correct analysis.
The Study of Language in its Social Context 233

simplicity have not always been convincing. In the study of linguistic

change in progress we often find rules expanding to yield more
complex forms (Cohen 1970), and it is possible for someone to assert
that no conclusive proof through relative simplicity has yet been
achieved. 34
The demonstration of Cedergren and Sankoff now provides a
crucial link in this argument. If the various subrules assembled in
the schema 21 were to be dissolved into individual rules, this would
be equivalent to stating that the variable constraints are not inde-
pendent, and that it is impossible to arrive at estimates of Vj which
would predict this data. The very existence of stable contributions
to the probability of a rule by variable constraints demonstrates the
validity of the rule schema. The hypothesis of the independence of
variable constraints is equivalent to the hypothesis that rule sche-
mata are significant. But it is only with quantitative data that we
can arrive at convincing demonstrations of this claim.
Beyond the internal evidence and reliability of the contraction and
deletion rules, there is other evidence for the validity of our account.
Subjects have no difficulty in repeating back the copula in imitation
tests (Labov et al. 1968:3.9, Labov 1972a, Ch. 2), and they show very
few problems in comprehension tests for this element. In these
respects, the copula contrasts sharply with third singular -s, which
does not correspond to any element in the grammatical structure of
BEV. 35
So far, we have said nothing about the input probability p This .

section is concerned with problems of internal linguistic structure,

but it should be evident from Table 8.3 that the Detroit speakers are

34. See the introduction to Lakoff (to appear), where he argues that there has not

been any successful explanation in syntax through simplicity or an internal evalua-

tion measure.

35. Torrey's experimental approach to the comprehension of syntax used pictures

which illustrated the difference between past and present or singular and plural. She
tested black second-graders for their ability to use various inflections in speech and
in comprehension, before and after a training period. The plural operated at a high
level from the beginning, and the copula responded rapidly to practice. But in the
case of third singular -s there was a sharp difference between two situations. The
second-graders had no difficulty in using -s to distinguish present from past as in
The man hits the dog vs. The man hit the dog. But they showed no success at all,
before or after training, in using -s to distinguish singular from plural as inThe cat
splashes vs. The cats splash. We can conclude that they have a rule which states
roughly, "Insert -s on present verbs in formal situations." (1972)
234 SOCIOLINGUISTIC PATTERNS

operating with a lower p , which is generally sensitive to such

sociolinguistic parameters as age, group memberships, and socio-

economic status. These variables operate directly upon p in our
representation and yield functions which can be isolated from the
more stable internal constraints upon linguistic rules. Such socio-
linguistic variables are the topic of Sect. 3 below.

3. Negative Concord
For the third example of the analysis of linguistic structure in its
social context, I will consider the problem which revolves around

the sentence:

27 It ain't no cat can't get in no coop.

This was said by Speedy, the leader of the Cobras in one of our
group sessions, in a discussion about pigeon coops. Speakers of any
other dialect of English besides BEV interpret 27 as meaning 'There
is no cat that can't get into any coop.' They are more than a little

surprised to discover from the context that Speedy was denying that
cats were a problem, and that his meaning was 'There isn't any cat
that can get into any coop.' At first glance, it seems that BEV speakers
are behaving in an illogical, contradictory way. do not If dialects
differ radically in their deep structure, as suggested above, how can
it be that a negative in one is a positive in another?

First, one might ask if Speedy had simply made a mistake. This
is not the case, for we have encountered a half dozen other examples

of the same construction in our work in the black community. Most

convincing is the example from a long epic poem of black folklore:
speaking of a whore, the narrator says, There wasn't a son of a gun
who this whore couldn't shun, meaning that she was so good that
'there wasn't any customer that could shun her'.
We noticed that the examples normally had three features in
common: (a) there were two clauses, and a contradictory negative
appeared in the second; (b) there was another negative in the first
clause, and (c) the first clause also contained an indeterminate adverb
such as ever or any (negative plus any = no). These three facts lead
us to connect the phenomenon with "negative concord," the process
by which negatives are attracted to indeterminates. In this investi-
gation, we again found it necessary to develop the argument further
in terms of our grammatical intuitions (as native speakers of standard
English and of several nonstandard white dialects). We can begin
The Study of Language in its Social Context 235

with the generative formulation provided by Klima (1964) of the

observation of Jespersen and others that in English the negative is
obligatorily attracted to the first indeterminate if it precedes the verb,
and optionally to the first indeterminate thereafter. Thus in place
of * Anybody doesn't sit there, we have by obligatory rule, Nobody
sits there. On the other hand, it is merely an optional (and stylistically
somewhat formal) rule which shifts He doesn't sit anywhere to He
sitsnowhere. If we continued to consider only standard English, we
might write a rule which takes the initial negative as a feature of
the sentence and distributes it directly to the various positions with
the conditions indicated. But study of a variety of English dialects
leads us to the conclusion that the negative attraction rule which
incorporates the negative with the first indeterminate is of a very
different character from the others. seems there are three distinct
It

rules, all operating after the negative is placed in its normal preverbal
position, which we can sketch in broad outline here. The first rule
is the categorical one of negative attraction:

28 NEGATTRAC (obligatory for all dialects)

Indet— X-Neg
1 2 3 -> 3 + 1 2

Not only is this rule obligatory for all dialects, but sentences where
it has not applied, such as * Anybody doesn't sit there are un-English
in a very striking way; in repetition tests (Labov et al. 1968:3.9)
sentences like this provoke only confusion and no one can repeat
them. There are conditions which suspend the obligatory force of
this rule, such as a preceding hypothetical or negative; an exact
statement requires an extended discussion (see Labov 1972a, Ch. 4).
But the detailed investigation of NEGATTRAC leads us to the con-
clusion that this rule reflects a cognitive requirement for the orga-
nization of the features of the indeterminates in relation to the
negative. The general condition is that universal quantifiers may not
precede a simplex negative predicate if they contain the features
[ + DISTRIBUTIVE] as with each and absolutely not if they also
contain [ — FACT] as with any.
The next two rules are of a different character:
29 Negative Postposing (optional; Standard Literary English only)

Neg - X - Indef
1 2 3 ->2 1 + 3
236 SOCIOLINGUISTIC PATTERNS

30 NEGCONCORD (optional; for nonstandard dialects only)

Neg - X - Indef
1 2 3 ->1 2 1 + 3

These two rules are complementary and perform the same em-
phatic function. Instead of He doesn't sit anywhere, the first rule
gives us He sits nowhere, and the second pleonastic rule yields He
don't sit nowhere. Rule 30 applies without regard to clause bound-
aries: thus we can have He don't like nobody that went to no prep
school = SE He doesn't like anybody that went to any prep school.
There are also some nonstandard white dialects [WNS 2 ] which can
transfer the negative back to preverbal position, so that we have
* Anybody doesn't sit there — Nobody
> sits there — Nobody
> don't sit

there. BEV shares this property.

Careful grammatical searching of our interviews and group ses-
sions now shows us that BEV differs surprisingly in one other way
from other dialects. Rule 30 is not variable, but obligatory for BEV
within the same clause. For core members of the BEV peer groups,
we find that negative concord operates not at a 95 or 98 percent level,
but at 100 percent, in 42 out of 42 cases, 63 out of 63, etc., whereas
our corresponding white groups show inherent variation. This means
that the emphatic function of negative concord is entirely lost for
BEV: if the rule is obligatory, it has no stylistic or contrastive sig-
nificance. This emphatic function is supplied by BEV through several
extensions of the negative rules not used by Northern dialects, such
as negative inversion: Nobody can do it^ Ain't nobody can do it.
Nobody saw him —> Didn't nobody see him. BEV also extends rule
30 to permit transferring the negative to the preverbal position in
a following clause. It is this emphatic usage which yields 27, It ain't
no cat can't get in no coop.
We see finally that the apparent "contradiction" of 27 is not a

difference in logical operations among dialects, but only a slight

readjustment of the conditions on a transformation. We may rewrite
30 now as

30' NEGCONCORD
(Verb)
Neg- X
I Indef J

1 3-> 1
The Study of Language in its Social Context 237

and set up a table of conditions as follows. This table is in the form

of values of <p, where means the rule never applies, X means it

is a variable rule with < <<p 1, and 1 means the rule is obligatory.

3 = Indeterminate Verb

1 and 3 clause mates? Yes No Yes No

Dialect
SE
WNSi X X
WNS 2 X X x
BEV 1 X X X

We here make use of the distinction between a variable rule and

an independent obligatory one in a new way: the variable rule has
a communicative function
—
'stylistic', 'expressive' or 'emphatic'

value in this case, while the invariant rule has none, it merely
facilitates the expression of choices already made. Once again, a
structural compensation appears as a variable rule becomes invariant
and information is lost: BEV extends negative concord to new envi-
ronments to supply this loss. BEV thus has the properties of a sepa-
rate subsystem, in that changes in one part of the system seem to
be inevitably accompanied with compensating changes in another
to maintain the same functional operation.

3. Sociolinguistic Structure

We may define a sociolinguistic variable as one which is correlated

with some nonlinguistic variable of the social context: of the speaker,
the addressee, the audience, the setting, etc. Some linguistic features
(which we will call indicators) show a regular distribution over
socioeconomic, ethnic, or age groups, but are used by each individual
in more or less the same way in any context. If the social contexts
concerned can be ordered in some kind of hierarchy (like socio-
economic or age groups), these indicators can be said to be stratified.
More highly developed sociolinguistic variables (which we will call
markers) not only show social distribution, but also stylistic differ-
entiation. As noted in Sect. 1, stylistic context can be ordered along
a single dimension according to the amount of attention paid to
speech, so that we have stylistic as well as social stratification. Early
studies such as those of Fischer (1958) or Kucera (1961) observed
238 SOCIOLINGUISTIC PATTERNS

linguistic variables only one dimension at a time, but more recent

studies (Labov 1966a; Wolfram 1969; Anshen 1969; Trudgill 1971) look
at the interrelation of both dimensions.

A Stable Sociolinguistic Marker: fingj.

One most general sociolinguistic markers in English is (ing):

of the
the presence or absence of a final velar for unstressed /ing/. The
formal variant is always the velar, with some interesting exceptions
in Southern areas. 36 There are a number of technical questions in
the definition of this variable which have not always been given the
attention they deserve. A naive approach focuses on the suffix -ing
as if the variable is an alternation in the shape of this morpheme,
neglecting unstressed -ing in something and nothing. In many
Southern dialects (e.g., Atlanta) the trisyllables anything and every-
thing are exempted from the rule and are always realized with the
[irj] variant: probably the result of tertiary stress in the last syl-

lable. On the other hand some dialects (Eastern New England)

realize [inj with tertiary stress in workin' [w^kin], fishin' [fijin].
Further complications can follow as some dialects differentiate
nouns from participles from progressive forms, so that inflectional
and derivational boundaries, # and + will appear as variable ,

constraints in the rule. In the most extreme Southern treatment of

this variable, all proper nouns are included as well, such as Manning
[me:nm].
Despite these variations in the selectional process the operation
of the variable itselfis extraordinarily uniform. Even if we do not

allow for the details noted above, we will obtain a regular pattern
of stylistic and social stratification quite similar to that presented
for (th) in Chs. 4 and 5. Fischer's early observations (1958) of (ing)
showed that this variable reflected sensitivity to sex, formality and
cultural orientation toward school. Fig. 8.3 shows the (ing) pattern
of the Lower East Side study (Labov 1966a: Ch. 10). Similar patterns
are obtained by other investigators for well-established sociolin-
guistic variables such as (ing), (th), (eh), and negative concord; these
patterns share a number of common properties with Fig. 8.3:

36. A study of English in the Austin, Texas area by Stanley Legum (pers. comm.)
showed an unusual reversal of the normal (ing) pattern. A number of subjects used
the velar variant [in] in the interview situation, but switched to [in] in the more formal

context of reading style. We seem to be dealing with a regional norm which elevates
the [in] variant as a symbol of local identity, recognized even by those who do not
use it in their own connected speech.
 The Study of Language in its Social Context 239

(ing) 40
Index

Fig. 8.3. Class and style stratification of (ing) in working, Jiving,

etc., New York City adults. Socioeconomic class scale:
for white
0-2, 3-6, 7-8, 9. A, casual speech; B, careful speech; C, reading
style.

a. In every context, members of the speech community are differ-

entiated by their use of (ing) so that higher and lower scores for
this variable are directly correlated with higher and lower posi-
tions on socioeconomic indices.
b. Yet every group is behaving in the same way, as indicated in the
parallel slope of style shifting of (ing) so that higher and lower
scores for this variable are directly correlated with higher and
lower positions on a scale of formality of the context.
240 SOCIOLINGUISTIC PATTERNS

c. Since Fig. 8.3 is not visible as a whole to members, facts (a) and
(b) are not part of general knowledge. The portion of Fig. 8.3
visible to any given individual is usually one vertical and one
horizontal section: the range of style shifting used by his own
group, and the stratified behavior of other groups in the few
contexts where he interacts with them. He is not aware that others
shift in the same way he does.
d. The same sociolinguistic variable used to signal social and
is

stylistic stratification. It may therefore

be difficult to interpret any
signalby itself — to distinguish, for example a casual salesman
from a careful pipefitter. 37
e. Although it is impossible to predict for any one utterance which
variant a speaker will use, the striking regularity of Fig. 8.3
emerges from samples with as few as five individuals in one
subgroup, and no more than five or ten utterances in a given style
for each individual.

The fact that (ing) is a monotonic function of social class and

contextual style is of great importance in characterizing it as a stable
sociolinguistic variable. The rule which governs this variable may
be stated as 31.

31 cont + VOC r -!

/^ — cons r + nas|
<0>/ -cons
/
tense
ant — stress
|_
+ cQr J

We are not concerned here with any variable constraints which may
influence this rule, but rather with the determination of the input
variable p The two major determinants of this quantity are socio-
.

economic class and contextual style: age, sex, and ethnic group play
a minor part. For (ing) and the other stable sociolinguistic variables,
the function takes this general form:

32 p = a • (SEC) + b • (Style) + c

The suggestion of 32 that this is a linear function goes beyond

37. This is one of the most striking findings of sociolinguistic research, since essays

about social usage, written from "common-sense" knowledge, have tried to distinguish
"functional varieties" and "cultural levels" as completely independent dimensions. But
their interdependence is shown in this and every other careful empirical study to
date. Though it may seem inconvenient to have one variable operate on both di-
mensions, it seems to be an inevitable result of the sociolinguistic processes involving
attention to speech and perception of norms, as outlined below.
The Study of Language in its Social Context 241

available data, since we

cannot yet quantify the dimension of style.
It may be possible to do this in future studieswhich develop the
notion that styles are organized by the amount of attention paid to
speech, but at the moment this lack of quantification is a serious
limitation.
The function 32 does state concisely the basic information stated
under (a) and (b) above: that stable sociolinguistic markers organize
linguistic variation in a way that is directly parallel to other hier-
archical indices of social status. In Ch. 9 we will see that they
contrast sharply in this respect with sociolinguistic variables reflect-
ing change in progress.
The five general traits a-e hold for a number of sociolinguistic
markers that have been studied in the research cited above. A com-
plete view of social and stylistic stratification is not available for
most of these studies: some provide data only on relatively small
sections of the pattern of Fig. 8.3 and its equivalents, while others
cover a wider range. But all of these data can be interpreted in terms
of this configuration and fitted into the pattern. Many studies have
demonstrated this for (th) and (dh), among white and black popula-
tions (Labov 1966a; Labov et al. 1968; Anshen 1969). Negative concord
and pronominal apposition have been studied by Shuy, Wolfram,
and Riley for Detroit (1967). A number of Spanish variables have
been studied by Ma and Herasimchuk for Puerto Rican speakers:
the most intricate and systematic variable is (S), the aspiration and
deletion of syllable-final /s/, which is generally of great importance
throughout the Spanish-speaking world. Ma and Herasimchuk give
a good view of stylistic stratification of this variable. Cedergren
studied (S) in her work on Panamanian Spanish, where she found
a linear correlation with socioeconomic status. The same may be
said for (R), the devoicing, fricativization, pharyngealization and
deletion of syllable-final /r/ (see Ch. 9 below). Sankoff and her
associates at Montreal are beginning to produce detailed reports on
a number of French variables which show the pattern of Fig. 8.3,
including the deletion of (1), presence or absence of (que) and many
others. Trudgill (1971) has studied a number of variables in the vowel
system of Norwich: again, the contrast between stable sociolinguistic
variables and on-going changes is of the greatest importance. The
generality and stability of the (ing) variable is reflected again in his
data, drawn from an area quite remote from those where this variable
was originally located. Figure 8.4 shows the linear pattern of (ing)
242 SOCIOLINGUISTIC PATTERNS

100

Class and style stratification of (ing) in Norwich (adapted

Fig. 8.4.
from Trudgill 1971). A, casual speech; B, careful speech; C,
reading style; D, word lists.

in Norwich, which matches all essential features of the pattern of

Fig. 8.3. The preserved intact. In addition we
five properties a-e are
can see that Fig. 8.4 is a good example of sharp stratification as
opposed to fine stratification. In this respect Norwich (ing) is more
similar to New York City (th) in Fig. 4.1, where working-class and
middle-class groups are sharply separated. In contrast, the pattern
of New York City (r) and (ing) shows a stepwise gradient relationship
where there is no clear discontinuity between one class group and
another.
The Study of Language in its Social Context 243

A superficial view of this regularity would lead one to conclude

that further sociolinguistic studies can do no more than uncover
more linear correlations. But each of these studies has led to deeper
insights into the relations of internal and external determinants of
language structure and change. Without a base line of stable socio-
linguistic markers, there is no basis for investigating more abstract
questions: the contrast between change and stability; between
phonological and grammatical features, between fine and sharp
stratification, or between abstract higher-level rules and low-level
phonetic patterns; the role of referential function vs. expressive
sociolinguistic information; the interaction of sex with social class
and ethnic status; the hypercorrect pattern of the second-highest
status group; and many other questions all presuppose that we have
laid out the basic sociolinguistic orientation of the community with
stable sociolinguistic markers.

Men vs. Women

There is a regular aspect of the social stratification of stable
variables is not shown on Fig. 8.3. In careful speech, women
which
use fewer stigmatized forms than men, (Labov 1966a:288), and are
more sensitive than men to the prestige pattern. They show this in
a sharper slope of style shifting, especially at the more formal end
of the spectrum. This observation is confirmed innumerable times,
in Fischer (1958), throughout Shuy and Fasold's work in Detroit, in
Levine and Crockett, and in Anshen's study of Hillsboro. The pattern
is particularly marked in lower-middle-class women, who show the

most extreme form of this behavior. In the example of (ing), we find

the usual pattern of sex differentiation and in Trudgill's study of
Norwich, women have lower values of (ing) for almost all styles and
social classes. In Fig. 8.4we have indicated this differentiation for
only one social Here the (ing) pattern for lower-middle-class
class.
women shows a smaller use of the nonstandard form than that of
the upper-middle class as a whole. Except for a small percentage
of [in] forms in casual speech, middle-class female speakers use the
standard [in] form exclusively. Here as elsewhere, it is clear that
women are more sensitive than men to overt sociolinguistic values.
Even when women use the most extreme forms of an advancing
sociolinguistic variable in their casual speech (Ch. 6), they correct
more sharply than men formal contexts. Chapter 9 will examine
in
in greater detail this pattern of sex differentiation in relation to the
process of linguistic change.
244 SOCIOUNGUISTIC PATTERNS

The Hypercorrect Pattern of the Lower Middle Class

One of the most solidly established phenomena of sociolinguistic

behavior is that the second-highest status group shows the most
extreme style shifting, going beyond that of the highest-status group
in this respect. Ch. 5 has presented the pattern in great detail, draw-
ing upon the data of the Lower East Side study of New York City.
Given this pattern, we can find parallel phenomena throughout all
the literature cited above. Shuy, Wolfram, and Riley (1967) and
Wolfram (1969) show many cases where the greatest style shifting
is exhibited by the second-highest status group. Most of these are
stable stigmatized linguistic features. The most clearcut examples
of the hypercorrect pattern are to befound in on-going change from
above, as with the importation of the new r-pronouncing norm
throughout the previously r-less areas of the Eastern United States.
New York City gave us
Fig. 4.2 for the class stratification of (r) in
the classic view of the crossover pattern. A striking corroboration
of this pattern is supplied by Levine and Crockett's study of (r) in
Hillsboro, N.C. The Hillsboro shift of (r) shown in Table 9.7 (p. 292)
is a stylistic shift between sentences in which (r) is not the main

focus, and word lists where it is (see above).

Here data from a completely independent study with a more
limited stylistic range shows the same crossover phenomenon as in
—
New York City. The second-highest status group in this case, high-
school graduates, shows a much greater shift towards the prestige
norm in the more formal style. The significance of this pattern for
the mechanism of linguistic change has been dealt with in Chs. 5
and 7 and will be considered in the more general context of Ch. 9.
The generality of our principle that "the second-highest status
group" shows the most extreme style shifting is challenged by the
Norwich data in Fig. 8.4, where the third-highest status group upper —
working class— shows the greatest shift. Between careful speech and
reading style the UWC
moves from (ing)-74 to (ing)-15. However, this
exception holds only for men; as far as women are concerned, it is
again the lower middle class which shows the sharpest slope, and
the correction shown by this group is indeed sharper than any other
group. The dotted line in Fig. 8.4 shows (ing) values for lower-
middle-class females who move from (ing)-67 in casual speech to
(ing)-03 in careful speech. From this point on they merge with the
upper middle class as far as the use of (ing) is concerned.
The Study of Language in its Social Context 245

Granted the stability and generality of this complex pattern, it will

be interesting to see if any formal simplification can be achieved.
The crossover pattern may be represented abstractly as:

The slope of style shifting is indeed complex. The highest and lowest
group have the shallowest slope; the interior groups follow behind
the lead of the second-highest group, which is the steepest. How can
this be formalized? The rule for the vocalization of (r) in the white
community has the general form:
33 [ + cen] -» < — cons) / [
— cons] [
— syl]
That is, a central segment loses its consonantal feature variably in
postvocalic position if the following element is not a vowel. The

problem here is to write a function for the basic constraint on the

input variable p comparable to the simple and straightforward
function 32. The solution lies in an understanding of the ultimate
significance of style shifting: it is governed by the recognition of an
exterior standard of correctness. The strength of such behavior can
be measured by the Index of Linguistic Insecurity. Table 5.1 shows
us the curvilinear pattern, with the lower middle class at a maximum
which we need to describe the slopes of style shifting shown above.
We can then write for 33 the function
34 p = a • SEC + b • (ILI) (Style) + c

Problems of Sociolinguistic Structure

As we have noted above, the most immediate problem to be solved
in the attack on sociolinguistic structure is the quantification of the
dimension of style. If quantitative studies of attention can be related
to style shifting, we will then be able to give more precise form to
rules such as 32 or 34 and specify the constants a, b, and c. Such
quantification may possibly be obtained by studies of pupil dilation,
246 SOCIOLINGUISTIC PATTERNS

or of systematic divisions of attention through mechanical and

measurable tests, or by quantitatively reducing audio-monitoring
through noise level.
It is also evident that many studies cited do not have enough data

from the direct study of the vernacular. The methodological task is

to combine surveys of individuals who give us a representative
sample with longer-term studies of groups. The ideal study of a
community would randomly locate individuals, and then study
several groups of which that individual was a member. That is quite
impossible in a normal social survey, given the numbers required,
but since we have established that sociolinguistic studies require a
smaller population to begin with, such a model is not beyond the
realm of possibility.
A third problem lies in dealing with rules which show irregular
lexical distribution. There is now good evidence that the course of
linguistic change involves the temporary dissolution of word
classes. 38 The most difficult problem here is that there are distribu-
tions across word classes which we would want to describe, but
which are not likely to be a part of the knowledge of the native
speaker. For example, only a certain proportion of English verbs with
Latinate prefixes show when
they appear as nouns
a shift of stress
like convict [V]: convict [N]; others retain end
stress, like consent
[V]: consent [N]. It can be shown that the proportion of words in
any given subclass is related to the length of the prefix, but this
regularity is of no use to the native speaker since most words have
a fixed accent. As another example, the tensing rule for short a in
New York City does not normally operate in CV environments,
though there are a number of exceptions. The linguist is interested
to discover that most of these exceptions have a sibilant as the medial
consonant C. But in such cases, the native speaker again only needs
to know in what class a given word falls. The proportion of the

38. Although Figs. 4.1 and 4.2 show word classes moving as a whole, we have

encountered some rules which show a great deal of irregular lexical variation. The
tensing of short a in bad, ask, etc., investigated in New York City by Paul Cohen
(1970),shows such irregularity, while the raising rule which follows the tensing rule
does not (Labov 1966a:51-52). It is the existence of a variable rule which allows the
word class to be reconstituted when the change is completed, since it is defined as
the class of lexical items which can vary between X and Y, as opposed to the classes
which are always X some structural causes of such lexical variation,
or always Y. For
see Wang 1969.
The Study of Language in its Social Context 247

original word class which has been affected by the incoming rule
is no immediate interest
of to him if he has no choice in the pronun-
ciation of any given item. It may be that we will enter rules into

our grammar which are not a part of the "knowledge" of native

speakers. This particular metaphor may have lost its value at this
point in our investigations.
A fourth major challenge is to enter more deeply into the study
of higher level syntactic variables, such as extraposition, nominaliza-
tion, placement of complementizers, negative raising, wh-attachment,
or relativization. The two chief stumbling blocks to investigating
these features in their social context is the low frequency of occur-
rence of the critical subcases, and the lack of certainty in our abstract
analyses. But some beginning has been made in our recent work in
urban ghetto areas, and the challenges to work with more abstract
matters cannot be ignored. The study of language in its social context
cannot remain at the level of such phonological variables as (ing),
if it is to make a significant contribution to the problems outlined

at the beginning of this chapter.

The fifth problem is to enlarge the scope of these studies beyond
individual speech communities, and relate them to larger grammars
of the English speech community as a whole. The discussion of
negative concord in Sect. 2 indicates one way in which this might
be done. The work of C. J. Bailey is most challenging here: particu-
larly his penetrating studies of phonological rules in Southern dia-
lects (1969a), and his broader attempts to incorporate all English
phonology into a single, pandialectal set of rules (1969b). Though
these studies of Bailey are not based upon the study of language in
context, one must eventually hope to provide reliable data to support
work of this generality and this level of abstraction.

The Relation of Norms to Behavior

So far, in our consideration of sociolinguistic structure, we have
taken into account only what people say, and only incidentally what
they think they should say. These are the "secondary responses" to
language that Bloomfield suggested that we might well observe (1944)
as one part of popular lore. There is a very small vocabulary avail-
able to most people for talking about language: the same few terms
recur as we hear that the other people's pronunciation has a "nasal
twang," is "sing-song," is "harsh" or "guttural," "lazy" or "sloppy."
Grammar is said to be "mixed-up" or "illogical."
248 SOCIOLINGUISTIC PATTERNS

A small number of sociolinguistic markers rise to overt social

consciousness, and become stereotypes. There may or may not be
a fixed relation between such stereotypes and actual usage. The
variables (ing) and (dh) are such stereotypes in the United States:
someone may be said to "drop his g's" or to be one of those "dese,
dem and dose guys". Most communities have local stereotypes, such
as "Brooklynese" in New York City which focuses on "thoity-thoid"
for thirty-third; in Boston, the fronted broad a in "cah" and "pahk"
receives a great deal of attention. Speakers of the isolated Cape
Hatteras (North Carolina) dialect are known as "hoi toiders" because
of the backing and rounding of the nucleus in high, tide, etc.
Such social stereotypes yield a sketchy and unsystematic view of
linguistic structure to say the least. In general, we can assert that
overt social correction of speech extremely irregular, focusing on
is

the most frequent lexical items, while the actual course of linguistic
evolution, which has produced the marked form of these variables,
is highly systematic. This is the basic reason why the vernacular,

in which minimum attention is paid to speech, gives us the most

systematic view of linguistic structure. For example, the evolution
of the New York City vernacular has led to the raising of the vowel
in off, lost, shore, more, etc. until it has merged with the vowel of
sure and moor. This high vowel has been stigmatized, and is now
being corrected irregularly by middle-class speakers. But the same
vowel, raised simultaneously in the nucleus of boy, toy, etc., is never
corrected. 39
But subjective reactions to speech are not confined to the few
stereotypes that have risen to social consciousness. Unconscious
social judgments about language can be measured by techniques
such as Lambert's "matched guise" test, and others described in Sect.
1. One basic principle emerges: that social attitudes towards lan-

guage are extremely uniform throughout a speech community. 40

39. We also find that the vowels of my

and mouth are affected by the rotation of
the long and ingliding vowels of bad, bar, lost. As bar moves to the back, my moves

with it, and mouth moves in the opposite direction towards the front. But of all these
systematically interrelated changes, only the raising of bad and Jost shows style
shifting and correction. Even for these cases, the correction is lexically irregular.
40. In fact, it seems plausible to define a speech community as a group of speakers
who share a set of social attitudes towards language. In New York City, those raised
out of town in their formative years show none of the regular pattern of subjective
reactions characteristic of natives where a New York City variable such as the vowel
of lost is concerned (Labov 1966a:651).
The Study of Language in its Social Context 249

Lambert's studies show, for example, that the negative attitude

towards Canadian French is not only quite uniform in the English-
speaking community, but almost as unanimously held among French
speakers in Quebec (1967). In our study of unconscious subjective
reactions to markers such as (r), we find the most extraordinary
unanimity in speaker's reactions, despite the great variation in the
use of [r] just described. There is a general axiom of sociolinguistic
structure which can be stated as: the correlate of regular stratifica-
tion of a sociolinguistic variable in behavior is uniform agreement
in subjective reactions towards that variable. This principle was
documented in Ch. 6, where we saw in Fig. 6.2 that the development
of social differentiation of (r) among younger speakers corresponded
to the uniform positive evaluation of (r-1).
As we reexamine the structures shown in Figs. 4.1 and 4.2, it is

apparent that the uniform slope of style shifting also reflects the
uniform attitudes held in the community. But for a stable socio-
linguistic marker like (ing) or (th), we can raise the question, what
maintains this structure for such a long period of time? Why don't
all people speak in the way that they obviously believe they should?

The usual response is to cite laziness, lack of concern, or isolation

from the prestige norm. But there is no foundation for the notion
that stigmatized vernacular forms are easier to pronounce 41 and ;

there is strong evidence of concern with speech in large cities.

Careful consideration of this difficult problem has led us to posit
the existence of an opposing set of covert norms, which attribute
positive values to the vernacular. In most formal situations in urban
areas, such as an interview or a psycholinguistic test, these norms
are extremely difficult to elicit. Middle-class values are so dominant
in these contexts that most subjects cannot perceive any opposing
values, no matter how strongly they may influence behavior in other
situations. In our recent work in the black community, we have been
able to uncover evidence of the existence of such opposing norms.
Fig. 8.5 shows responses to the first two items on our subjective
reaction test, opposing a working-class speaker to a middle-class
speaker on "zero" sentences (which contain none of the variables
to be tested). The upper line shows the percent of those who rated

41. Someof the extreme developments of vernacular vowel shifts in New York City,
Detroit, orChicago are tense vowels which seem to involve a great deal of muscular
effort compared to the standard. Spectrographs analysis indicates that such vowels
as short /a/ rising to the height of here are extremely fronted.
 250 SOCIOLINGUISTIC PATTERNS

Job

Friend

40
Fight

20

No. So. No. So.

Middle Upper Lower
Class working working
Class Class

Fig. 8.5. Percent rating middle-class speaker

higher than
working-class speaker on 3 scales for 5 social groups, from
Labov et al. 1968:242.

the middle-class speaker higher on the scale of "job-suitability." It

begins very high with middle-class subjects, and falls off slightly as
we move to lower socioeconomic groups. The lower line is the
converse: this registers reactions to the "fight" or "toughness" scale:
"If thespeaker was in a street fight, how likely would he be to come
out on top?" There is a simple inverse relationship here: a stereotype

that is probably reinforced by school teachers but also shows some

recognition of social reality. But the third set of reactions to the
"friendship scale" shows that there is more involved. This scale is
in response to the question "If you knew the speaker for a long time,
how likely would he be to become a good friend of yours?" For the
three upper social groups, this follows the job scale closely; but for
the lower working class, it switches abruptly, and follows the light
scale. The same phenomenon can be observed for a whole range of
variables tested (Labov et al. 1968:3.6).
We have therefore some empirical support in positing the opposi-
tion between two sets of values as the normative correlate of stable
sociolinguistic markers such as (th) and (ing). In this type of study,
we agree with Homans (1955) that the proper object of study should
The Study of Language in its Social Context 251

not be behavior alone, or norms alone, but rather the extent to which
(and the rules by which) people deviate from the explicit norms that
they hold. It is at this level of abstraction that we can best develop
linguistic and sociolinguistic theory.

The Role of Social Factors in Linguistic Change

Although this chapter is not primarily concerned with the prob-

lems of language change, we have already introduced some data
which bear on this question. In speaking of the role of social factors
influencing linguistic evolution, it is important not to overestimate
the amount of contact or overlap between
and the social values
structure of language. Linguistic and by no means
social structure are
coextensive. The great majority of linguistic rules are quite remote
from any social value; they are part of the elaborate machinery
which the speaker needs to translate his complex set of meanings
or intentions into linear form. For example, the rules governing the
quantifiers and negation discussed above are well below the level
of social affect, and their irregular, idiosyncratic distribution in the
population reflects this fact.
Variables closer to surface structure frequently are the focus of
social affect. In fact, social values are attributed to linguistic rules
only when there is variation. Speakers do not readily accept the fact
that two different expressions actually "mean the same" and there
42
is a strong tendency to attribute different meanings to them. If a

certain group of speakers uses a particular variant, then the social

values attributed to that group will be transferred to that linguistic
variant. Sturtevant (1947) has proposed a general model of linguistic
change showing the opposition of two forms, each favored by a
particular social group. When the issue is resolved, and one form
becomes universal, the social value attached to it disappears.
As far as the synchronic aspect of language structure is concerned,
it would be an error to put much emphasis on social factors. Gen-

42. When New York City cruller (Dutch kroeller) was replaced by the standard term
doughnut, the term cruller was variously assigned to other forms of pastry. Similarly
the local pot cheese (Dutch pot kees) was replaced by cottage cheese and was
differentiated to indicate a drier form. The oscillation of socially marked pronun-
ciations of vase led one informant to say, "These small ones are my [veziz] but these
big ones are my [vaziz]."
252 SOCIOLINGUISTIC PATTERNS

erative grammar has made great progress in working out the invari-
ant relations within this structure, even though it wholly neglects
the social context of language. But it now seems clear that one cannot
make any major advance towards understanding the mechanism of
linguistic change without serious study of the social factors which
motivate linguistic evolution. Ch. 7 outlined a proposal for the basic
mechanism by which social factors interact with internal, linguistic
factors; Ch. 9 will review a much broader field and take as primary
focus the range of social variables which directly influence the
course of linguistic evolution.

4. Some Invariant Rules of Discourse Analysis

This presentation has so far concentrated almost entirely upon the

variable rules of language: their use in providing decisive evidence
on questions of linguistic structure, their place in sociolinguistic
structure, and more briefly, their role in the evolution of language.
But a very great number of linguistic rules are not variable in the
least:they are categorical rules which, given the proper input, always
apply. More than any other field concerned with human behavior,
linguistics has succeeded in isolating the invariant structures under-
lying the surface phenomena, and it is upon this achievement that
we have been building in the work outlined in Sects. 2 and 3. The
formal representation of variable rules presented there depends
upon, and interlocks with, a number of invariant rules of grammar
derived from studies of language quite apart from any social context.
There are some areas of linguistic analysis in which even the first
steps towards the basic, invariant rules cannot be taken unless the
social context of the speech event is considered. The most striking
examples are The fundamental problem
in the analysis of discourse.
of discourse analysis show how one utterance follows another
is to
in a rational, rule-governed manner — in other words, how we un-
derstand coherent discourse. We rely upon our intuitions to distin-
guish coherent from incoherent discourse; for example, the following
is plainly not governed by any rules that we can immediately recog-

nize (from Laffal, 1965:85).

35 A: Whatyour name?
is

say you might have thought you had some-

B: Well, let's
thing from before, but you haven't got it any more.
A: I'm going to call you Dean.
The Study of Language in its Social Context 253

This is an excerpt from a conversation between a doctor and a

schizophrenic patient. Our first data in dealing with such a passage
will be our intuitive reactions to it, and the first challenge in dis-
course analysis is to account for our intuitions (as confirmed by the
response of participants as in 35). The question is: how much and
what kind of data do we need in order to form sound judgments
and interpret sequences of utterances as the participants in the
conversation do? The simplest case is that of elliptical responses,
as in 36.

36 A: Are you going to work tomorrow?

B: Yes.

Here our normal knowledge of English syntax is sufficient to allow

us to derive B's utterance from Yes, I am

going to work tomorrow.
There is a simple rule of discourse of the following form:

37 If A utters a question of the form Q-S x and B responds ,

with an existential E (including yes, no, probabiy, maybe,

etc.), then B is heard as answering A with the statement

E-S r
But now let us consider sequences of the following form:

38 A: She never helps at home.

B: Yes.

39 A; She told you what we are interested in.

B: Yes.

40 A: You live on 115th St.

B: No. I liveon 116th.
We encounter many such examples in our analyses of therapeutic
interviews and in everyday speech. Rule 37 obviously does not apply:
there is no Q^in the A form. Is it true that any statement can be
followed with a yes or no? The following sequences seem to indicate
the opposite.

41 A: I don't like the way you said that.

B: *Yes.

42 A: I feel hot today.

B: *No.
It is not only that 41-42 do not require or tolerate a yes or no

answer, but even more strikingly that statements like 37-40 seem to
254 SOCIOLINGUISTIC PATTERNS

demand such a response. We find many cases where speakers will

not the conversation continue unless a yes or no answer is given
let
to such statements. The rule which operates here is one of the
simplest invariant rules of discourse. Given two parties in a con-
versation, A and B, we can distinguish as "A-events" the things that
A knows about but B does not; as "B-events" the things which B
knows but A does not; and as "AB-events" knowledge which is
shared equally by A and B. The rule then states:

43 If A makes a statement about a B-event, it is heard as a

request for confirmation.

Note that in 41-42, A is making a statement about an A-event, but

in 38-40 about a B-event. Anyone can immediately test this rule in
an ordinary conversation and observe the force of its operation. This
rule contains the social construct of "shared knowledge" which is
not normally part of a linguistic rule. This is merely one of many
rules of interpretation which relate "what is said" questions, state- —
— —
ments, imperatives to "what is done" requests, refusals, assertions,
denials, insults, challenges, retreats, and so on. There are no simple
one-to-one relations between actions and utterances; rules of inter-
pretation (and their nearly symmetrical rules of production) are
extremely complex and relate several hierarchical levels of "actions"
to each other and to utterances. Sequencing rules do not operate
between utterances, but between the actions performed with those
utterances. In fact, there are usually no connection between succes-
sive utterances at all. The overall pattern of discourse analysis may
be sketched as:

Actior^ (Action 2 ) -h> Action 3

Utterancej Utterance 3 —> sequencing rules

B: : Utterance 2 * rules of interpretation

i *
(Action^ —> Action 2 (Action 3 ) > rules of production

It may be helpful to consider a more difficult case, drawn from

a therapeutic interview we have been investigating in some detail. 43

43. From studies of therapeutic interviews being conducted by the author and David
Fanshel of the Columbia School of Social Work.
The Study of Language in its Social Context 255

44 A: Well, when do you plan to come home?

B: Oh why-y?
There is no syntactic connection between these two questions and
no amount of abstract analysis will relate them correctly. One cannot
interpret B as Q-S 1 "Why do I plan to come home?" One might
:

interpret B as addressed to an implicit underlying form, A: [I ask

you] when do you B: Why [do you ask me.
. . . .]? But this would . .

be a wrong interpretation; without detailed knowledge of the

speakers and the situation, one could not hope to arrive at the
appropriate intuitive judgments to begin analyzing. We must be
aware that A is a college student, and that B is her mother; that B
has been away for four days helping a married daughter; that A and
B both know that A wants B to come home; and that B has said many
times in the past that A cannot take care of herself, which A denies.
It is then clear that 44-A is a request for action, not for information:

A is requesting that her mother come home.

There is a general rule for interpreting any utterance as a request
for action (or command) which reads as follows:

45 If A requests B to perform an action X at a time T, A's

utterance will be heard as a valid command only if the
following preconditions hold: B believes that A believes
(= it is an AB-event that)

1. X needs to be done for a purpose Y

2. B has the ability to do X
3. B has the obligation to do X
4. A has the right to tell B to do X.
Where the four preconditions do not hold in some obvious way,
we have jokes or joking insults such as: "Drop dead!" "Go jump in
the lake!" or "Get this dissertation finished by the timeI get back

from lunch!" 44 These preconditions appear in almost every rule of

interpretation and production which concerns making or responding
to commands. Note that the primitive terms of 45 include rights and

Harvey Sacks has pointed out that the first decision to be made in the inter-
44.
pretation of any utteranceis whether it is serious or not (or we might say, the degree

of seriousness involved). Appropriate reactions to jokes are limited, and almost

independent of context, but if the utterance is serious more complex rules must be
invoked. Rule 45 shows us one formal basis for this decision.
256 SOCIOUNGUISTIC PATTERNS

obligations which are plainly social constructs. Given rule 45, there
is a rule of interpretation operating for B in responding to A's ques-
tion in 44:

46 If A makes a request for information of B about whether

an action X has been performed, or at what time TX will

be performed, and the four preconditions of (45) hold,
then A will be heard as making a request for action with
the underlying form B: do X!

B's response "Oh, why?" is then aimed not at the surface request
for information, but rather at the precondition 1 of the more abstract
request for action: "Whyyou asking me
are to come home?" By
asking a question about precondition 1, B puts off A's request: since
if any of the preconditions are not shared knowledge, the request
is obviously not valid by rule 45. A's next move in this discourse

is to respond to B's request for information: she explains that the

housework and her studies are altogether too much for her to do.
Thus the content of A's response shows that she interprets B's ques-
tion as we do here.
We now find intuitively that the original request of 45 is still in
force, under the operation of a further invariant rule which states
generally that

47 If A has made a request, and B responds with a request

for information, A reinstates the original request by
supplying that information.

Since the original request is put again, B must now respond a

second time. This time B puts off the request by asking another
question involving precondition 2— by implying that Helen is the
person who should have been asked, she indicates that she herself
should not have been asked, implying that she (B) does not have the
ability to comply with the request from A.

48 A: Well, things are getting just a little too much, [laugh]

—
This is it's just getting too hard . . .

B: Well, why don't you tell Helen that?

It is obvious that the complexities of the situation do not end here.

These illustrations of discourse rules should serve to show the form

of such rules and the kind of primitive elements which they require.
The Study of Language in its Social Context 257

Although based upon several years of analysis of

this exposition is
therapeutic interviews and other speech events, it is not put forward

with the same confidence as the solutions to the problems of Sects.

2 and 3. On the contrary, discourse analysis is at a much more
primitive stage, analogous to the earliest developments in syntax and
morphology. It is some interest that the most significant
a matter of
advances in have not been made by linguists, but by
this field
sociologists. The work of Sacks (1972) and Schegloff (1968) has
located many fundamental questions concerning the selection of
speakers and the identification of persons, and isolated a number
of sequencing rules. Linguists have been handicapped in their ap-
proach to this field by their inability to utilize essential social con-
structs involving the roles of speaker and listener, obligations, power
relationships, membership categories, and the like.
It should be evident that the approach to the study of language

in its social context outlined in Sects. 2-3 of this paper can easily
accommodate the full range of elements which we need for discourse
rules. The linguistic approach can contribute a number of concepts
which are not well developed in anthropology or sociology. First
there is the distinction between utterances and actions, and the
hierarchical relations of actions whereby a question may be seen
as a request for information, which is in turn interpreted as a request
for action, which may appear on a higher level as a challenge.
Further advancement of this field may depend upon the linguistic
concept of an invariant rule, and the linguistic approach to the
formalization of such rules.
Eventually, the exploration of discourse rules will reach a quanti-
tative phase in which variable rules may be constructed and in which
large bodies of data can be introduced to confirm or reject the
tentative rules we have written. One area which plainly involves
variable rules is in the degree of mitigation or aggravation which
governs the selection of rules for making requests. We observe that
in 44 the daughter must mitigate her request; to say to her mother
"Come home right now!" would be violating a strong social con-
straint, although a mother can easily say this to her daughter. The
exact degree of mitigation, and the way in which the request is
executed involve a number of variables: age, socioeconomic class,
relative status of speaker and listener, and the form of the preceding
utterance. Such variable constraints will eventually appear in rules
comparable to those written in Sects. 2 and 3. But our present
258 SOCIOLINGUISTIC PATTERNS

knowledge is too fragmentary to make such attempts fruitful. 45

Quantitative research implies that one knows what to count, and this
knowledge is reached only through a long period of trial and ap-

proximation, and upon the basis of a solid body of theoretical

constructs. By the time the analyst knows what to count, the problem
is practically solved.
In recent years, there have been many attempts by social psychol-
ogists to characterize differences in the use of language by middle-
class and working-class speakers (Bernstein 1966; Lawton 1968).
There is little connection between the general statements made and

the quantitative data offered on the use of language. It is said that

middle-class speakers show more verbal planning, more abstract
arguments, more objective viewpoint, show more logical connec-
tions, and so on. But one does not uncover the logical complexity
of a body of speech by counting the number of subordinate clauses.
The cognitive style of a speaker has no fixed relation to the number
of unusual adjectives or conjunctions that he uses. As the example
given above shows, no useful purpose would be served by counting
the number of questions that someone asks in an interview. The
relation of argument and discourse to language is much more abstract
than this, and such superficial indices can be quite deceptive. When
we can say what is being done with a sentence, then we will be
able to observe how often speakers do it.

5. The State of Linguistics

In the introduction to this chapter, it was suggested that linguistics

was suffering from difficulty in coming to grips with the fundamental
data of language. In this respect, our field is no different from any
other social science. Linguists did take the somewhat unusual step
of redefining the field so that the everyday use of language in the
community would be placed outside of linguistics proper— to be
called speech, not language. Rather than worry about the difficulties
of dealing with this material, linguists found it quite unnecessary,
on theoretical grounds, to account for it; indeed, it was argued that
a linguist should not be concerned with the facts of speech.
Just how long such a program can continue to be productive is

45. The most thorough examination of a speech event which we have carried out
so far is the analysis of ritual insults in the black community (Labov 1972a: Ch. 8).
Although the discourse rules given there seem to be sound, we do not have the means
of corroboration which are available in our studies of linguistic structure.
The Study of Language in its Social Context 259

an open question. Clearly linguistics has benefited from a restriction

of its field of view. But if at this point linguistics is more advanced

than any other study of social behavior, it is no doubt due to the

highly structured nature of our object rather than the peculiar ex-
cellence of our strategy. In this paper, I have taken up a number
of problems where progress has been blocked, where a broader field
of view seems to be required to come to a decisive solution. The
analysis of language out of context will undoubtedly continue as a
separate pursuit; as before, we will find some linguists who spend
all of their time analyzing their intuitions about language, as others
will work with texts or carry out laboratory experimentation. My
own view is that such activity will be increasingly valued as a
necessary preliminary to the development of linguistic research. But
linguistic theory can no more ignore the social behavior of speakers
of a language than chemical theory can ignore the observed prop-
erties of the elements.
The penalties for ignoring data from the speech community are
a growing sense of frustration, a proliferation of moot questions, and
a conviction that linguistics is a game in which each theorist chooses
the solution that fits his taste or intuition. I do not believe that we
need at this point a new "theory of language"; rather, we need a new
way of doing linguistics that will yield decisive solutions. By enlarg-
ing our view of language, we encounter the possibility of being right:
of finding answers that are supported by an unlimited number of
reproducible measurements, in which the inevitable bias of the
observer is cancelled out by the convergence of many approaches.
There are many linguists who do not believe that there is a right
or wrong side to theoretical alternatives: the nature of their data does
not allow convergence on any one view or a decisive confirmation
of it.

do not mean, of course, that a particular solution offered is right

I

in any absolute sense. No one can doubt that his best effort will be
criticized, modified, replaced, or perhaps re-emerge in an almost
unrecognizable form. But within the framework provided in this
chapter, we can say that the kind of solutions offered to problems
such as consonant cluster simplification, copula deletion, and nega-
tive concord represent abstract relations of linguistic elements that
are deeply embedded in the data. It is reasonable to believe that they
are —
more than constructions of the analyst that they are properties
of language itself. The state of linguistics is indeed promising if we
can assert this about any single result of our research.
 The Social Setting of
9 Linguistic Change

THE study of language change in its social context has been

described by some as a virgin field; by others, as a barren territory. 1
A brief examination of what has been written in the past on this
subject shows that it is more like an abandoned back yard, over-
grown with various kinds of tangled, secondary scholarship. The
subject has been so badly treated with voluminous, vacuous, and
misleading essays that one can sympathize with linguists who say
that it is better left alone. But the consequences of avoiding the social
dimension of language change are serious. We are then left with such
a limited body of fact that we are condemned to repeat the arguments
of our predecessors; we find ourselves disputing endlessly about bad
data instead of profiting from the rich production of new linguistic
change around us.
Nevertheless, historical linguists have adopted and vigorously
defended a thoroughly asocial policy in the past half-century. To
understand why, it will be helpful to re-examine briefly the history
of the relations between linguists and society, and how they have
succeeded in avoiding each other. This review will raise three ques-
tions: whether the expressive and directive functions of language
are important determinants of change; whether highly abstract rules
of grammar can be affected by social forces; and whether linguistic
evolution is entirely dysfunctional. This chapter may be considered
the diachronic correlate of Ch. 8, which was concerned almost
entirely with the synchronic aspects of the first two questions. Data
from recent studies of change in progress will be presented to show
1. This chapter is a revised version of the chapter on this subject prepared for
Current Trends in Linguistics, Vol. 11, edited by Thomas Sebeok (The Hague: Mouton,
to appear).

260
The Social Setting of Linguistic Change 261

how linguistic changes may be embedded in a social context, how

they are evaluated, and how change may be activated at a particular
time and place. These findings will then be applied to the three
questions raised above.

The Linguistic View of Language as a Social Fact

Every linguist recognizes that language is a social fact, but not

everyone puts an equal emphasis on that fact. When linguists write
about language change, we find a very different degree of concern
with the social context in which these changes occur. Some broaden
their view to include a wide range of facts about the speakers and
their extralinguistic behavior, while others constrict their view to
exclude as much as possible. We can generally predict from an
author's definition of language how much he will be concerned with
the social factors in linguistic change. Furthermore, those who focus
upon the communication of cognitive or referential information will
deal more with the individual, those who become involved with
affective and phatic uses of language, with social matters.
It is not difficult to find 19th-century spokesmen for the importance

of the social factors in linguistic change. Whitney was firmly com-

mitted:

Speech is not a personal possession, but a social; it belongs, not to the

individual, but to the member of society. (1901:404)

In considering the functions of language, Whitney emphasized com-

municative function in a social sense, rather than the primacy of
"ideas":

Man speaks, then, primarily not in order to think, but in order to impart
his thought. His social needs, his social instincts, force him to expression,
(p. 401)

For an opposing view, we can look to Hermann Paul, whose

individualist approach is most current theories of lan-
reflected in
guage change (see Weinreich, Labov, and Herzog 1968). Paul saw the
language of the community as a rough mixture of the well-formed
speech of individuals. On this basis, he dismisses as quite transparent
the problem of explaining the diversification of language:

Ifwe start from the undeniable truth that each individual has his or her
own language, and that each such language has its own history . . . the rise
262 SOCIOLINGUISTIC PATTERNS

of variations seems a mere matter of course [die Entstehung der

Verschiedenheit scheint ja danach selbstverstandlich]. (Paul 1889:23 =
1886:37)

For Paul, the function of language was to organize "groups of ideas

(Vorsteilungsgruppen), a process which "takes a peculiar develop-
ment in the case of each individual" (1889:6 = 1886:22). Sweet stud-
ied Paul, and absorbed and endorsed this point of view; he cautions
that all general principles of change are subordinate "to the main
function of language . .the expression of ideas." (1900:34) We
.

should not be surprised, then, that Sweet defines language without

any reference to social context, as "the expression of thought by
means of speech-sounds." (1900:1). His explanations of language
change accordingly revolve about such individual traits as "laziness"
or "carelessness."
The emphasis on cognitive or representational functions of lan-
guage was maintained by the Prague school in their synchronic
studies. Other functions were of course recognized: following Biihler
and Laziczius, Troubetzkoy would set up three divisions of
phonology: the expressive, appellative, and representational, or
phonology proper. The net effect of this division was to free the
linguist from any concern with social factors and noncognitive
functions. After devoting a few pages to an anecdotal account of
these matters, the linguist could proceed with the real business at
hand (1957:16-17). Martinet (see below) seems by his manner of
dealing with linguistic change to be a direct descendant of this
tradition.
Bloomfield inherited from Paul the same individualist psychology,
though he did object to its subjective character (1933). Bloomfield's
S-R model shows language as the property of the individual; his
model of sound change imagines a perfectly regular but unobserv-
able process taking place within the individual's speech pattern.
Larger social factors are presented as relatively vague and confused
processes, under the chapter headings of "Fluctuation in the fre-
quency of forms" and "Dialect borrowing."
Chomsky and Halle, who differ from Bloomfield and Paul on so
many other points, continue the tradition of speculation on the basis
of individual models of the speaker-hearer relationship. Chomsky
would deliberately exclude all social variation from the subject
matter of linguistics (1965:3); Halle (1962) presents a model of lin-
The Social Setting of Linguistic Change 263

guistic change in which the individual child restructures his parents'

speech.
Although Paul's individualist views have been followed in the
main stream of linguistics, there has been considerable opposition
from many who shared Whitney's concern with the social context
of language and its wider range of social functions. Witness the
well-known position of Meillet:
From the fact that language is a social institution, it follows that linguistics
is a social science, and the only variable to which we can turn to account
for linguistic change is social change, of which linguistic variations are only
consequences. (1921:16-17)

Meillet's associate Vendryes continues in the same vein, some 20

years later:

Language is thus the social fact par excellence, the result of social contact.

It has become one of the strongest bonds uniting societies, and it owes its
development to the existence of the social group. (1951, p. 11)

Jespersen followed Sweet on many matters, but he laid heavy em-

phasis upon the role of language in social interaction:

The language of a nation is the set of habits by which the members of the
nation are accustomed to communicate with one another. (1946:21)

Although Jespersen was deeply involved in his "notional" theory of

grammar, we find throughout his writings a concern with the ex-
pressive and directive functions of language that enters strongly into
his discussions of linguistic change. Sturtevant wrote in the same
tradition:

A language is a system of arbitrary vocal symbols by which members of

a social group cooperate and interact. (1947:2)

Sturtevant's proposals for the mechanism of linguistic change

(1947:74ff) laid a great deal of emphasis on the assignment of social
and affective values.

It is clear that we understand the regularity of phonetic laws until

shall not
we learn how between phonemes leads to the victory of one of
rivalry
them. Before a phoneme can spread from word to word ... it is necessary
. .

that one of the two rivals shall acquire some sort of prestige. (1947:80-81)

But Sturtevant represented a late survival of Meillet's fading notion

that we might search for an explanation of the fluctuations of lin-
264 SOCIOLINGUISTIC PATTERNS

guistic change in the fluctuating course of social events. The pre-

dominant trend is expressed by many eminent linguists who fiercely
resist any such involvement, and insist that we must confine our-
selves to purely internal, linguistic explanations. Martinet, for ex-
ample, declares that only the results of outside impact that the
it is

linguist is competent to study. In his capacity as a linguist,

he will
decline to investigate "sociological" conditioning (1964:52).
Kurylowicz takes an even harder line:

Once we leave language sensu stricto and appeal to extra-linguistic factors,

a clear delimitation of the field of linguistic research is lost. Thus, e.g., the

physiological (articulatory) aspect may be a consequence of social factors,

the latter being themselves due to certain political or economic facts (con-
quest, migrations involving bilingualism) ... seems that the field of
It

linguistic explanation in the literal sense must be circumscribed by the

linguistic aspect of the change in question, i.e., by the actual state of the
system before and after the change ("l'etat momentanee des termes du
systeme"— Saussure). (1964:11)

Kurylowicz wishes to purify linguistic argument of all contaminating

support; he renounces the use of dialect geography, phonetics, psy-
chology, and cultural anthropology in the reconstruction of the
history of a language, in order to rise to a "higher conceptual basis"
(1964:30).
Thus linguists seem to fall into two major groups in this matter.
Group A, the "social" group, would pay close attention to social
factors in explaining change; see expressive and directive functions
of language closely intertwined with the communication of referen-
tial information; study change in progress and see on-going change

reflected in dialect maps; and emphasize the importance of linguistic

diversity, languages in contact, and the wave model of linguistic
evolution.
Linguists of Group B, the "asocial" group, focus upon purely
internal — structural or psychological — factors in explaining change;
segregate affective or social communication from the communication
of "ideas"; believe that sound change in progress cannot be studied
directly, and that community studies or dialect maps show nothing
but the results of dialect borrowing; they would take the homogene-
ous, monolingual community as typical, working within the Stamm-
baum model of linguistic evolution.
It would be unfair to argue that Group B linguists would disregard
The Social Setting of Linguistic Change 265

social factors entirely in explaining linguistic change. Rather they

define the influence of society as alien to the normal operation of
language, and view the operation of social factors as dysfunctional
interference with normal development (Bloomh'eld 1933), or as rare
and unsystematic interventions. Thus Martinet develops what we
may call a "catastrophic" view of the relations of social and linguistic
events. He argues that extraordinary social upheavals can disturb
the linguistic equilibrium at rare intervals, setting off a wave of
linguistic readjustments in which purely internal factors govern
the succession of changes over "years, centuries, and millennia"
(1964:522). Thus the influence of Norman French on
English in the
12th and 13th centuries had profound effects, being felt today
still

through a long series of internally linked readjustments. Within their

own perspective, Chomsky and Halle (1968) share this point of view.
They argue that the underlying forms of English have undergone very
little change since Middle English, and that the last serious change
in the system was probably the substitution of the Romance stress
rule for the Germanic stress rule as a result of the Norman invasion.
Thus there are areas of general agreement about the effects of
certain violent social changes upon language. No one would deny
the importance of conquests, invasions, and massive immigration,
with consequent extinction, superposition, or merger of whole lan-
guages. We can distinguish three subtypes, following Lehmann (1963):
(1) an invasion in which the language of the conquered people all
but disappears, as with Celtic in Britain; (2) a conquest in which the
conquerors eventually adopt the language of the conquered, with
consequent large-scale alteration in a class-stratified vocabulary, as
with the Norman hegemony; (3) an invasion which results in an
intimate mingling of two populations, with intimate borrowing of
vocabulary and even function words, as with the Scandinavian
invasions of England. It would be interesting to add if we could the
conditions for each of these outcomes, but the problem seems to be
an historical and political one, appropriate for the larger focus of
an interdisciplinary "sociolinguistics." 2

2. Any discussion of the history of the study of language change in its social setting

must take into account the one field where there has never been any doubt about
the importance of social context: the study of pidgins and Creoles. From the time of
Schuchardt (1909) Creolists have found it necessary to learn as much as possible about
the social conditions under which these languages were formed and reformed (see
in particular Sidney Mintz, "The socio-historical background to pidginization and
266 SOCIOLINCUISTIC PATTERNS

The issue is therefore not the importance of social factors, but

rather whether they are deeply involved in the most systematic
processes of phonological and grammatical change. Are such changes
sensitive to the socialand stylistic stratification of speech, and the
expressive information conveyed by social and stylistic variation?
Do we have to take these factors into account in order to understand
the observed regularities of linguistic change? On these questions,
Groups A and B diverge sharply, answering "yes" and "no" respec-
tively. Without doing too much violence to individual views, we can
line up in Group A such linguists as Whitney, Schuchardt, Meillet,
Vendryes, Jespersen, and Sturtevant. In Group B, we would find Paul,
Sweet, Troubetzkoy, Bloomfield, Hockett, Martinet, Kurylowicz,
Chomsky, and Halle.
But so far, we have not yet placed in this dichotomy Meillet's
teacher and associate Saussure. At first glance, Saussure's definition
of Jangue seems to place him squarely in Group A:
la partie sociale du langage, exterieure a l'individu elle n'existe qu'en
. . .

vertu d'une sorte de contrat passe entre les membres de la communaute

(1962:31).

Since Saussure is said to be the most influential linguist of the

century, and Meillet one of the most eminent in historical linguistics,
and Jespersen is currently read and cited with the greatest attention,

creolization" in Hymes 1971, and the Hymes volume throughout). Many of the sys-
tematic processes of language change that we would like to trace in "normal" linguistic
evolution can be observed in accelerated form in Creoles. Some of the most detailed
studies of systematic morphological shifting are those of Bickerton (1971a, b). He
demonstrates that speakers of the English-based Creole of Guyana move through a
very wide range of pronoun paradigms, copula rules, and complementizer placement
rules in a regular implicational serieswhich seem to reflect the historical process of
de-Creolization which affects the community as a whole. Bickerton argues with Bailey
(1970) that such regular distributions through style and social class levels are direct
reflections of change in progress. The synchronic-diachronic dichotomy of Saussure
would then be collapsed, and the new version of the wave model of linguistic change
would show symmetrical distributions through time, space and society (1971:182).
Bailey and Bickerton argue further that the Creole examples are not special cases — that
the history ofmost languages shows parallel processes. This recent revival of Group
A forces coincides with the view of Wang and his associates that grammars must
be seen as extended in time and space (Wang 1969: Chen and Hsieh 1971). The call
for a final abolition of the synchronic-diachronic distinction of Saussure represents
a sharp opposition to the Group B policy of dividing linguistic data and activity into
discrete and limited categories.
The Social Setting of Linguistic Change 267

it is not at all obvious why Group A should not be the dominant

element in 20th-century linguistics. In 1905, Meillet predicted that

this century would be devoted to isolating the causes of language
change within the social matrix in which language is embedded. But
this did not happen. In fact, there were almost no empirical studies
of language change in its social context in the 50 years following
Meillet's pronouncement. It is clearly Group B that dominates cur-
rent linguistic theory and practice; most linguists would agree with
Chomsky in taking as the object of linguistic description the "ideal
speaker-listener in a completely homogeneous speech community,
etc." (1965:3).
Without good evidence to the contrary, the great achievements and
the good authority of the linguists cited under Group B would argue
for the correctness of their asocial view. We can isolate four general
conditions which favor the predominance of Group B's outlook in
recent decades, which have to do with the linguistic climate of
thought rather than the substantive issues involved.
1. The first element in the success of Group B involves what we

may call the Saussurian Paradox. Saussure argues that langue is a

social fact, knowledge possessed by practically every member of the
speech community. It follows that one can find out about iangue
by questioning any one or two speakers of the language even —
oneself. On the other hand, parole reveals individual differences
among speakers that can be examined only in the field, by a kind
of sociological survey. Thus the social aspect of language can be
studied in the privacy of one's own office, while the individual aspect
would require social research in the heart of the speech community.
The Saussurian Paradox explains how Bloomfield could analyze
the English "spoken in Chicago" from knowledge of his own speech
(1933:90-92). The popularity of Saussure's Jangue/paroie dichotomy
was further assured when it was transformed into Chomsky's com-
petence/performance distinction. Both treatments illustrate the way
that linguists can adapt their methodology to suit their personal style
of work without abandoning principle; there is no doubt that intro-
spection is a congenial methodology for many linguists.
2. In their relations with other disciplines, linguists have tradi-

tionally leaned to the side of psychology rather than sociology. The

road between language and thought is a well-traveled one; psychol-
ogists of language have always appeared prominently in the lin-
guistic literature, from Wundt and Paul to Buhler and Jean Piaget.
268 SOCIOLINGUISTIC PATTERNS

On the other hand, Emile Durkheim's influence on Meillet seems to

have been a matter of historical accident which has not been re-
peated.
Thus nothing could be more natural for linguists than to explain
language change in terms of the parent-child relationship. To explain
the acquisition of language, one need only consider the mother as
speaker and the child as hearer; to understand linguistic change, one
considers the child first as hearer, then as speaker. Many linguists
like to perform "thought-experiments" in which they put themselves
in the position of an imaginary child grappling with fictional data
from an imaginary mother. This image follows naturally from the
small amount of data that we have on children's speech tradition-—
ally derived by parents studying their own children. More recent
psycholinguistic experimentation confronts the child with the test
situation; just as the linguist traditionally summons the informant
to a formal elicitation session, so the psychologist of language sum-
mons the child to the laboratory to wrestle with blocks or matrices,
hoping to find some meaning in answers to meaningless questions.
Since no one follows the child to observe his day-to-day interaction
with other members of society, it would be strange if our explana-
tions of his linguistic behavior took such social factors into account.
3. In the latter half of the 19th century, historical linguists were
quite open As one striking example,
to the influence of dialectology.
one can cite the impact on Osthoff and Brugmann of Winteler's
monograph on the Swiss German dialect of Kerenzen (Weinreich,
Labov, and Herzog 1968:115). But in the 20th century, dialectology
as a discipline seems to have lost any orientation towards theoretical
linguistics, and dialect geographers have generally been content to
collect their materials and publish them. Bloomfield's chapter dealing
with this subject is strangely disjunct from his other chapters
(Malkiel 1967); as others have noted (Sommerfelt 1930), the most solid
and startling results of dialectology, such as those of Gauchat (1905),
could not be fitted into the framework of neo-grammarian thinking.
It was not until the 1950's that the work of Martinet (1955), Moulton

(1962), and Weinreich (1954) demonstrated again the theoretical force

of areal linguistics.
4. The eclipse of the social group of linguists is due primarily to
the limitations of their own work and writings on the social context
of language. They relied almost entirely upon an intuitive explana-
tion of a few anecdotal events drawn from their own general knowl
The Social Setting of Linguistic Change 269

edge. When we read the comments of Whitney, Meillet, Jespersen,

or Sturtevant cannot argue that any of these authors knew more
we
about society's impact on language than anyone else; they were
simply willing to talk about it more. Typical of Whitney's arguments
about the social nature of language is one that he raises to prove
that "external circumstances" are the most important factor in lin-
guistic change:

While Family Robinson keep up their language, and enrich it with

a Swiss
names for the new and
strange places and products with which their novel
circumstances bring them in contact, a Robinson Crusoe almost loses his
for lack of a companion with whom to employ it. (1901:405)

We encounter hundreds of "thought-experiments" in these writers,

taking the place of actual data. Again, Whitney says:

Let two children grow up together, wholly untaught to speak, and they
would inevitably devise, step by step, some means of expression for the
purpose of communication; how rudimentary, of what slow growth, we
cannot tell. . .
(p. 404)

The "castaway" thought-experiment occurs over and over:

Suppose an illiterate English family to be cast away upon a coral islet in

and to be left there isolated through a succession of generations.
the Pacific
How much of our language would at once begin to become useless to them!
(p. 138)

We find to
our dismay that Whitney lives in a world of "facts" which
him but not to us, the kind of commonsense "experi-
are obvious to
ence" which has never been questioned:

The fact of variation in the rate of linguistic growth is a very obvious one.
(P. 137)

The mode of argument which the 20th century inherited to deal

with social matters is remarkably similar. We find most often a series
of anecdotes that are designed to prove ideas already accepted as
true. Thus Vendryes, expounding the Saussurian notion of the uni-
formity of iangue, reports that

We know what misadventure befell Theophrastus of Lesbos in the market

at Athens. When he asked the price of some commodity, a woman of the
people recognized him as a stranger by his speech. (1951:240)

With distressing regularity we raise the familiar desert islands,

populated with the castaways of thought-experiments:
270 SOCIOLINGUISTIC PATTERNS

When a solitary Frenchman meets a Persian on a desert island, each will

forget the differences between them and they will naturally seek to make
common cause, (p. 239)

We need only look up the word "social" in the index of any of the
—
authors cited above in Group A or Group B to encounter more—
thought-experiments and more anecdotes. Thus Bloomfield explains
the diversification of language by imaginary differences in density
of communication, discovered in an elaborate thought-experiment
which charts every utterance of every speaker in a community
(1933:46). Since the experiment cannot be performed, Bloomfield
admits that he is "forced to resort to hypothesis." The hypothesis
is then enriched with a further hypothesis about linguistic
—
change that the relative prestige of speaker and hearer determine
borrowing and the fluctuation of forms. The two hypotheses both —
stated without evidence —
are then combined in another thought-
experiment which would provide a final accounting for the direction
and rate of linguistic change:
If we had a diagram with the arrows thus weighted [by gradations repre-
senting the prestige of the speaker with reference to each hearer] we could
doubtless predict, to a large extent, the future frequencies of linguistic
forms. (1933:403)

It is difficult to believe that Bloomfield, with his fine sense of evi-

dence, would give much weight to such arguments — even his own.
By following the traditional mode of dealing with social facts, he
was also arguing that this field lies outside the domain of the lin-
guist'scompetence.
Although some of our most precise linguists are still prone to
thought-experiments and anecdote, they tend to limit these to fields
outside their normal linguistic activity. They are certainly sensitive
to such vacuity on the part of others. To understand the strong Group
B position taken by Martinet, Kurylowicz, and Chomsky, we have
to know what they were reacting against. Arguments about race and
climate in the earlier literature strike us today as too absurd to take
seriously, but these are the kinds of "external" arguments that Marti-
net and Kurylowicz were denouncing. 3

3. Even Sweet had a favorite argument about the effect of climate on language that

he returns to over and over again. He attributes the Germanic raising of I-E long a
to 6 to the fact that speakers in the northern climate tried to keep the cold, damp
air out of their mouths by not opening their lips and jaws as wide as Mediterranean
speakers (1900).
The Social Setting of Linguistic Change 271

Three Substantive Questions on Linguistic Change

Historical and strategic considerations influenced the linguists'

decision to resist being drawn into a study of the social setting of
change. But there is also a solid basis for their position in the sub-
stantive issues involved. If we decide to engage in a study of the
social setting of linguistic evolution, contrary to the dominant posi-
tion in current linguistic practice, we must raise at least three serious
questions that must eventually be resolved.

a. The Place of Social Variation

Does the social and language play an impor-

stylistic variation of
tant role in linguistic change? By I mean those language
"social,"
traits which characterize various subgroups in a heterogeneous
society; and by "stylistic" the shifts by which a speaker adapts his
language to the immediate context of the speech act. Both of these
—
are included in "expressive" behavior the way in w hich the speaker
T

tells the listener something about himself and his state of mind in
addition to giving representational information about the world.
Social and stylistic variation presuppose the option of saying "the
same thing" in several different ways: that is, the variants are identi-
cal in referential or truth value, but opposed in their social and/or
stylistic significance.
Martinet takes a strong negative position on the question of social
variation.Under the heading, "Communication alone shapes lan-
guage," he argues:

It is therefore the communicative uses of language which must engage our

attention if we wish to discover the causation of linguistic changes. What
we and be able to formulate will not necessarily hold good
shall establish
for those linguistic utterances which do not serve the purpose of communi-
cation. But we shall be disposed to disregard these since they are modelled
on communicative utterances and so offer nothing which will not be found
in these. (1964b:170)

Of course the term "communication" could include all kinds of

expressive communication, but a reading of the wider context makes
itclear that it is intended to exclude any information about the
speaker contained in the linguistic form, as well as excluding
"phatic" expression. The nonrepresentational information we are
considering under this question is usually delivered simultaneously
with other messages, so it is a question of neglecting one aspect and
allowing others.
272 SOCIOLINGUISTIC PATTERNS

To simplify our analysis, we shall assume that the language in process of

evolution is that of a strictly monoglot community, perfectly homogeneous
in the sense that observable differences represent successive stages of the
same usage and not concurrent usages ... we must disregard these [social
and geographical] variations as we did above in the case of descriptive
linguistics. (1964b:164)

Martinet's position allows us to specify the phrase "play an important

role." If theanswer to the original question a stated above is positive,
then the Chomsky-Martinet strategy will be defective, and produce
vacuous or erroneous statements about linguistic change and its
causes. In the main body of this discussion, I will assemble evidence
to show that this is the case.

b. The Level of Abstraction

Can abstract rules of phonology and grammar be

high-level,
affected by social factors? One of the factors that has led to the
decline of the Group A approach is that linguists have steadily
shifted their attention away from lexicon, phonetics, and inflectional
morphology to rules of abstract phonology and syntax that operate
at a "higher level" —
that is, they are higher up in the ordering se-
quence, if changed would affect the output of many other rules, and
contain more abstract information. We are increasingly aware that
most rules of grammar are quite remote from conscious awareness.
To state that social factors bear heavily on the systematic develop-
ment of a language therefore raises a major conceptual problem,
since speakers are not even vaguely aware of most of the deeper
relations. How are such social factors brought to bear upon the
process of language learning?
Members of Group A tended to focus strongly upon the role of
the word as receiving and reflecting social influences. The contention
of dialect geographers that every word has its own history (Meillet
1964:29, Malkiel 1967)fits in with this orientation. At the same time,

it was recognized that grammatical particles are less subject to

borrowing, more stable in the face of outside impact upon language.
This would seem even more true for the rules that relate surface
structure to underlying forms. Even if social factors should alter
profoundly the phonetics and vocabulary of a language, and possibly
the surface formatives as well, we might still argue that linguistic
change in higher-level rules is purely an internal readjustment, not
The Social Setting of Linguistic Change 273

even remotely related to the immediate social context. We will

consider this question in relation to some complex phonological
rules; the answer is not at all obvious, but it can be seen to be located
on the side of doubt.

c. The Function of Diversity

Is there any adaptive function to linguistic diversification?

Throughout the 19th century, there were many analogies drawn

between linguistic and biological evolution, not least by Darwin
himself who saw the two as "curiously parallel." He observed in
language and in biological species homologies of form due to com-
munity of descent, and he compared the generalization of phonetic
change to correlated growth in plants and animals. Languages and
species both show the reduplication of parts, the effects of long-
continued use, vestigial elements, hierarchical grouping, typological
vs. genetic taxonomies, dominance and extinction, hybridization, and
variability (1871:465). But Darwin felt it necessary to complete the
analogy by arguing for the survival of the fittest in language, quoting
as his authority Max Muller:

A struggle for life is constantly going on amongst the words and grammatical
forms in each language. The better, the shorter, the easier forms are con-
stantly gaining the upper hand, and they owe their success to their own
inherent virtue.

No linguist would today endorse this point of view, which runs

exactly counter to our notion of the arbitrariness of the linguistic
sign. Languages do not seem to be getting better and better, and we
see no evidence for progress in linguistic evolution (Greenberg 1959).
Except for the development of vocabulary, we cannot argue for
adaptive radiation in any area of language. The diversification of
languages is not immediately and obviously functional, as the diver-
sification of species may be. We receive no immediate benefit from
not being able to understand the Russians or the Gaels, and the time
taken to learn their languages does not seem to help in the survival
of our own. We must seriously consider the possibility that the diver-
sification of language is dysfunctional, and that we are worse off than
if we all spoke a mutually intelligible version of post-Indoeuropean.

As far as systematic phonetic change is concerned, most linguists,

both Group A and B, felt that the process of diversification was
entirely negative. It was only the effects of analogy, or even conscious
274 SOCIOLINGUISTIC PATTERNS

social intervention which restored the balance. 4 The notion that

sound change destroys and analogy reconstructs is so widespread
that it is almost assumed by most linguists. It is true that Martinet
(1955) finds some value in the symmetry produced by the hole-filling
tendency, and other linguists would find value in the simplification
and generalization of rules. As far as grammar is concerned, it still
must be demonstrated that analogy can be systematic, or that sys-
tematic grammatical change exists (Kurylowicz 1964).
On the whole, linguists still seem to be confronted with the prin-
ciple that diversification of language is due to the systematic and
destructive effects of sound change (usually laid to the principle of
least effort) and the breakdown in communications between isolated
groups. This finding not only destroys the parallelism between lin-
guistic and biological evolution, but is strangely conservative as well:
it projects the view that the unchanging, homogeneous speech com-

munity of Chomsky-Martinet is the ideal towards which we should

be striving, and that any degree of heterogeneity subtracts from our
communicative powers. Because this is an unattractive and seem-
ingly unrealistic result, I am inclined to reject it; some of the avail-
able evidence for this point of view will be marshalled in the final
section.

3. The Study of Sound Change in Progress:

the Uniformitarian Principle

If we are to apply ourselves to solving the three questions of

linguisticchange outlined above, we must immediately face one
grave we have too little information on the state of society
difficulty:
in which most linguistic changes took place. The accidents which
govern historical records are not likely to yield the systematic ex-
planations that we need. Some historical linguists have achieved
remarkable and insightful results with the texts on hand; H. C. Wyld
will be cited as the most brilliant example. But such efforts have
never been more than moderately convincing, and one is always free
to disagree on the basis of other fragmentary evidence. The only
strong solutions to the problems of language change just posed are
through the study of change in progress.
Group B linguists have often claimed, in defense of neo-gram-
4. In this sense, Darwin's argument was reversed. He thought that the phonetic
processes which shortened words necessarily improved them, so that the fittest or
shortest survived.
The Social Setting of Linguistic Change 275

marian principles, and on the basis of their thought-experiments,

that linguistic change is too slow, too subtle, or too elusive to be
studied as it occurs around us. In the work to be cited below, there
is overwhelming evidence to the contrary, that the study of on-going

change is a practical strategy. This evidence will reinforce and help

us interpret the results of historical investigations. To use it in this
way, we necessarily have to operate under a uniformitarian princi-
ple. 5 We posit that the forces operating to produce linguistic change
today are of the same kind and order of magnitude as those which
operated in the past five or ten thousand years.
There are certainly new factors emerging, with the growth of
literacy, the convergence of widespread languages, and the develop-
ment of scientific vocabulary. Yet these represent minor interven-
tions in the structure of languages. If there are relatively constant,
day-to-day effects of social interaction upon grammar and phonol-
ogy, the uniformitarian principle asserts that these influences con-
tinue to operate today in the same way that they have in the past.
If this principle is to any extent unjustifiable, our interpretations of

the past by the present may be wide of the mark; but from present
indications, the principle will be as successful in linguistics as it is
in geology. To quote Gauchat, the most brilliant of the earlier
workers in this field:

. les dialectes paries sont les representants vivants de phases que les
. .

langues litteraires ont parcourues dans le cours des temps. Les patois . . .

pourront nous servir de guides pour arriver a une meilleure comprehension

de l'histoire des langues academiques. (1905:176)

It may immediately be argued that we do not literally observe the

change "in progress." In most of the studies to be reported here, the

investigator observed distribution in apparent time that is, the —
differential behavior of speakers in various age levels. We distinguish
this behavior from regular and repeated age-grading by obtaining
at least one measurement at some contrasting point in real time. In
the case of Martha's Vineyard, we had the Linguistic Atlas records

5. A term borrowed from geology; the concept introduced into geological theory

by James Hutton at the turn of the 18th century. Hutton showed that the mountains,
volcanoes, beaches, and chasms we now have are the result of observable processes
still taking place around us, rather than violent convulsions at some remote time in

the past ("catastrophism"). The uniformitarian doctrine is one of the accepted princi-
ples of current geomorphology —
perhaps its fundamental tenet.
276 SOCIOLINGUISTIC PATTERNS

of 30 years before. Gauchat's study of Charmey (1905) was oriented

to real time 24 years later by the supplementary observations of
Hermann But even if we did repeated studies of the same
(1929).
area every few years, it might still be argued that we were only

studying discrete stages and not the change in progress.

Such an argument is based on a vision of linguistic behavior as
a set of uniform, homogeneous rules which can change uniformly,
much as a beam of yellow light can gradually be changed to orange;
the assumption is that we could theoretically observe the change in
the same way that we watch the sunset turn to various colors. But
this view is based on a faulty model of the homogeneous speech
community. As Gauchat demonstrated, 'Tunite phonetique de
Charmey ... est nulle." We find instead differentiated behavior, in
which there is a gradual change in the frequency with which certain
rules are utilized in various environments (Ch. 8). The internal
evolution of linguistic rules involves shifts in the ordering and
prominence of certain "variable constraints" — too abstract an object
to be directly observed in the ordinary sense. We can take measure-
ments of this process at discrete points in time, but we cannot
observe the rules change from one moment to another. Consider the
rule which diphthongized the low back /a.:/ in Charmey.

— cons [ + cons]
tense] — central)
0-^< — voc
-I- (
>/[ -I- low J <+stress> +ant
(-(-ant)
+ back <( )>
I— corr

This is which states that a back glide appears variably

a variable rule
after a low tense vowel, and that the rule applies more often to
stressed (final) words, more often if the final consonant is not an
/r/ (i.e., central), and as a lesser effect, more often if the final conso-
nant is a labial. The reflexes of Lat. porta, corpus, were rarely
diphthongized from [pworta, kwo] to [pwa°rt8, kwa ]. 6 When Her-
mann visited Charmey in 1929, he found the rule in a more advanced
state: it had gone to completion in all environments except before
/r/, where it was now quite common.

6. Though the vowel before r is given by Gauchat as open o[o], and the main

dipthongization rule affects a, two phones can be analyzed as conditioned

these
variants, and so we are dealing with the same rule. A later rule deletes final r.
The Social Setting of Linguistic Change 277

— cons
r+iow + cons]
—
0^< -voc
voc >/
[

L+ tense J <* — central)

+ back
The rule now vowel is diphthongized variably, but
states that the
that wherever the vowel not followed by an /r/, the rule applies
is

without exception. 7 It is clear that we could not follow the change

of this rule from one utterance to another, for a discrete period of
time would have to elapse before the new ordering of the variable
would be evident even for an observer who never left
constraints
The qualitative changes shown in the rule involve (1) the
the scene.
appearance of the "categorical" symbol *, for all but noncentral
environments, and (2) as a consequence, the disappearance of the
stress and labial constraints which are now obliterated in the general
advance.
Finally, something must be said about the distinction which is
often drawn between the "origin" and the "propagation" of a change
(Postal 1964:284, Sturtevant 1947, Sommerfelt 1930). Speaking on
behalf of those who investigate change in progress, I do not find
such a distinction coherent. What is the origin of a linguistic change?
Clearly not the act of some one individual whose tongue slips, or
who slips into an odd habit of his own. We define language, along
with other Group A linguists, as an instrument used by the members
of the community to communicate with one another. Idiosyncratic
habits are not a part of language so conceived, and idiosyncratic
changes no more so. Therefore we can say that the language has
changed only when a group of speakers use a different pattern to
communicate with each other.
Let us assume that a certain word or pronunciation was indeed
introduced by one individual. It becomes a part of the language only
when it is adopted by others, i.e., when it is propagated. Therefore
the origin of a change is its "propagation" or acceptance by others.
From that point on, we only have a continuation of the same pattern.
We do not rule out the possibility of independent simultaneous
innovation by a number of speakers; but we do find absurd the
notion that an entire community would change simultaneously
without reference to each other, without a gradual transfer of the

7. The asterisk notation indicates a feature in a variable rule which, when present,
causes the rule to apply categorically, without exception (Labov 1972a: Ch. 3).
278 SOCIOLINGUISTIC PATTERNS

pattern from speaker to speaker. Every empirical study, beginning

with Gauchat, shows such systematic differentiation in even the most
isolated and closely knit communities.
In this chapter, I will utilize data from eight of these empirical
studies of speech communities:

1. Louis Gauchat's investigation of phonetic diversity among three

generations of Swiss French speakers in the village of Charmey

(1905), with Hermann's following report (1929).

The main sound changes observed in progress are summarized in

Table 9.1, showing (a) the palatalization of 1 —> y, proceeding variably
in the middle generation, completed in the youngest; (b) the mon-
ophthongization of /ao/, variable in the oldest generation, completed
in the middle; (c) the diphthongization of back d—> a beginning in
the middle generation, completed in the youngest except before /r/;
(d) the diphthongization of e —> e beginning in the oldest generation,
1
,

variable in the middle age group, and completed in the youngest

generation except for the word class with underlying /r/ following
the vowel.

TABLE 9.1.

FOUR SOUND CHANGES IN PROGRESS IN THE

SWISS FRENCH OF CHARMEY, 1899

/ // ///

90-60 yrs 60-30 yrs under 30

(*) i \~y y
(aw) o°-(o-) a* a*
(ey) e - (e
j

)
e ~~ e
1
e
1

W t> t)-a° a°

Source: Gauchat 1905.

2. Ruth Reichstein's investigation of phonemic variables among

Parisian school children (1960), against the background of Martinet's
World War I inquiry.

Reichstein tested some 570 schoolgirls for phonemic contrast with

nine minimal pairs, revolving about /a~a/, /e^e:/, /e^-rie/; these
contrasts seemed to be disappearing rapidly, and comparison by area
and class structure shows that certain interior working-class districts
led the way.
The Social Setting of Linguistic Change 279

3. My own study of the centralization of (ay) and (aw) on Martha's

Vineyard, compared to earlier phonetic records made by Linguistic
Atlas interviewers.

This study showed (a) progressive centralization of /ay/ in the oldest

generation, with (b) a later centralization of /aw/ in the middle
generation, overtaking the process for younger speakers: see
first

Table 9.2 (a rearrangement of Table 1.2). The Atlas data show no

/aw/ in 1933. Recent spectrographs studies of the
centralization of
same speakers (Labov 1972c) have confirmed the original view of
the mechanism, and added considerable detail.

TABLE 9.2.

CENTRALIZATION OF (ay) AND (aw)

IN 3 GENERATIONS OF ENGLISH
SPEAKERS: MARTHA'S VINEYARD,
MASS.

Generation (ay) (aw)

la (over 75 yrs) 25 22
lb (61-75 yrs) 35 37
I la (46-60 yrs) 62 44
Mb (31-45 yrs) 81 88
III (14-30 yrs) 37 46

4. My own study of the evolution of New York City vowels (1966a),

as confirmed and enlarged by our current instrumental studies
(1972c). Relation to real time is shown by comparison with four other
reports extending back to 1896.

The New York City investigations show (a) a rise in the stratification
of final and preconsonantal /r/ among speakers below 40; (b) a
tensing and raising of short a to form the variable (eh) and raising
of long open o to form the variable (oh), from low to high position,
with accompanying merger of mid and high ingliding vowels; (c),
a backing and raising of the nucleus of /ay/ and /ah/ in guy and
God, with a corresponding fronting of the nucleus of /aw/. Fig. 9.1
shows four stages in the chain shift of /ahr—> ohr—> uhr/ relating
the word classes of lard, lord, and iured, from the spectrograph^
measurements of the vowel systems of four working-class New
Yorkers.
 A
o
LO _

o
o
o
o o
o o
o o
o o
o
o
o oo
oo o
o
o
o
o
o
o
o o
o
o
o
o
o
"t LO o r\ co 0^ o ^^r LD o l\ co 0> o «—

Fig. 9.1. Four stages in the New York City chain shift
ahr —» ohr —» uhr.
The Social Setting of Linguistic Change 281

5. The study of Hillsboro, North Carolina, carried out by Levine and

Crockett, primarily through questionnaires and formal tests of pro-
nunciation (1966); a study of the black population of the same town,
using the same methods, by Anshen (1969).

Levine and Crockett have reported so far only the pronunciation of

finaland preconsonantal /r/. They found a strong shift towards a
new norm of r-pronunciation, parallel to 3a, but evidence for an older
high-prestige norm of r-lessness surviving also.

6. A
recent investigation of the English of Salt Lake City and its
environs by Stanley Cook (1969) showing an early stage in the
development of an urban dialect.

The most prominent feature in the process of change was the fronting
of (aw), which was strongest among young college people, and
spreading gradually outwards from Salt Lake City. Cook also studied
the rural merger (and stereotype of reversal) of /ar/ and /or/ in far
and for; it was shown to be an advanced change undergoing a certain
amount of overt correction. Cook was able to trace the history of
the merger in a rural community, its stigmatization and reversal
among younger speakers, and a tendency towards hypercorrect
raising of /or/ to [oV] among younger suburban speakers. The
separation of the two word classes among rural speakers is shown
in Table 9.3.

TABLE 9.3.

PHONES USED FOR (ar) AND (or) IN CASUAL SPEECH: MINERSVILLE, UTAH

Speakers N a o ov o* fi o

Older informants 5 (ar) 89 11

(60+ yrs) (or) 40 8 53

Noncollege youth 2 (ar) 100 — — — — —
(ca. 17 yrs) (or) 14 29 54 4
College students 2 (ar) 100 —
(ca. 21 yrs) (or) 100

Source: Cook 1969.

7. An investigation of the social stratification of Spanish in Panama

by Henrietta Cedergren (1970), showing five linguistic variables
distributed across four socioeconomic groups.
282 SOCIOLINGUISTIC PATTERNS

Cedergren found one of these variables — the retroflexion and frica-

tivization of /c/ — was in process of rapid change, while the others
showed fairly stable distributions over age levels.

8. The study of the social stratification of language in Norwich,

England by Peter Trudgill (1971), using the same basic interview

techniques as in 4 and eliciting a range of styles from casual to word
list from a sample of 60 informants drawn from five social classes.

Trudgill also found several linguistic variables in the process of

change, both from "above" and "below" in the social hierarchy. The
backing of short /e/ to [a] before /l/ is the case we will consider
in detail, as well as Trudgill's findings on the differentiation of the
sexes in the process of change.
There are of course many studies of secondary data which give
us considerable insight into the social setting of linguistic change.
The most useful examination of historical texts, for our purposes,
is Wyld in A History of Modern
that of H. C. Colloquial English (first

published 1921). Wyld had a strong interest in the influence of class

dialects on the history of English, and their origins in regional
dialects, because he himself was a speaker of a class dialect, "Re-
ceived Pronunciation." We can profit from his close examination of
the spelling in such documents as the Cely papers, the Paston letters,
Machyn's diary, and the Verney Memoirs, by writers with a wide
range of social background. 8
We might enlarge the list of sources indefinitely by including
qualitative and fragmentary accounts of change in progress. But
quantitative studies of actual speech communities must necessarily
take precedence. We can draw some conclusions from qualitative
observations which are framed in a three-category system: that a
given form or rule is never found, occurs variably, or is always found.
But much more profitable use can be made of linguistic variables
ranging freely from to 1, so that < x < 1 where x represents the
proportion of all permitted environments in which the rule actually
applies. Here we are in a better position to detect the progress of

It has been pointed out by Asta Kihlbom (1926) that there are many serious defects
8.

in Wyld's treatment of interpretation of this data, since he did not consult the original
manuscripts, and many of the letters are in the hands of secretaries rather than those
of the principals themselves. But Wyld's larger reconstruction of the processes involved
seems to stand.
The Social Setting of Linguistic Change 283

a change or give a convincing account of how the change is corre-

lated with social factors.A review of Tables 9.1-9.3 will show that
most of the relationships would disappear in a qualitative report.
In the case of Gauchat, we will see that the most interesting and
important findings add quantitative detail to the qualitative Table
9.1. In our most recent studies of sound change, we can move to a

higher level of precision by the use of instrumental measurements,

and this has now been done for two of the six studies outlined above
(see Labov 1972c; Labov, Yaeger, and Steiner 1972).

4. The Embedding of Linguistic Change in its Social Context

We can identify at least five different problems connected with

the explanation of linguistic change ( Weinreich, Labov, and Herzog
1968) but not all of them are related to the social setting of the change.
The universal constraints upon linguistic change are by definition
independent of any given community. The question of locating the
transition between any two stages of a linguistic change is an internal
linguistic problem. The embedding problem has two aspects: the
change is seen as embedded in a matrix of other linguistic changes
(or constants), and also as embedded in a social complex, correlated
with social changes. There is also an important social component
to the evaluation problem —
to show how members of the speech
community react to the change in progress, and discover what ex-
pressive information the variants convey. Finally, we can expect that
social factors will be deeply involved in the actuation problem: why
ittook place at the particular time and place that it did.
It should be clear then that a full understanding of linguistic

change will require many investigations that are not closely tied to
the social setting, and other studies which plunge into the network
of social facts. Our other studies of the universal constraints on the
expansion of mergers (Herzog 1965:211), on universal principles of
vowel shifting (Labov, Yaeger, and Steiner 1972 Ch. 4), and the
internal transition of rules (ibid, Ch. 3) are concerned with the
speech community only as a source of data. In order to assemble
the data we need to answer the three questions raised in Sect. 2,
we will have to take full advantage of the available data on the
social embedding, evaluation, and actuation of the linguistic changes
under study.
Our first problem is to determine the aspects of the social context
284 SOCIOLINGUISTIC PATTERNS

of language that are most closely connected with linguistic change.

We might begin with a complete account of the immediate social
setting of the speech event, following the descriptive program of
Hymes (1962). We would consider all social relationships holding
among speakers, addressees, audiences, and inhabitants of the social
domains of the speech event (school, church, job, family . .). We
.

could then ask whether changes in the language reflect changes in

the relationships between these participants and settings. For exam-
ple, we are now witnessing a steady series of changes in the use of
second-person pronouns of respect in Spanish (Weinberg and Najt
1968), French (Lambert 1969), Serbo-Croatian, and other languages.
We suspect that there is some truth to the conventional reaction of
older people: that young people don't have the respect for their elders
that they used to. But what independent measures of respect behav-
ior would show us that this is more than a change in conventions
or surface modes of expression? We have increasingly sophisticated
means of recording and measuring linguistic behavior, relating our
observations to the language used in everyday social interaction. But
we do not have equally well-developed measures of authority, re-
spect, or intimacy. We would therefore be wise to correlate our
linguistic data with whatever measures of social position or behavior
can be repeated reliably by others at other points in time.
It then seems reasonable to connect linguistic behavior with

measures of ascribed and achieved status of the speakers. Whereas

shifts of linguistic expression can register momentary changes of
social attitudes, we will be more concerned with well-established
—
ranges of linguistic expression the way in which the individual
habitually presents himself in various social settings. From one
moment to another, our language gives information to answer the
listener's question, "How are you feeling about me?" But the
speaker's language provides general information about himself as
well, answering the other's questions, "Who are you?" and "What
—
are you?" These are matters of ascribed status ethnic and reli-
—
gious membership, caste, sex, family and achieved status educa- —
tion, income, occupation, and possibly peer-group membership.
Changes in language can then be correlated with changes in the
position of the subgroups with which the speaker identifies himself.
Current evidence shows that most incoming changes follow signif-
icant social distributions before they register any stylistic shifting
(see Cook 1969 on /aw/).
The Social Setting of Linguistic Change 285

Socioeconomic Class
The social status of an individual is determined by the subjective
reactions of other members of society, but it is easier for outsiders
to use objective social and economic indicators to approximate the
position of given individuals. In the United States, we obtain the
sharpest overall stratification with various combinations of occupa-
tion, education, income, and residential area. In studying historical
records, we usually judge upper-class figures by their family con-
nections and title; less prominent individuals are easiest to classify
by their occupations and habitual associations.
Henry Machyn, the Diarist, seems from his own words to have been a simple
tradesman, possibly an undertaker, with a taste for pageants especially —
for funerals (as was natural) —
and for gossip. Of the great persons whom
he mentions, he knew no more than their names and faces, scanned as they
rode past him in some procession, and an occasional piece of gossip picked
up, one is inclined to think from some other spectator among the crowd.
(Wyld 1936:141)

Machyn's status is an important issue: he provides us with evidence

for the lower-middle-class treatment of several important linguistic
variables in 16th-century London
English. For example, most of his
-er- words armyn, 'ermine'; hard 'heard'; sarmon,
are spelled as -ar-:
'sermon'. Wyld takes this alternation of -er and -ar- as a classic case
of the correlation of social class movement and linguistic change.
He raises the question: How did it come about that many words now
pronounced with a mid central vowel, [&], like clergy, heard, etc.,
were pronounced by good speakers in the 17th and 18th centuries
with [ci3\|? If the [a^] pronunciation was "wrong," how was it adopted
by the aristocracy in the first place?
Wyld collects a considerable body of data to show that the [W]
pronunciations entered the London dialect in the 15th century from
Southeast dialects, and appear strongly in the private writings of
middle-class Londoners like Machyn up to the middle of the 16th
century. From this point, such spellings appear with increasing
frequency in upper-class English until the end of the 18th century,
when they begin to recede.

We have here a linguistic feature which found its way from a Regional
dialect intoMiddle Class London speech, passed thence into Received
Standard, only to be ousted later by a fresh wave of Middle Class influence
(1936:11).
286 SOCIOLINGUISTIC PATTERNS

The ultimate rejection of the [W| pronunciation by middle-class

speakers in favor of the standard spelling pronunciation was a
natural byproduct of upward social mobility. 9 The new bourgeoisie
had no way of knowing which words spelled -er- were pronounced
[aa\| by elegant aristocrats, and they naturally could only adhere to

the solid orthographic standard.

We see in this development three processes of social embedding:
(1) the transformation of a regional dialect into an urban lower-class
spread of a linguistic feature upward from a lower
dialect; (2) the
to an upper class; (3) the recourse to spelling pronunciation by an
upwardly mobile group. We can easily find further support for
Wyld's view of the matter. The [a^] pronunciation still appears in
Southeast dialects today. 10 And we will see, there is ample evidence
for the upward movement of lower-class innovations or importations
into the standard language.
Oddly enough, a great deal of the speculative literature on dialect
borrowing is based on the notion that all movement of linguistic
forms is from the higher-prestige group to the lower. 11 When
Group B linguists deal with this topic, they inevitably assert such a
principle.

Among his occupational companions, for example, a speaker will imitate

those whom he believes to have the highest "social" standing. (Bloomfield
1933:476)

This is simply a remark, with no more justification than any of the

other general observations in Bloomfield's treatment of dialect bor-
rowing. Studies of current sound changes show that a linguistic
innovation can begin with any particular group and spread outward
and that this is the normal development; that this one group can be
the highest-status group, but not necessarily or even frequently so.
We can find two other examples in Machyn's speech of incoming
changes that eventually reached the upper class. One is the high
pronunciation of -ea- words from M. E. § as [i:], indicated in

9. An alternate analysis of the New York City data (Labov 1967) based on the social
mobility of the speakers provides as good or better correlation with linguistic behavior
than the analysis based on the speakers' socioeconomic positions.
10. As shown in data gathered by Howard Berntsen and analyzed in our current
studies of sound change in progress.
11. Expressed as a general principle by Gabriel Tarde in his Lois de l'imitation in
1890, and since known as Tarde's law.
The Social Setting of Linguistic Change 287

Machyn's spelling by y in prych, spyking, bryking, brykefast. Even-

tually, this was detached from the mid-vowel position
word class
of long a words so that meat fell together with meet rather than mate.

Secondly, we note in Machyn a pronounced tendency for the loss

of -r- before -s, giving Woseter 'Worcester', Dasset 'Dorset', continu-
ing a tradition which ultimately led in the 18th century to a com-
12
pletely r-less pronunciation for all Londoners.
In more recent studies, Reichstein (1960) found that several on-
going phonemic changes she was studying were most advanced in
working-class districts of Paris. Our own studies of the suburbs of
Paris find the most extreme raising and backing of /a/ among work-
ing-class youth, so that casser la tete can be raised to [koselatet], and
13
j'sais pas to [spo]. In our other exploratory studies of the dialects
of Boston, Rochester, Detroit, and Chicago, we find the most ad-
vanced forms of many vowel among working-class youth. In
shifts
Chicago, working-class youth who show the most extreme
it is

tensing and raising of short a and fronting of short o, together with

the lowering of lax short vowels /i/ and /e/ to mid and low position.
Fig. 9.2 shows such an extreme vowel system, that of a 17-year-old
Va
working-class girl. In her sentence, That ended that [Si: t endid
Va
Si: t], we observe the low vowel /ae/ raised to high position, and

the mid vowel /e/ opened to low position.

Dialect studies based on the speech of college students fail to
capture such remarkable transformations of vowel systems until they
are already far advanced, and already being repressed.
It is not always the working class which is at the leading edge

of linguistic change. Often it is a higher-status group, comparable

to Machyn's. Table 9.4 gives systematic data on one of the ongoing
changes in the New York City vowel system: the backing and raising
of /ay/ along with the fronting of /aw/. The numbers here refer to a
scale in which zero would represent a low central nucleus for either
/ay/ or /aw/: [a1 au ], characteristic of the oldest speakers. Consistent
,

12. By r-less, we mean that there is a categorical rule for the vocalization of
consonantal and preconsonantal position: whenever not followed directly
r in final

by a vowel. This style of speech was adopted in the 19th century in all those cities
of the United States which looked to England as a center of cultural prestige: Boston,
New York, Richmond, Charleston, and Atlanta, but not Philadelphia.
13. These data are from the field work of Bryan Simblist, of Yale University.

Although the working-class Parisian youth that we have studied preserve a great
phonetic distance between front and back a, the lexical distribution of this distinction
is quite different from that of the standard.
 o
o

g> E>

Z> >>

1
D>

• @ ^

OOOO

"SO©© ^0£>[I> * DD^lH

Fig. 9.2. Vowel system of CM., 16 years, Chicago. (Reproduced

by permission of the Indiana University Press from W. Labov,
"The Internal Evolution of Linguistic Rules" in R. Stockwell
and R. Macaulay (eds.) Historical Linguistics and Generative
Theory.)
The Social Setting of Linguistic Change 289

TABLE 9.4.

AVERAGE AND
VALUES FOR ALL
(ay) (aw)
WHITE NEW YORK CITY INFORMANTS BY HALF-GENERATIONS

(ay) (aw)

Socioeconomic class Socioeconomic class

Age Generation 0-2 3-5 6-8 9 0-2 3-5 6-8 9

5-19 ll-B 7 23 22 12 8 20 17 8
20-34 ll-A 5 18 24 10 — 7 10 4
35-49 l-B 8 17 18 20 4 7 8 1

50-64 l-A 5 10 10 15 2 7 5 5
65 +

Source: Labov 1966a.

Index of (ay) and (aw) = average of phonetic ratings x 10
Phonetic scales for rating fronting of nucleus of (aw) and backing of nucleus of (ay):

£asf.
*.£.

I 2 3
< ££#«
X*'
ve'
W 3 %^O
<>
o 2,
9- oSk

o
o
— °>
i

%
290 SOCIOLINGUISTIC PATTERNS

pronunciation of /ay/ as [d 1 ] would yield (ay)-40, and consistent /aw/

as [ae u ] would also show a score of 40. Table 9.4 shows that the
movement of /ay/ seems to have begun in the middle-class groups,
but continued most strongly in the lower middle class, spreading
gradually to the working class while upper-middle-class speakers
avoided it. The fronting of /aw/ is strongest in the lower middle
class, but is gradually expanding in this generation to affect the other
groups.
Innovation by the highest-status group normally a form of is

borrowing from outside sources, more or with some

less conscious;
exceptions, these will be prestige forms. 14 The original spread of
r-less pronunciation in the United States was modelled on the Lon-
—
don patterns of the early 1800's a "change from above" spreading
outwards from the centers of Anglophile influence. 15 The present
reversal of this trend throughout the eastern United States can be
seen in Boston and the South (Levine and Crockett 1966) as well as
New York (Labov 1966a). This is another form of change from above,
reversing the prestige relations. There is a sharp differentiation
between older speakers, who still reflect the earlier Anglophile
norms, and the younger speakers who have adopted the new broad-
cast norm of "general American" r-pronunciation. Change from
above can be quite regular; it can affect each subgroup in proportion
to its distance from the center of prestige, and the formality of the
speech situation. A linguist who is interviewing subjects will observe
these importations from above more clearly in reading texts than
connected speech; in the unattended speech of everyday life, we
observe more dramatically the effects of change from below, as in
the movements of (ay) and (aw) cited above. 16 Table 9.5 shows the
average indexes of r-pronunciation in casual style for New Yorkers:
it is plainly an upper-middle-class phenomenon which has not

affected the natural speech of other groups. Table 9.6 shows the way

14. In The Pickwick Papers, Dickens marks Sam Weller and his father with the
stigmatized v — w confusion; he also ridicules a foppish young lord with a stereo-
typed labialization of prevocalic r, all printed as w's in the text.
15. This expansion reached a radius of approximately 150 miles from Boston,
Richmond, and Charleston, but was confined to the immediate vicinity of New York
City.
16. In the early stages of a linguistic change, a consistent pattern can be observed
in formal styles as well as casual speech. This is the case with most New Yorkers'
pronunciation of front /aw/ and back /ay/. It is not until the more advanced stages
of the change that formal styles show correction and irregular distribution.
The Social Setting of Linguistic Change 291

TABLE 9.5.

(r) IN CASUAL SPEECH BY AGE AND CLASS

Socioeconomic classification

Age 0-1 2-5 6-8 9

8-19 00 48
N
00 01
20-29 00 35 4 11 4 4
00 00
3 5 5 3
30-39
40-49
00
00
00
06
00
10
32
18
-243
5 18 7 3
50- 00 08 00 05 5 7 13

TABLE 9.6.
PERCENT OF SPEAKERS IN WORD
(r)-00
LISTS (STYLE D) BY AGE AND CLASS

Soc ioeconomic class

Age 0-1 2-5 6-8 9

N
8-19 50 50 50 25
2 8 2 4
20-39 67 75 00 20 9 4
5 17
40-49 20 18 00 25 3 7 1 3
50- 64 27 00 33 11 11 1 3

Source: Labov 1966a.

in which form affects speakers in the most formal style:

this prestige
the percentage who
use the completely r-less norm in reading word
lists. The second-highest status group —
the lower middle class is —
most affected; no adult speaker stayed with the older norm. This is
one of the many indications of the strong pattern of "hypercorrect"
behavior on the part of the second status group, an important element
in the mechanism of linguistic change (see Ch. 5). A parallel example
of this hypercorrect pattern appears in the quantitative study of
r-pronunciation of Hillsboro; as Table 9.7 shows, the greatest shift
from "sentence style" to word list style was on the part of the
high-school educated group, rather than the college group. In the
first case, subjects read sentences with blanks; their attention was

drawn to the problem of filling these with a lexical item, rather than
to the words of the text that contained /r/. In the word list, attention
was focused directly on the variable, and the second-highest status
group responded more sharply to this difference than any of the
others. Returning to the age relationships in Table 9.6, we find that
292 SOCIOLINGUISTIC PATTERNS

TABLE 9.7.

FREQUENCY OF FINAL AND PRECONSONANTAL [r] FOR WHITE SPEAKERS:

IN HILLSBORO, N.C

Sentence-list Word-list Net increase

Age
21-39 yrs 56.6 65.1 8.5
40-59 yrs 54.2 60.3 6.1

60 yrs and over 44.5 49.3 4.8

Education
Any college 52.7 58.9 6.2
High school graduate 54.6 65.6 11.0
Some high school 50.0 57.0 7.0
Grade school or none 52.6 57.3 4.7
Sex
Male 52.3 57.4 5.1

Female 52.9 61.1 8.2

Source: Levine and Crockett 1966:223.

in the working-class groups, the older speakers tend to hold to the

earlier norms, and the younger speakers as well; it is only among
the middle-aged groups that the new prestige norms are adopted.
As we will see below, overt correction tends to be rather unsystem-
atic when it occurs late in it focuses on individual words
life, and
rather than on general rules. 17 We
might ask: can the highest-status
group ever innovate unconsciously?
In so far as any habit of speech is associated with a high-prestige
group, it is apt to be remarked. Conservative critics will quickly call
attention to it, as Gill ridiculed a fashionable 17th-century pronun-
ciation which raised long a to [i:], giving [ki:pn] for capon (Wyld
1936).
Not every linguistic change is attached to a particular social group.
The raising of M.E. e to seems to have been "common to the
[i:]

speech of all areas and classes in London" (Wyld 1936:207). In areas

of the United States, the merger of short open o and long open o
seems to affect everyone in certain areas. 18 But here I am speaking

17. Speakers who have acquired the norm of formal r-pronunciation late in life

show regular and predictable patterns of shifting, but do not achieve consistency.
The word classes containing /r/ are fairly well defined, yet there is a fair amount
of hypercorrection, in idear, lawrand order and occasionally Gard for God.
18. There are exceptions to the uniformity of this merger of hock and hawk, don
and dawn. In Phoenix, for example, it appears to be more characteristic of the Anglo
population than blacks or Spanish-Americans.
The Social Setting of Linguistic Change 293

from general impressions; it cannot be denied that in every case that

has been studied closely, one social group or another has been found
to lead strongly in the development of a linguistic change.
The difference between a change in progress and an advanced
change may sometimes be seen clearly in the pattern of social distri-
bution. A change may begin first in a social group located anywhere
in the social hierarchy. As long as it is developing and spreading
outward, one can still see the pyramidal pattern through various age
levels, with the highest values in the youngest speakers of the original
group. But when the change reaches an advanced state, and all social
classes are affected, it is often stigmatized, and the social correction
of formal speech begins to obscure the original pattern. In this case
we get a linear distribution, with the highest social class showing
the least amount of the stigmatized feature in ordinary conversation.
We can see this clearly in Cedergren's study of the Spanish of
Panama City (1970). Table 9.8 shows Cedergren's findings on the
social stratification of five variables in Panamanian Spanish across
four social classes.

TABLE 9.8.

SOCIAL STRATIFICATION OF FIVE

SPANISH VARIABLES IN PANAMA

Social groups

Variable / II III IV

(R) 1.62 1.88 2.29 2.29

(PARA) 1.11 1.37 1.39 1.69
(ESTA) 1.26 1.56 1.62 1.71
(S) 2.03 2.24 2.31 2.36
(CH) 1.88 2.24 2.13 2.00

Source: Cedergren 1970

The linguistic variables in Table 9.8 may be defined briefly as

follows:

(R): the devoicing, fricativization, pharyngealization, and dele-

tion of syllable-final /r/, with values ranging from 1 to 6
in the direction of these processes.
(PARA): the alternation of the full form of the preposition para with
pa, with values of 1 and 2 respectively.
(ESTA): alternation of the full form esta with ta, assigned values
of 1 and 2 respectively.
294 SOCIOLINGUISTIC PATTERNS

(S): the syllable-final alternation of [s], [h] and [0], with values
of 1, 2 and 3 respectively.
(CH): palatal vs. retroflex and reduced stop onset of /c/, with
values of 1 and 2 respectively.

The figures in Table 9.8 are the arithmetical means of the values
of the variables. The social classes range from the highest (I) to the
lowest (IV), with the first four variables showing a linear distribution
with lowest scores for the highest social class. But the (CH) variable
shows a curvilinear pattern which suggests
change originates
that the
in the second-highest status group
That (CH) represents an on-
II.

going change in progress is clearly shown by Table 9.9, which gives

the distribution of scores by age groups for the same five variables.
We can see that (CH) is the only variable with a unidirectional linear
progression across age groups, with a dramatic advance in the
youngest groups quite comparable to the values shown for change
in progress in Martha's Vineyard and New York City.

TABLE 9.9.
DEVELOPMENT OF FIVE SPANISH VARIABLES BY AGE GROUPS

Age

Variable 11-20 21-30 31-40 41-50 61-70

(R) 2.28 1.90 1.95 2.23 1.46

(PARA) 1.31 1.34 1.48 1.33 1.39
(ESTA) 1.64 1.50 1.67 1.57 1.41
(S) 2.34 2.22 2.15 2.18 2.19
(CH) 2.15 2.29 2.05 1.81 1.31

Source: Cedergen 1970.

The situation in Panama City would not in itself justify the princi-
ple that a curvilinear pattern of social stratification corresponds to
early stages of a change in progress. But in New York City we also
find that, of five major variables studied, two show a curvilinear
distribution: the raising of (eh) in bad, ask, dance, is most advanced
among upper-working-class speakers, and the raising of (oh) in off,
all, water is most advanced in the lower middle class. Distribution

in apparent time clearly shows that these two variables are still
developing among the oldest speakers; the raising is found chiefly
in the social class which now shows the most extreme forms. Trud-
gill's study of the sociolinguistic structure of Norwich (1971) con-
The Social Setting of Linguistic Change 295

siders a number of phonological variables which show linear social

stratification — matching the hierarchical distribution of social
classes. But one variable shows a curvilinear distribution the —
backing of /e/ to [a] before /l/ in belt, held, etc. Trudgill shows that
the distribution of this variable in apparent time clearly indicates
change in progress, with upper-working-class speakers in the lead.
An understanding of this principle demands an appreciation of
the difference between the regular progress of change from below
in the early stages of a sound change, and the later correction from
above which occurs when the changing feature comes to the atten-
tion of those who set social norms. If the new linguistic element is
associated with a lower social group, it will usually be stigmatized,
and eventually its use will be inversely related to social status. The
later stages of a change may then show a linear distribution, although
the leading edge was originally in an interior group. If the linguistic
innovation spreads upward from the very lowest social group, it will
be aligned with the social hierarchy from the beginning, but this
seldom happens. It does sometimes happen that a feature will be
introduced by the highest class in the social system, though as a rule
this is not an innovating group.
We have thus documented in some detail the embedding of lin-
guistic change in one type of social structure: the socioeconomic
system of class differentials. Change does not occur without regard
to class patterns; instead, the incoming pattern enters like a wedge
with one group or another acting as the spearhead. The feature is
rarely confined to a particular class (unless it is stigmatized and
—
recedes see Sect. 5 below). It is therefore difficult to see how
Martinet or Chomsky would handle this situation. Would the homo-
geneous society posited by the abstract linguist contain the new form
or not? If not, at what point would the abstractor make the decision
to include it in his homogeneous system and for what reason?
I can imagine two directions in which the abstract linguist might
go. He might simply abstract from social class, and allow the in-
coming form to be optional for everyone. He would then argue that
its association with a particular class of people is below the level

of linguistic significance. In that case, the new option would appear

suddenly as a blind fact, without direction or interpretation. Our
discussion of the evaluation of such well-developed linguistic
changes will illustrate how much significance the option can have
in fact.
296 SOCIOLINGUISTIC PATTERNS

A second alternative would be to abandon the notion of writing

a description of langue as a community property, and simply de-
scribe the speech of one class. If the community is differentiated
according to class, we could then seize the homogeneous groups
which do exist, and describe their language. As a first step this might
be practical, but complications ensue when we discover that there
are other social structures such as caste or ethnic group which in-
tersect with socioeconomic class.

Ethnic Group and Caste

The beginning of the current interest in sociolinguistic studies can
be seen in the 1960 publication of Ferguson and Gumperz, Linguistic
Diversity in South Asia. Some of the papers concern literary-
colloquial oppositions, but others describe caste differences: in
Kannada (Bright and McCormack), Tamil (Pillai), Bengali (Dimock
and Chowdhury), and Hindi (Gumperz and Nairn). Bright was prim-
arily concerned with the effect of caste differences on linguistic
change. Though his comparisons of two dialects of Kannada are
— —
based on only two informants college students his conclusions are
of the greatest interest. The non-Brahmin dialect (NB) accepted
foreign words and foreign phonemic patterns (English and Sanskrit)
less readily than the Brahmin dialect (B); B was more resistant to
grammatical change and phonological change from within.
For example, B has imported /f, z, d/ from English, where NB has
substituted native /p, j, a/, so that B kofi 'coffee' = NB kapi, and B
has Sanskrit /s/ in santi, 'peace', while NB has santi. On the other
hand, B shows original /e/ in words where NB has differentiated
to /ya/ if a mid or low vowel occurs in the next syllable: B pete,
NB pyate.
A later secondary study of Tamil and Tulu showed the same
general pattern.

The Tulu evidence shows the Brahmins as chief innovators in the more
—
conscious varieties of change semantic shift, lexical borrowing, and
phonological borrowing. In the less conscious processes of phonological and
morphological change involving native materials, both B and NB dialects
innovate. (Bright and Ramanujan 1964)

Bright and Ramanujan conclude that upper- and lower-class dialects

innovate independently; the more conscious importations are regu-
larly the mark of the upper class, while the less conscious changes
The Social Setting of Linguistic Change 297

affect both classes. They then point out that the Brahmins can show
overt correction of these changes, as inKannada and Tamil. These
findings from South Asia with the general principles of
fit in well
change which emerge from our New York City studies, contrasting
19
the behavior of the highest-status group with the others.
In the development of the New York City vowel system, we find
that ethnic identity plays an important role more important than —
socioeconomic class, for some items. The ethnic differentiation of
the tensing and raising of short a in bad, ask, dance — the (eh)
variable — is Table 9.10. The numbers represent the average
shown in
degree of openness of the vowel. Consistent pronunciation of bad
with [ae*] at the level of unraised bat would show a value of (eh)-40.
If the nucleus of all such short a words that were tensed were then

raised to the level of [e:], the index would show (eh)-20, and for the
level of [i:], (eh)-10. Although all groups show a gradual decrease
in the openness of the vowel with age, New Yorkers of Italian
background show the greatest decrease for each age level. This ethnic
differentiation shows up in all but the lowest social group, which
has not been drawn into this process.

TABLE 9.10.
DISTRIBUTION OF (eh) BY AGE AND
ETHNIC GROUP IN NEW YORK CITY

Age Jews Italians Blacks

8-19 22 20 24
20-39 23 19 28
40-49 27
18 33
50-59 29

Source: Labov 1966a:357.

When we speak of ethnic identity as a part of the social context

of a linguistic change, it immediately raises the question of a sub-

stratum. Is the underlying parent language the cause of this differen-

tiation? The Martinet view would lean toward this explanation, since
we are considering outside impact as reflected in "the pressure of
another language." But the language-contact explanation fails by

19. Current studies of Maxine Berntsen in Phaltan are designed to examine the social
distribution of nonstandard features of Marathi, using many same techniques
of the
as in the New York City studies. Berntsen's preliminary findings show that education
is now a more important determinant of linguistic behavior than caste membership.
298 SOCIOLINCUISTIC PATTERNS

itself. Italian has no and first-generation Italian-Americans tend

[ae],

to use their native Italian low vowel [a] for the class of English short
a words. It would then appear that the second-generation Italian
tendency to raise short a is not the response that a structural version
of contact theory would predict: it reverses the direct influence of
Italian. Yiddish has no [ae] either, but first-generation Yiddish
speakers tend to raise English short a to [e]. Second-generation Jewish
speakers of English have somewhat less tendency to raise this vowel
than the Italians. This result can be interpreted in the light of other
sociolinguistic data on second-generation speakers, who reach for
native status by removing themselves as far as possible from the
low-prestige pattern of their parents. This is another version of the
"hypercorrect" or reverse influence that appears to play a major role
in linguistic change.
We now see that linguistic change may also be differentiated by
its association with a particular ethnic or caste group, and that
various ethnic groups may treat the same variable
in different ways.
An abstract treatment of these data might decide to ignore the
still

association with caste and ethnic group as beneath linguistic notice.

—
The second approach contracting one's focus to homogeneous
—
groups is now somewhat more difficult. A group of working-class
men will no longer do; we must be sure to insist on working-class
Italian men.

Local Identity
and caste
In addition to the crossclassifying assignments of class
status,communities often develop more concrete categories by which
individuals are placed. In rural communities (or in urban villages),
local identity is an extremely important category of member-
—
ship one which is often impossible to claim and difficult to achieve.
In many small towns of New England, there is a large subcategory
of "summer people." Then there are "incomers" or "outsiders,"
people who have moved into the town permanently but who are
shouldered aside for many decades before they are accepted. There
are immigrant groups, like the "damn Portagees," Indians, blacks,
and other ethnic groups who are not immediately accepted into the
main stream. The eighth- or tenth-generation Yankees form the core
of the local population, but members of other groups may gradually
assume local identity.
On the island of Martha's Vineyard (Ch. 1), it was this network
The Social Setting of Linguistic Change 299

of social categories which was most closely correlated with the

linguistic change in —
progress more important than occupation,
geography, education, or sex. In the final analysis, the sound change
being studied was associated with the assertion "I'm a Vineyarder."
The Martha's Vineyard study focused on the relation of social factors
to linguistic change; it demonstrated that the direction and develop-
ment of this change could not be understood without relating it to
the basic categories of local identity.

The Transformation of Regional Dialects into

Urban Class Dialects
Wyld observed that a regular pattern in the development of English
was the transformation of rural, regional dialects into class dialects
in the cities. This process often involves the movement of rural
speakers into low-prestige urban occupations, and into rapidly
growing ghettos.
When the rural speaker arrives in the city, he usually finds that
his country talk is Even if it was a marker of local identity,
ridiculed.
and a source of prestige home, he may already have been con-
at
scious of the provincial character of his speech before he came to
the city. As a result, we often see a rapid transformation of the more
salient features of the rural dialects as speakers enter the city.
In the United States, the movement of the southern black popula-
tion into the northern cities has brought about the creation of a
uniform caste dialect —
the black English vernacular of Harlem and
other inner cities.Black speakers in smaller communities, unaffected
by this process of dialect swamping, tend to participate in the lin-
guistic changes taking place around them. But in the larger ghetto
areas, we find black speakers participating in a very different set
of changes bearing no direct relation to the characteristic pattern
of the white community (Labov et al. 1968).
First we find the extreme rural forms stigmatized: lexical items
such as tote, verbal auxiliaries such as done or liketo (= 'almost')
are used less often. The characteristic vowel patterns of the regional
southern dialects are modified. These differences are so extreme that
the sound [dae] can mean 'die' in Alabama but 'deer' in South Caro-
lina. But such radical sound shifts are levelled out in the northern
cities: The /ay/ of die is pulled back to [da e], [da:' £ ] or [dae* :], and
deer is [di: 8 ] or [de: a ]. The basic phonological patterns which intersect
—
with grammar consonant cluster simplification, r and I vocalization,
300 SOCIOLINGUISTIC PATTERNS

—
copula deletion remain remarkably constant throughout the north-
ern ghetto areas. Inflectional morphemes which are absent in the
original southern black English, like third singular -s, remain absent.
Syntactic patterns of southern colloquial speech are maintained or
extended, as with negative concord and inversion (Ain't nobody see
that). The end result is the black English vernacular— a consistent
caste dialect, relatively constant for speakers from 6 to 20 years old,
with remarkable geographic uniformity and resistance to standard
English importations in the school system. Speakers in Boston,
Newark, Cleveland, Detroit, Chicago, New Orleans, St. Louis, Hous-
ton, San Francisco, and Los Angeles show these grammatical patterns
with startling regularity.
Similar processes appear to be operating in other languages,
wherever large capital cities are developing at the expense of the
hinterland. In the traditional literature described in Sect. 1, the social
setting of language change is discussed in terms of the spread of the
prestige patterns of urban capitals such as London and Paris. The
creation of low-prestige working-class dialects is a pattern of equal
linguistic interest; it embodies two major linguistic trends of the past
several centuries: the decline of local dialects and the growth of
vertical stratification in language.
This rapid language mixing seems to follow a kind of classic
structural reductionism, and it would not be difficult to argue that
it is a subtype of the same process that produces contact languages.

As the history of various Creoles shows, the rapid result of sudden

contact of two dissimilar structures is frequently the lowest common
denominator of both, with a strong push towards inflectional sim-
plification (see Hymes 1971 for various views on pidgin genesis and
Bickerton 1971 for an opposing position). One of the universal con-
straints on change seem to be operating here —
that in contact situa-
tions, mergers expand at the expense of distinctions (Herzog
1965:211). Yet if we are to apply our linguistic insight to predict such
mergers, we must first recognize the existence of heterogeneous
dialects, as the common, even normal form of language system. Not
every linguist is willing to do this. Wyld and Kokeritz tried to explain
the realignment of the -ea- words as the result of the upward move-
ment of a regional dialect within the London system. But Halle
preferred to abstract from any social evidence on this point (1962)
and argue the history of mate, meat, and meet as if they were
elements in the homogeneous speech community constructed by
Chomsky and Martinet.
The Social Setting of Linguistic Change 301

The Role of Women

Gauchat's elegant and convincing study established the variability
of Charmey patois, the existence of change and the role
in progress,
of women in furthering linguistic change. In case after case, Gauchat
discovered that women used more of the newer linguistic forms than
men did (1905:205, 209, 211, 218, 219, 224-26).

1. The palatalization of 1 —» y was found variably among speakers

30 to 40 years old, and regularly for those under 30. Above 40
years old, only women showed this trait. 20
2. 9 — > h variably for postverbal pronouns so that for veux-tu
[vu6o^> vuho] among the youngest generation, under 30, "espe-
cially women."
3. a —» a is variable in the oldest generation, and women apply the
rule more than the men. Laurent Rime, 59, pronounced douce as
[da°9a]; his wife Brigide, 63, said [ddGa].
4. o —> a , variably in the middle generation, regularly in the young-
est."As always, women take this route more readily . . . than
men."

Gauchat reinforced his finding by citing other examples from the

history of French in which the women of Paris were portrayed as
initiators of linguistic changes. We can point to similar behavior in
the evolution of New York City English, and here the pattern of sex
differences is even richer. In case after case, we find that women
use the most advanced forms in their own casual speech, and correct
more sharply to the other extreme in their formal speech. Table 9.11
compares men and women in the raising of tense short a, (eh), in
three styles. In casual speech, women use more of the high vowels
around (eh)-10 =
and show a modal value around (eh)-20, or
[iV],
[e:" ]. But in reading lists of words with (eh) items, women shift to
8

the opposite extreme, with a modal value around (eh)-40, = [ae:].

Men, on the other hand, shift their modal value only one notch, from
(eh)-22-26 to (eh)-27-32. Our instrumental studies confirm these im-
pressionistic tables: women
whole generation further
are almost a
along in the raising of (eh) than the men (Labov, Yaeger, and Steiner
1972; Ch. 3).
We find the same pattern in Detroit, where women clearly lead
in the more extreme raising of this vowel, and in Chicago as well.

20. Gauchat reports that one woman, 63 years old, pronounced her list of 1 words
regularly with y: viyo (veclu), Pyare (plorat), byatse (blanca) etc.
302 SOCIOLINGUISTIC PATTERNS

TABLE 9.11.

COMPARISON OF MEN AND WOMEN FOR (eh) AND (oh)

IN NEW YORK CITY

Styh e*

A B C

Variable Men Women Men Women Men Women

(eh) 10-13
— 1
— — — —
14-18 1 4 — 2 1
—
19-21 3 10 3 9 1 2
22-26 4 6 7 9 — 5
27-32 3 4 11 12 8 9
33-39 4 4 3 5 4 14
40-42 1 2 — 6 4 16

16 33 24 43 18 46

(oh) 10-13 3 4 — 1
— 3
14-18 3 10 4 10 2 5
19-21 4 14 7 7 5 13
22-26 3 5 8 16 5 4
27-32 3 4 5 10 5 8
33-40 1 1 — 2 1 10

17 38 24 46 18 43

Source: Labov 1966a:313.

* Contextual styles: A Casual speech, B Careful speech, D Reading word lists

Fig. 9.2 shows the extreme rotation of the vowel system in Chicago
which we cited as an example of young working-class speakers; again
it is the women in this group who show the more extreme forms.

Throughout Shuy's studies of Detroit speech (Shuy, Wolfram, and

Riley 1967), we find the same pattern that women show more —
shifting than men. Shuy's original report showed a greater shift
towards prestige forms in formal style, and our current spectro-
graphs studies confirm the advanced position of women on the
vernacular raising of /ae/ and fronting of /a/.
Why do women do cannot be only their sensitivity to
this? It

prestige forms, since that explains only half of the pattern. We can
say that they are more sensitive to prestige patterns, but why do they
move forward faster in the first place? Our answers at the moment
are not better than speculations, but it is obvious that this behavior
of women must play an important part in the mechanism of linguistic
change. To the extent that parents influence children's early Ian-
The Social Setting of Linguistic Change 303

guage, women do so even more; certainly women talk to young

children more than men
do, and have a more direct influence during
the years when children are forming linguistic rules with the greatest
speed and efficiency. It seems likely that the rate of advance and
direction of a linguistic change owes a great deal to the special
sensitivity of women to the whole process.
It would be a serious error to construct a general principle that
women always lead in the course of linguistic change. The central-
ization of /ay/and /aw/ on Martha's Vineyard was found primarily
in male speakers; women showed a much weaker tendency here.
Trudgill (1971) shows in Norwich that women are more influenced
by standard forms than men, but that men are in the lead in the
use of new vernacular forms in casual speech. This seems to be
generally true for a number of sound changes taking place in English
cities. The correct generalization then is not that women lead in

linguistic change, but rather that the sexual differentiation of speech

often plays a major role in the mechanism of linguistic evolution.
We now faced with data that pose even greater difficulties for
are
those who claim that language change is independent of social
variation. The sexual differentiation we are dealing with clearly
depends upon patterns of social interaction in everyday life. The
differentiation of men and women cannot depend upon weaknesses
in communication networks as Bloomfield would have it, or the law
of least effort as Sweet would argue. In the communities we have
been studying there is no barrier to men and women talking to each
other, and no reason to think that women are lazier than men. If
anything, they put more effort into speech, as we observe from the
extreme fronting in women's vowel systems in current spectro-
graphic work. We are dealing with some positive factor here, oper-
atingupon a subtle set of conventional social values. There are of
course physical differences between the vocal tracts of men and
women to be taken into account, and the shorter length of women's
vocal tracts does predict higher formant positions. But I. Mattingly
demonstrated that there must also be conventional, social factors
involved in the differentiation of men's and women's speech. A
review of Peterson and Barney's data on vowel identification showed
a relatively low correlation of these formant differences from one
vowel to another: a purely physical explanation would produce
comparable shifts in all vowels (1952). Our spectrograph^ studies
show that in many dialects the difference is much more than an
304 SOCIOLINGUISTIC PATTERNS

upward shift: women use a wider range of formant positions, over-

lapping men's formants in all directions, with much greater distances
between vowel locations. The sexual differentiation of speakers is
therefore not a product of physical factors alone, or of different
amounts of referential information supplied by speakers, but rather
an expressive posture which is socially more appropriate for one
sex or the other. On Martha's Vineyard, men are more "close-
mouthed" than women, and use more contracted areas of phono-
logical space; conversely,women in New York City and Philadelphia
use wider ranges of phonological space than men with more extreme
lip-spreading, lip-rounding, for vowels, more blade-affrication and
palatalization of consonants.

Restructuring of Early Patterns under Peer-Group Influence

Most models and studies of the acquisition of language take the

mother-child interaction as the social context of language learning
(Brown and Bellugi 1964; Bloom 1971). The influence of other chil-
dren and peers outside of the family is usually not considered. Thus
we do not know who Adam and Eve played with in Brown's studies,
and when Adam suddenly turns up with negative concord to both
indefinites and verbs at the age of four, it is interpreted as a purely
internal development since these forms are not heard within his
family. Halle's model of linguistic change is built entirely upon
parent-child interaction: the parent adds new rules to his grammar
and the child forms a new grammar which incorporates
later in life,
this more complex structure into a simpler one (1962).
The difficulty with this model is that children do not speak like
their parents. In the great majority of cases that we have studied
or encountered, children follow the pattern of their peers. I can cite
a great many examples from my own observation, and so can every
linguist who has looked into the matter. The evidence of dialectology
is strong on this point. Though on principle dialectologists prefer
third-generation residents in a given area, it is rare to find a second-
generation speaker showing the effect of his parents' alien rules. In
the study of the Lower East Side (Labov 1966a) the great majority
of informants were second-generation New Yorkers, but the regular
evolution of the New York City vowel system appeared in their
speech in the same way as with third- and fourth-generation subjects.
The critical cases were those who had moved into New York at an
early age. If we consider the formative period for a native speaker
The Social Setting of Linguistic Change 305

to be roughly 4-13 years, reasonable to assume that a person

it is

had to spend at least half of those years inNew York to acquire the
New York City pattern. This proved to be the case: those who entered
the United States after the age of 8 stood out as exceptional cases
in any table or graph. For those who moved into New York City from
another dialect area in the United States, the critical cut-off period
appeared to be at 10 years.
Restructuring an oddly unspecified notion. Are the early rules
is

learned from parents simply abandoned, unlearned, rewritten? Or

are they set aside but remain as potential rules of the grammar? It
is also possible that restructuring never takes place within the native

language itself: that no one unlearns vernacular rules, but simply

adds new ones. It may be that the dialects learned from peers are
only those rules that were not specified in the grammars learned from
parents. A detailed empirical study of this process of restructuring
is therefore required to resolve such questions if we are to under-

stand how the basic vernacular changes.

A strategic research site for such a study is a community with
many families moving in from other dialect areas of equal or higher
prestige. Here we would be able to observe the gradual restructuring
or additional development in the dialect patterns of children who
enter the community with a dialect already formed to a certain
extent. One such pilot study was carried out in the fifth grade in
Radnor township, a suburb of Philadelphia in which roughly half
of the parents come from other dialect areas. I have studied the
phonological patterns of several self-selected peer-groups of 11-
year-old boys; the relevant linguistic patterns of one such group are
shown in Table 9.12. Jim and Charlie have local parents and were
raised in Radnor; Ken came from Worcester, Mass., at the age of 8;
Tim came from Cleveland at the age of 7. Ken and Tim are two of
—
the most popular and prominent members of the class Tim is class
—
president and Charlie too is among the best at a number of sports.
Table 9.12 shows that Ken and Tim have learned the specific
Philadelphia rules for fronting /uw/ and /ow/; they show the Phila-
delphia area's sharp difference between the central allophones used
for most vowels and the back vowels before /l/, which are not
fronted. Furthermore, Tim has acquired the backed and centralized
allophone of /ay/ before voiceless obstruents, with the sharp con-
trast between this and the somewhat fronted nucleus when final. On
the other hand, neither Ken nor Tim has picked up the complex
306 SOCIOLINGUISTIC PATTERNS

TABLE 9.12.

ACQUISITION OF PHILADELPHIA PHONOLOGY

BY FOUR PREADOLESCENT BOYS

Jim W. Charlie C Tim M. Ken A.

Age 11 11 11 11
Born in Radnor, Pa. Radnor, Pa. Cleveland, O. Spring-
field,

Moved to Radnor at the Mass.

age of — — 7 8
Tensing and raising of
short a: before
voiceless fricatives 11/12 4/7 0/7* 1/5
(anterior)
nasals (anterior) 19/19 4/4 3/3 6/6
voiceless stops 0/9 1/7 0/2 0/2
Centralization of /ay/
before:
voiceless consonants 19/19 11/11 12/12 0/2
voiced and final 0/22 3/9 0/4 0/2
Distinction of low back
vowels
short open /p/ phones a' , a a' , a a n
long open /p/ phones o~, o~ d", o" D t>

Fronting of high back

vowels
/ow/ nuclei 8 a a a
/owl/nuclei
/uw/ nuclei -u- -H- +t -t*

/uwl/ nuclei u u u u

*Underlined items show original dialect pattern preserved and different from Philadelphia

conditions for the tensing of short a in Philadelphia. They do not

tense a before front nasals, voiceless fricatives, and mad-bad-glad
before /d/, excluding function words and irregular verbs with due
regard to following morpheme boundaries, as good Philadelphians
do, but follow much simpler patterns characteristic of their original
areas: tensing and raising before nasals, without regard to function
vs. lexical words, closed or open syllables, or any of the other
deep-seated apparatus of the Philadelphia rule. Furthermore, Ken
has not reversed the eastern New England merger of short and long
open o in hock and hawk, even after three years. The higher-level
rulesseem to remain fixed, while the lower-level conditions seem
The Social Setting of Linguistic Change 307

to have shifted to the Philadelphia pattern. The merger of hock and

hawk is its application, and not dependent on any
quite general in
prior rules; when complete, there is wholesale restructuring in
it is

the lexicon which is not easily unlearned. The tensing rule for short
a is relatively abstract, and is followed by a number of rules, includ-
ing the raising rule, the formation of interdental flaps, and the dele-
tion of grammatical boundaries.
When children move to a new dialect area at the age of 3 or 4,
they seem to adopt the basic patterns of the new area. But we do
not have any systematic studies on this point, and there are many
open questions. For the Radnor fifth-graders, there is no evidence
for restructuring. It is when
younger child falls under
possible that a
the influence of his peer group, he has not yet formed most of the
rules which differ from one dialect to another, and that he merely
adds the rules that do not conflict with his own. But it is also possible
that 6-year-old children will actually abandon one set of rules for
another. This is certainly a critical area for further research.
A case where family
influence outweighs the peer group has been
citedby Kostas Kazazis from personal experience (1969). It involves
Athens-born middle-class teenagers whose parents or grandparents
came from Istanbul. The Istanbul dialect differs prominently from
Athenian Greek: one salient point is the use of the accusative for
indirect objects, rather than the genitive. Though the Istanbul ado-
lescents mixed and were under considerable
freely with the others,
pressure to change their speech, they did not do so. The strength
and prestige of Istanbul family ties and the value of Istanbul identi-
fication seem to have been great enough to resist such pressures.
Yet the forces pressing for uniformity are apparently quite general.
The survival of distinct dialects in neighboring Swiss German vil-
lages an interesting case in point. Enderlin (1913) reported from
is

Kesswil that up to 40 percent of the wives came from other villages,

and thus spoke a different dialect. They were sharply ridiculed for
their outlandish speech, and soon adopted most of the features of
the Kesswil dialect. Here the remarkable heterogeneity of the area
as a whole is supported by pressures for homogeneity within the
village.
We
must recognize, in any case, that there are many kinds of
prestigeand many kinds of support for minority dialects. Numbers
alone do not account for the direction of linguistic change, nor does
mere frequency of interaction.
308 SOCIOLINGUISTIC PATTERNS

5. The Evaluation Problem:

Subjective Reactions to Linguistic Change

In our study of the embedding of linguistic change in its social

matrix, we have seen many examples of the new form being ad-
vanced most rapidly in a particular class of speakers, often a lower-
status group, and spreading outward in a wavelike fashion from that
center. If we consider that the other groups are "borrowing" the new
form from the originating group, we must question Tarde's law (1913)
that borrowing always takes place from higher to lower prestige
groups. If Bloomfield was right, the solution to the evaluation prob-
lem would be simplistic: people would simply imitate behavior
associated with their superiors. They do not; in fact, the primary
sociolinguistic problem was posed for me by an upper-middle-class
woman who said, "Why do I say [d1 ] when I don't want to?"
Even the notion of "borrowing" falls short when we consider two
other major factors in linguistic change. We have seen in a number
of areas that women adopt the newer forms more readily than men,
and noted the strong effect of preadolescent peer groups in changing
the language of its members. These objective correlations strongly
argue that some machinery of social interaction is at work that
cannot be the product of simple structural pressures or simple
imitation. It seems that social variation does play a systematic role
in linguistic change; to see how, we must see what social information
is communicated by these variations. The notion of "prestige" must

be defined in terms of the people using it and the situation in which

it is used; that is, brought out of the area of speculation and made

the focus of empirical investigation. Other dimensions of expressive

social information carried by incoming linguistic forms must be
explored as well.
Not every linguistic change receives overt social evaluation or even
recognition. Some seem to lie far below the level of overt social
reactions, as with the changes studied by Gauchat. The speakers of
Charmey were unwilling to believe in these differences even when
they were confronted with them: "Nous parlons tous la meme chose!"
(1905:202)
The same can be said of the response of Vineyarders who were
confronted with changes in their pronunciation of /ay/ and /aw/.
When the islanders talk about language, they concentrate on special
marine jargon ("talking salty") rather than systematic rules of pro-
The Social Setting of Linguistic Change 309

nunciation. 21 It is more difficult for grammatical changes to escape

public notice, but that is the case with the midwestern development
of the positiveanymore, meaning 'nowadays', in "That's the way
with airplanes anymore." This regional feature is strongly en-
it is

trenched throughout the Midland area, but most speakers are quite
unconscious of it. I can record the following typical exchange from
Cleveland:
W. L.: Around here, can you say, "We go to the movies
anymore?"
Salesgirl: No, we say "show" or "flick."

As striking as this extension of the indefinite may seem to a linguist,

it isbeing accomplished without public notice. A story caption in
a 1969 issue of Life read, "What it Takes to be a Lady Author Any-
more." But I could capture no spontaneous reaction to this from any
readers I met.
There are a number of systematic changes taking place in the
English of the western United States that have no evident social
significance. In most areas, the unconditioned merger of the low back
vowels in hock and hawk, which affects two very large word classes,
proceeds without notice or comment. Cook shows that the fronting
of (aw) in Salt Lake City is of little social concern, and shows almost
no stylistic shifting (1969). In our own recent studies in Utah, we
have observed an on-going merger of a number of vowels before
final /l/, so that fool = full, feel — fill, and further pairs merged in
some working-class areas. But these mergers are not reported by
anyone and seem to make no impact on social consciousness.
To sum up, incoming linguistic changes rarely rise to the level of
social comment in their initial stages, and not all changes become
the focus of conscious attention even in their advanced stages. But
there is much more to social evaluation than the overt responses that
native speakers can summon The problem
as posed so far is
up.
governed by the superficiality of the observations. The traditional
literature on social evaluation is limited, on the one hand, to anec-

21. There is some indication of a social awareness of Vineyard styles of pronun-

ciation. One examples of centralization of /ay/ and /aw/ is a young
of the strongest
man who went and returned to develop several businesses by the docks
to college
of Chilmark. His mother remarked that he only started to talk like the men on the
docks when he came back, but she wasn't able to point to any particular feature of
his pronunciation (See p. 31).
310 SOCIOLINGUISTIC PATTERNS

dotal evidence of overt reactions to gross stereotypes, and on the

by linguists as to efficiency or economy. In the
other, to speculation
past two decades, however, considerable progress has been made
in measuring unconscious social reactions to language through the
"matched guise" technique.
The work of Lambert, Tucker, and their associates at McGill
University has provided us with a firm methodology and a number
of empirical principles for the study of subjective reactions (Lambert
1967).The basic technique consists of exposing listeners to a series
which include bilingual speakers recurring
of tape-recorded excerpts
in "matched guises." These guises contrast English and Canadian
French (Lambert et al. 1960), Canadian French and continental
French (Preston 1963), Arabic and Hebrew (Lambert, Anisfeld, and
Yeni-Komshian 1965), network English, southern black and southern
white English (Tucker and Lambert 1969), and so on. The listeners
record their judgments of the speakers in terms of a series of person-
ality traits: intelligence, honesty, reliability, ambition, sincerity,
kindness, sociability, sense of humor, and so on. The differential
rating of the same speakermatched guises gives us a measure
in his
of the listener's unconscious social evaluation of the two dialects
or languages, and a few central factors can be isolated from the many
attributes tested.
From the work of Lambert and his associates we can derive several
important principles:

1. Subjective evaluations of social dialects are remarkably uniform

throughout the speech community. Canadian French speakers
agree with English speakers in rating their own language lower
on most of these personality traits: in the French guise the speaker
was heard as less intelligent, less reliable, etc.
2. Evaluations of language are usually not available to conscious
elicitation, but are readily and consistently expressed in terms of
personality judgments about different speakers (Lambert, Anisfeld,
and Yeni-Komshian 1965). Thus a study carried out in Texas of
several well-established sociolinguistic variables failed to record
any significant reactions because all of the 16 dialect variations
constructed were recorded by the same speaker (Baird 1969).
3. All subjects acquire these norms in early adolescence, but
upper-middle-class children show a stronger and more permanent
The Social Setting of Linguistic Change 311

Brown extended the matched-guise technique to socio-

reaction.
economic differences in Canadian French (1969), concentrating on
teenage judges. His findings give "strong support to the idea that
speech is expressive of motives and values." Judges from all
backgrounds favored upper-class speakers over working-class
speakers of French, and English over French, on a general factor
of "competence." Boys from high-prestige schools showed ratings
closer to adult judges than other boys, indicating that adult norms
are acquired earlier among upper-middle-class youth. 22

Lambert is primarily concerned with the effect of underlying social

values upon the bilingual child, the language learner, and the bilin-
gual community. These matched-guise tests do not contrast individ-
ual features of language, but record undifferentiated responses to
the language or dialect as a whole. A comparable series of subjective
reaction tests carried out in New
York City differed in two respects.
The speakers recurred in the test series with sentences which were
the same referentially, and in their own dialect: but with different
values of one linguistic variable. Second, the scales used for meas-
uring reactions were drawn from situations which commonly arise
in everyday social interaction: "What is the highest job this person
could hold, speaking as he does?"
The New York City studies (Ch. 6; Labov et al. 1968) gave full
confirmation to the three principles which emerged from Lambert's
work. One further principle appeared:

4. Speakers who use the highest degree of a stigmatized feature in

their own natural speech show the greatest tendency to stigmatize
others for their use of this form.

We can see this principle operating in regard to the raising of short

a, and long open o, the (oh) variable. As noted
the (eh) variable,
above, the Italians show
a greater tendency to raise the front vowel,
while the Jews favor the back vowel. In subjective reaction tests,
Italians stigmatize a speaker in his high (eh) guise (saying [be: a d] for
bad, etc.), this feature more consistently than other
and downgrade
groups. Since New
Yorkers show a high level of agreement on this
point, the differences are small, but Table 9.13 shows that Italians

22. On the other hand, French speakers were scored higher on a number of attributes
which can be analyzed as containing the factor of "benevolence".
312 SOCIOLINGUISTIC PATTERNS

are somewhat more consistent than the Jews. 23 The socioeconomic

differentiation of judges in subjective reaction tests follows the same
principle. Whereas working-class and lower-middle-class speakers
show the greatest tendency to raise the vowel of bad, ask, dance,
etc.,they are also most consistent in stigmatizing this behavior, as
shown in Table 9.14. In both Table 9.13 and Table 9.14 we see that
younger speakers show a definite increase in sensitivity to this
well-established linguistic variable, just as they show a greater ten-
dency to use the stigmatized form (see Table 9.10).
TABLE 9.13.

PERCENT STIGMATIZING
RAISED (eh) IN SUBJECTIVE
REACTION TESTS BY AGE
AND ETHNIC GROUP
Age Level Jews Italians

20-39 86% 100% N

40- 81 88 "
14 6
All ages 82 91 28 15

TABLE 9.14.
PERCENT STIGMATIZING RAISED (eh) IN
SUBJECTIVE REACTION TESTS BY AGE AND
SOCIOECONOMIC CLASS

Socioeconomic class

Age 0-2 3-5 6-8 9

8-15 100% 75% 100% (100)%

7 8 6 1
16-19 86 100 100 75
7 12 4 4
20-39 75 90 100 80 4 10 11 5
40- 75 80 70 71 16 15 10 7

The most dramatic shift in subjective reactions which we have

recorded response to an incoming prestige feature: the pronun-
is in
ciation of final and preconsonantal /r/ in New York City. Fig. 6.2
(p. 152) showed the pronunciation of /r/ by New Yorkers in casual

23. There is an interaction here between class and ethnic group. The Jewish sub-
group shows greater upward mobility, with a heavier concentration of lower-middle-
class speakers, while the Italian sample shows a greater concentration of working-class
subjects. Ethnic membership appears to be the predominant influence, however, in
the tendency to favor the raising of (eh) or (oh).
The Social Setting of Linguistic Change 313

speech (the same data as Table 9.5) as compared to their subjective

reactions to this variable in the speech of others. In casual speech,
those over 40 show a sprinkling of r's without any particular direc-
tion; those under 40 show a sudden increase in stratification, with
the highest-status group using [r]. In the subjective reaction tests,
we see a shift from random response over 40 years old to 100 percent
agreement for those under 40: they all unconsciously rate the same
speaker higher on the job scale when [r] is pronounced than when
it is not.
The subjective reaction have been reactions to
tests cited so far
dominant
a single job suitability scale, operating strictly within the
middle-class set of values. This brings us back to the original ques-
tion: why don't people behave in away consistent with the normative
values that they express? There are four possible answers that we
might consider:

1. They are too lazy or careless to use the norms that they recognize.
2. Differences in communication patterns mean that lower-class
speakers would not be aware of the subjective norms of upper-
class speakers.
3. Even if lower-class speakers do learn the norms, they do not do
so until it is too late for them to acquire consistent productive
control of the prestige forms.
4. Lower-class speakers do not want to adopt the norms of the upper
class; although they do endorse the dominant norms in the test
situation, there are opposing sets of values that support the ver-
nacular forms, and that do not appear in subjective reaction tests.

We can reject (1) as a product of class bias; there is no reason to

think that any one class has a monopoly on laziness. The evidence
cited above disqualifiesThere seems to be some support for (3)
(2).

in the differential age atwhich speakers acquire subjective norms,

and the differential clarity or strength (Brown 1969, Labov 1966a).
We find strong support for (4) in our recent work on subjective
reaction tests in Harlem, where we introduced two other rating
scales: "fight" and "friend" (see Ch. 8, especially Fig. 8.3).
Granted that the social setting of linguistic change is a hierarchical,
stratified society, not all prestige forms sweep through the commu-
nity, and not all change from above succeeds. The best general
formulation of this opposition of values is still that of Ferguson and
Gumperz (1960):
314 SOCIOLINGUISTIC PATTERNS

1. Any group of speakers of language X which regards itself as a

close social unit will tend to expressgroup solidarity by favor-
its

ing those linguistic innovations which set it apart from other

speakers who are not part of the group.
2. Other things being equal, if two speakers A and B of a language
X communicate in language X and if A regards B as having more
prestige than himself and aspires to equal B's status, then the
variety of X spoken by A will tend towards identity with that
spoken by B.

Indicators, Markers, and Stereotypes

We can classify the various elements involved in linguistic change
according to the kind of social evaluation that they receive. Indi-
cators are linguistic features which are embedded in a social matrix,
showing social differentiation by age or social group, but which show
no pattern of style shifting and appear to have little evaluative force.
The merger of hock and hawk, the extension of anymore, may be
taken as cases in point. Markers such as (eh) or (r) do show stylistic
stratification as well as social stratification. Though they may lie

below the level of conscious awareness, they will produce regular

responses on subjective reaction tests. Stereotypes are socially
marked forms, prominently labelled by society. In the following
subsection, we will consider the linguistic changes brought about
by such labelling.

The Social Stigmatization of Linguistic Forms (Stereotyping)

A social stereotype is a social fact, part of the general knowledge
of adult members of the society; this is true even if the stereotype
does not conform to any set of objective facts. Stereotypes are
referred to and talked about by members of the speech community;
they may have a general label, and a characteristic phrase which
serves equally well to identify them.

1. "Brooklynese." A stereotype which is based on earlier forms of

working-class New York City speech, without regard to geography.
A characteristic label is toiiy-toid street, based on the older form
of the preconsonantal mid central vowel with a palatal upglide,
[a'].

2. "Deses, dems, and doses." General characteristic of working-class

speech in the United States, based on the use of interdental stops
for interdental fricatives.
The Social Setting of Linguistic Change 315

3. "Bostonian", often labelled as "Pahk your cah in the Hahvahd

Yahd", based on ther-less Boston pattern with fronted low central
vowel [a:].
4. "Broad a" pronunciation, particularly for the words aunt [ant] and
bath [ba9] sometimes caricatured as "Fahncy that" [fansi Sat]
attributed to upper crust British and New England speakers.
5. "Southern drawl," based on various imitations of southern
monophthongs, long and ingliding vowels, and in particular the
word Y'all [jt>:l].

6. "Outer Banks" of North Carolina, known regionally as "Hoi

Toiders;" based on the pronunciation of high tide as [ho 1 tD x d],
and the general raising and backing of /ay/.
7. "Put the harse in the born," a standard way of referring to the
stigmatized stereotype of rural Utah speakers, alleged to reverse
horse and barn and other /ahr/ and /ohr/ words.
8. "Parigot," working-class Parisian speech, based on such features
as the absence of ne, backing of /a/, and often characterized with
j'saispas [sepo] or [spo].

This list of examples shows the variety of relations to fact and the
variety of social values associated with stereotypes. Some stereo-
typed features are heavily stigmatized, but remarkably resistant and
enduring, like dese and dose. 24 Others have varying prestige, posi-
tive to some people and negative to others, like Bostonian or south-
ern drawl.
some of these stereotypes has led to rapid
Social stigma applied to
linguistic change, with almost total extinction. A good example is
the [a 1 ] pronunciation in New York City. Table 9.15 shows how
systematically and how thoroughly this feature is being extinguished
in New York. For those born before World War I, it is a regular
feature for all but middle-class speakers. For those born after World
War II, it is found only among lower-class speakers.
In the last stages of the decline of a stigmatized variable, it may
appear as a form of ritual humor. One example of this process that
we can document over a long stretch of time is the Cockney confu-

24. In the United States, all native speakers show the ability to discriminate the
word classes of /d/ and pronounce the fricative in isolated words. But
/S/, and to
the strong Gaelic substratum of Ireland produces a different result. Witness the story
cited to me by Jerry Crowley of Cork: the school teacher who says, "Now today we're
going to study English pronunciation. And we're not going to say dese, and dem, and
dose.' We're going to say dese, and dem, and dose/.'"
316 SOCIOLINGUISTIC PATTERNS

TABLE 9.15.

PROPORTION USING ANY [ai] IN INTERVIEWS

BY AGE AND SEC CLASS IN NEW YORK CITY

Age 0-1 2-3 4-5 6-8 9 0-9

8-19 2/7 0/11 0/12 0/16 0/5 2/51 04

20-39 3/4 3/7 3/10 1/11 0/7 10/39 24
40-49 1/3 5/14 4/8 4/13 0/4 14/42 33
50-59 3/3 2/4 3/3 2/4 0/3 10/17 59
60- 7/7 5/5 1/1 13/13 100
8-60 16/24 15/41 11/34 7/44 0/19 49/162
% 67 38 32 16 00

sion of /v/ and /w/. Most Americans know this social variable only
through Dickens' 1837 version of the speech of Sam Weller and his
father. It is evident that the v/w alternation was already a stereotype,
since Dickens was using it as heavily as possible to characterize the
Wellers, exchanging almost every /v/ and /w/. 25
This here money . . . he's anxious to put somevers, vere he knows it'll be
safe,and I'm wery anxious too, for if he keeps it, he'll go a lendin' it to
somebody, or inwestin' property in horses, or droppin' his pocket book
down a airy, or makin' a Egyptian mummy of hisself in some vay or another.
— Pickwick Papers, p. 800.

The by Henry Machyn three centuries earlier showed

spelling used
was then a regular characteristic of the respectable
that this variable
middle class: wacabondes 'vagabonds', waluw 'value', wue 'view',
welvet, woyce, vomcm, veyver 'weaver'; Volsake 'Woolsack',
Vetyngton 'Whittington', Vosseter 'Worcester' (Wyld 1936:143). The
/v~~w/ merger was apparently well-stigmatized in the 18th century;
Walker says that it is a "blemish of the first magnitude" that occurs
among Londoners, "not those always of the low order". Eighty years
later it was dead in London. Wyld notes that he never heard it

actually used in the 1870's, but he did hear middle-aged people use

25. The confusion of oil and Earl in New York City ("Brooklynese") forms a
stereotype quite parallel to the v — w confusion cited here. Though there are speakers
who use [oil] for the Earl class, it was more common to use a centralvowel [ail] for
both, or to keep the two classes apart by preserving a rounded vowel in the oil class.
But to outsiders, the effect of a merger is that oil sounds like Earl and Earl sounds
no one notices those variants that are the same as
like oil. If there is free variation,
the standard; it is the pronunciations which differ from the standard pattern which
are noticed and remembered.
The Social Setting of Linguistic Change 317

the ritual joke of saying weal for 'veal' and vich for 'which'; a joke
that he never understood until he read Dickens. Long after it is
actually extinct in speech, a linguistic variable can survive as a
stereotyped use of certain words, then as a standard joke, and finally
as a fossil form whose meaning has been entirely forgotten.

6. The Actuation Problem

Though we have assembled a fair amount of data in response to

theembedding and evaluation problems, there is comparatively little
that can be said about the particular social or linguistic events that
trigger a particular change. We can point to some general circum-
stances that are not irrelevant to the temporal location of some
linguisticchanges discussed above. For instance, the reversal of New
York City attitudes towards r-pronunciation can be seen as merely
one prominent feature of a general shift away from British and New
England models in favor of a general American broadcast standard.
At some point in time, the older prestige dialect was redefined;
instead of an "international standard" it became a "regional peculi-
arity". This event seems to coincide with the period of World
War II, and one might argue that the experience of men in the serv-
ice was somehow involved. It would be difficult to prove; all we
can do time is to point to the war as the most prominent
at this
social disturbance which coincided with the period of the linguistic
change.
In examining the alternations of the New York City vowel devel-
opment, we cannot help being struck by the differentiation of the
Jewish and Italian ethnic groups in the raising of (eh) and (oh). The
raw materials for this process seem to have been present before 1890,
but the arrival of these two large immigrant groups about that time
may well have put new energy into the raising process. It appears
that the Jews defined the raising of (oh) in coffee, lost, and all as
the basic change, and the following shift of (ah) —> (oh) —> was
accordingly emphasized by Jewish speakers as a part of a chain shift.
Italian New Yorkers focused on the front vowel, continuing the
earlier tradition of treating this as theprimary variable involved.
We can see a parallel tendency on Martha's Vineyard. The earlier
of the two raising processes affected (ay) in the speech of the Yan-
kees, at a time when the Portuguese population was hardly consid-
ered to be a part of the community. After 1930, Portuguese more and
318 SOCIOLINGUISTIC PATTERNS

more took the place of the Yankees as merchants and politicians.

The Portuguese then began to use centralization in their own English,
but they raised (aw) to [a 1 more prominently than they shifted (ay)
]

to [e 1 ]. Their emphasis on the raising of (ay) would lead us to expect

further generalization, with the back vowel now pulling the front
one after it.
These are the kinds of fluctuations which Meillet referred to when
he characterized society as
an element, in which circumstances induce continual variation, sometimes
rapid, sometimes slow, but never completely suspended. (1905:16)

Observing the impact of the Jewish and Italian immigrations into

New York, we might look backwards at the entry of the Irish and
Germans in the 1860's; that movement might well have affected the
evolution of the city's dialect by reemphasizing elements already in
flux. At present, we are witnessing the rise of two new entering
groups, blacks and Puerto Ricans, who do not seem to participate
in or directly influence the rotations of the vowel system just de-
scribed. But it is possible that the influence of the black speech
pattern is acting to check or reverse the raising of the tense vowels
among white speakers.
As a basis for comparison in our studies of black youth in Harlem,
we selected two white working-class groups at the northern
(Irish)
tip of Manhattan, in Inwood, a fairly solid white residential area
where considerable hostility towards blacks was expressed. We
could not easily find a white working-class group in Manhattan with
less direct black or Puerto Rican influence (Labov et al. 1968). 26 The
intricate tensing rule for short a words, which selects vowels before
voiced stops and voiceless fricatives, front nasals, etc., is fairly intact
among members of the Inwood groups. For the general population
the tensing rule is followed by a raising of [ae:] to mid and then high
position, involving the successive mergers of /ash/ with /ehr/ and
then with /ihr/. That is, bad merges with bared in the early stages,
and then both bad and bared with beard. But the /aeh/ of the Inwood
group is not much higher than [ae:] and it has not merged with /ehr/.

26. The fact that a group expresses hostility towards another group does not
preclude the likelihood of linguistic influence. We have observed many cases of the
contrary effect: that white groups surrounded by blacks, and engaged in hostile combat
with them, acquire many linguistic features from them. Such a phenomenon can be
observed on the Lower East Side of New York City, or in Highland Park in Detroit.
The Social Setting of Linguistic Change 319

The merger of /ehr/ and /ihr/ in bared and beard, bear and beer,
iscomplete. This second merger is widespread among black speakers,
reflecting the coastal South Carolina pattern, but in BEV the raising
of /aeh/ is quite limited. 27 We
must therefore consider the high
probability of black influence in the disturbance of the New York
City pattern for the Inwood group. Examples such as these give
weight to Meillet's contention that explanations for the irregular
course of linguistic change are to be found in the fluctuating social
composition of the speech community.

7. The Place of Social Variation in the

Life History of a Linguistic Change
In the three preceding sections, we have assembled enough data
to give an answer to the first of the three substantive questions raised
in this chapter: does social variation play an important role in
linguistic change? A review of the asymmetrical history and evolu-
tion of most linguistic changes, and of their remarkably uniform
evaluation by society, shows that an asocial account would be
incoherent. The histories that I have outlined would not exist if social
differences were abstracted from the grammars involved, for the
accounts of change would then be vacuous. The overall view is best
expressed by outlining a typical life history of a sound change.
The change first appears as a characteristic feature of a specific
subgroup, attracting no particular notice from anyone. As it prog-
resses within the group, it may then spread outwards in a wave,
affecting first those social groups closest to the originating group.
Inevitably, the linguistic feature is associated with the expressive
characteristics of the originating group, with whatever prestige or
whatever social values are associated with that group by other
members of the speech community (Sturtevant 1947). Whether such
association is outward movement is difficult
sufficient to explain the
to say. We do growth of the affected area may be
know that the
checked by linguistic factors (Herzog 1965) or by social factors or
historical discontinuities (Bloomfield 1933:344), or by the negative
prestige of the group as a whole (as in the case of New York City). 28

27. In our studies of adolescent groups in south-central Harlem, we find that about
half show the merger of (ihr) and (ehr) in beer and bear, cheer and chair, etc. (Labov
et al. 1968).
28. For the negative prestige of New York City, see Labov 1966a, Ch. 13. The severely
limited scope of the New York City dialect area has been noted above.
320 SOCIOLINGUISTIC PATTERNS

At this point, the linguistic feature may be an indicator of age and

social distance from the originating group.
As the linguistic feature develops within the original group of
speakers, it becomes generalized in several senses. Over the course
of time (three or four decades) a wider range of conditioned sub-
classes may be involved, and more extreme (less favored) environ-
ments. Furthermore, the structural symmetry of the system leads to
generalization to other vowels or consonants, or members of the
same natural class, which begin to move in the same (or opposing)
direction. In the meantime, new social groups enter the community,
and through historical accident or the influence of their original
dialect, reinterpret the on-going change to reemphasize other ele-
ments in this complex system.
This discussion has not focused on the internal processes involved
(see Ch. 7, Labov 1972c), but it is important to note that structural

generalization in linguistic systems is far from instantaneous. It is

a slow process, with considerable time lag, and between the move-
ment of one item and the associated movement of another, several
decades with their accompanying social changes may have passed
(Chen 1971).
As the original change acquires greater complexity, scope, and
range, it comes to acquire more systematic social value, and is

restrained or corrected in formal speech (a marker). Eventually, it

may be labeled as a stereotype, discussed and remarked by everyone.

The future prospects depend upon the fortunes
of this stereotype
of the group it is associated with. Ifmoves into the main-
the group
stream of society, and is given respect and prominence, then the new
rule may not be corrected but incorporated into the dominant dialect
at the expense of the older form. If the group is excluded from the
mainstream of society, or its prestige declines, the linguistic form
or rule will be stigmatized, corrected, and even extinguished.
But such correction does not have the regular character of the
change itself. Instead, correction focuses irregularly on certain
prominent sounds or words, with consequent disturbances in the
regularity of the linguistic system. Although the New York City tense
back vowel is corrected and lowered in lost, coffee, water, and
chocolate, it is never noticed or corrected when it occurs as the first

part of the diphthong in boy or Lloyd.

The irregularity of this secondary social correction, compared with
the regularity of the original change, may provide part of the expla-
The Social Setting of Linguistic Change 321

nation for the massive irregularity and word-class splitting docu-

mented by Wang, Chen, and Hsieh in the history of Chinese dialects
(Cheng and Wang 1971; Chen and Hsieh 1971; Chen 1971). It is quite
probable that the original movements were more regular than the
final result indicates, if they were at all similar to the sound changes
in progress which we are now studying by instrumental means. But
at the same time we cannot neglect the effect of competing and
overlapping sound changes pointed out by Wang (1969) as an expla-
nation for such irregularities; and even without such competition,
there is evidence for lexical irregularity in these on-going changes
which is inconsistent with the neo-grammarian hypothesis and yet
free of any social affect.

8. Doubts on the Level of Abstraction

Affected by Social Factors

We can now return to the second of the three general questions

raised:can social factors affect abstract rules of grammar and pho-
nology? First of all, it is clear that members of the speech community
do not develop social reactions to abstract rules of phonology such
as the tensing rule for ae. The highly marked social feature is

the second rule — the raising of the vowels that have been tensed.
For the first rule, we find considerable individual variation in New
York City in the treatment of words ending in voiced fricatives (razz,
jazz); or of weak words ending in nasals (am, can, and and variable
as against has, as, had which are always lax). There is a great deal
of variation in the treatment ofwords like passage, or Ahbie, which
may be analyzed with a derivational boundary /. aeC + V ./ as . .

opposed to the consistent tensing of words like passing or stabbing

with inflectional boundaries /. aeC#V ./ or consistent lax treat-
. .

ment of castle or cabbage with no such boundary at all: /. aeCV ./. . .

This variation is idiosyncratic; it fails to show any regular social

distribution, no matter how carefully we study the data (Cohen 1970).
Husbands and wives, brothers and sisters may differ on these points,
but unpredictably, and neither will notice the difference. 29 In general,
we find that social affect is not attached to such variations in abstract
rules, but rather to individual words (cf. Whitney 1901) or to low-

29. In a recent interview in New York husband read Abbie with a tense,
City, a
raised vowel, while his wife used a lax, unraised vowel in the same word. They were
both from the same dialect background and showed no other systematic differences
in their vowel systems.
322 SOCIOLINGUISTIC PATTERNS

level rules of phonetic behavior that involve very frequent items.

Certainly there are important grammatical and phonological con-
sequences of sociolinguistic variables. Consonant cluster simplifica-
tion is a typical social variable, showing both stylistic and social
stratification. One of the constraints on the rule is the existence of
a morpheme boundary before the final consonant, signalling that it
is a separate morpheme. If there is no such boundary, the rule will

apply more often:

[-cont]-> <0> / [ + cons] <0> ## <-syl>

The variable constraint indicated by <0> in the environment registers
the fact that without a morpheme boundary, the cluster in last is
simplified more often than the one in pass#d, the cluster in old more
than the one in rolled. The rule in this form may have no immediate
effect upon grammar, for it indicates that the past tense morpheme
#d indeed preserved and recognized. But when such sociolin-
is

guistic rules are extended and become obligatory, we find com-

pensating and dramatic changes in the grammar. Thus when final
clusters -pt and -kt were obligatorily simplified in Scots, the rule
of epenthesis apparently changed to yield slippit and workit instead
of slip' and work', allowing the regular past to be expressed. (For
more detailed discussion and other examples, see Ch. 8.)
We can look back at the phonetic processes in late Old English
which led to the loss of final inflections: reduction of unstressed
vowels, deletion of final nasals, and loss of final schwa. All of these
appear to have been optional sociolinguistic variables for some time
in the history of English. Before the inflections were entirely lost,
we can hazard the guess that some of the compensating changes in
word order had already occurred. But this is still a speculative area,
for we have not yet been able to study any such systematic gram-
matical change in progress.
On the whole, our evidence then points to a negative answer for
our second question, but suggests that the connection between social
variation and the higher-level change may be fairly rapid as the
variable rule develops into a categorical one. It is important to note
that social significance depends on variability. In that sense, social
meaning is parasitic upon language: it is confined to those areas of
variation, usually on the leading edge of a generalizing linguistic
change, where there exist alternative ways of saying "the same
thing."
We can observe the results of a process of syntactic generalization
The Social Setting of Linguistic Change 323

in the black English treatment of negativeconcord (Labov 1972a, Ch.

4). This is an expressive optional rule for white dialects, giving a

choice between emphatic Nobody gave him nothing and unmarked

Nobody gave him anything. Some white dialects optionally redupli-
cate the negative particle after an indeterminate subject but only —
within the same clause; thus Nobody didn't give him nothing is a
way of saying the same thing as the two preceding sentences. Black
speakers have the additional option of extending this subrule across
clause boundaries, as in It ain't nobody didn't give him nothing,
meaning the same as Nobody gave him anything. This extension of
the expressive negative concord rule coincides with the extension
of normal negative concord to all indeterminates within the clause,
now obligatory in most BEV; the new subrule adds the expressive
feature which the old one has lost.
These last two examples are discussed after the fact, and must
be taken as speculative reconstructions of an historical process
which we have not yet observed in progress. Only when we have
the good fortune to seize such a syntactic change as it is occurring
will we be able to give a good solution to the transition problem,
and provide a sound basis for other arguments about the evaluation
and actuation of change.

9. Is There An Adaptive Function to

Linguistic Diversity?

It is plain to most linguists that the "destroy and rebuild" theory

of linguistic evolution is equivalent to claiming that the whole
process is dysfunctional. For the systematic part is the destructive
one, and the analogical reshaping seems to be making the best of
a bad job. And if the principle of least effort is the evil genius behind

the destruction, we can only look at language change as a kind of

massive testimony to original sin.
Gauchat made it had been
clear that the principle of least effort
overpromoted. He pointed out (1905) that the diphthongization of
a showed every sign of being an increase in effort, not a decrease.
We can say the same for the tensing and raising rules which have
played such a large part in this discussion. The stigmatized phones
are longer, more peripheral, with greater intensity than the older
A
forms, and develop an offglide as well: contrasting [bae- d] with
9
be:* d]. These developments give one a sense of the strong motivating
force behind sound change, one that is almost the opposite of the
economy or least effort argument. We observe sound change length-
324 SOCIOLINGUISTIC PATTERNS

ening vowels, rotating systems, merging word classes, destroying

distinctions, overriding structural constraints. Why?
A classic confrontation on the issue of progress in the evolution
of language may be seen in the papers of Greenberg (1959) and
Hymes (1961). Greenberg takes the position that there is evolution
only in the sense of diversification, but no progress in the sense of
increase of complexity or adaptive radiation. Hymes points to the
development of complex layers of vocabulary, scientific vocabulary,
metalanguages, and other attributes of world languages as evidence
in favor of continuing development. Since this chapter has been
concerned with the central core of grammar and phonology, rather
—
than lexicon, it would seem to favor Greenberg's position that there
is only diversification.

But what if this diversification were itself an important element

in cultural evolution? Inevitably, we are drawn back to the third
question: does diversification have an adaptive function, or is it the
product of multiple local failures in the communication network?
My own studies of on-going linguistic changes indicate that dialect
diversification is continuing in the face of saturation by the mass
media, and in spite of close contact among the social groups in-
volved. The fact that diversity is not automatically connected with
isolation suggests that it may also be connected with the normal
processes of face-to-face communication.
Most linguists who work with small, diverse groups must recognize
in themselves a natural prejudice in favor of the survival of their
subjects. The anthropological or comparative linguist will intuitively
fight for the existence of his group, and he resists the notion that
the cost of bilingualism is too great to be borne. He refuses to weigh
the value of a language or a dialect in terms of its attractiveness to
printers, the size of its literary output, or how well it prepares
children to fit into a European school system. Linguists must recog-
nize that they are interested parties in this argument.
With this precaution, I am inclined to believe that the development
of linguistic differences has positive value in human cultural evolu-
tion — and that cultural pluralism may even be a necessary element
in the human extension of biological evolution.
If we are to risk speculation about the general character of human
evolution, it have a point of comparison with other
will be helpful to
species that have developed communication systems. My thinking
on this point has been stimulated by a recent overview of work on
the development of bird song by Nottebohm (1970). Much of Not-
The Social Setting of Linguistic Change 325

tebohm's work has involved the vocal learning of chaffinches; they

are one of several species that differ from most birds in that their
songs are not stereotyped and genetically controlled in any simple
sense. Chaffinches pass through a critical period of ten months during
which they learn their songs from other birds of their species. If they
are deafened early in this period, they produce degenerate and
extremely abnormal songs; if deafened after ten months, they con-
tinue to produce normal song. It is of immediate interest to linguists
that birds like the chaffinch and the white-crowned sparrow have
developed dialects. Nottebohm suggests that the shift from "self-
centered" to "environmentally dependent" song production may
have accompanied the rapid expansion of the birds into differen-
tiated habitats. The examples he gives of the behavior of the white-
crowned sparrows suggest that dialects may play a role in the mating
system by providing a relative degree of genetic isolation but without
any "necessary irrevocable commitment to actual speciation."

Whereas genetic isolation of small populations may lead to high rates of

extinction and even possibly to excessive inbreeding, differences in vo-
calizations are probably rarely insuperable barriers to breeding, and thus
the microevolutionary process is kept more flexible and open. (1970:955)

Nottebohm himself raises the possibility that human dialects may

have an evolutionary function, and may have influenced the emer-
gence of "local physiological adaptations." It is also possible that
the "linguistic puberty" through which we pass, seriously reducing
our language-learning ability, may itself be the product of evolution-
ary selection. A language-learning "block" in human development
would have the same general function as in birds, encouraging
diversity of dialect patterns.
In speculating about these possibilities, I would prefer to look
forward rather than backward, and focus on the connection with
cultural diversity rather than physiological adaptation. The value of
Nottebohm's contribution is not in setting forth a theory or an
hypothesis that we can test immediately, but rather suggesting an
alternate view to broaden our thinking about linguistic evolution.
It is an interesting point to consider that language diversity may have

value for humans other than linguists, providing relative cultural

isolation and maintaining cultural pluralism. And linguists them-
selves may be encouraged to look more deeply into the mechanisms
which differentiate languages as well as the limiting conditions
which form the content of a universal grammar.
 BIBLIOGRAPHY

Allen, P. 1968. /r/ variable in the speech of New Yorkers in department

stores. Unpublished research paper SUNY, Stony Brook.

Anshen, F. 1969. Speech variation among Negroes in a small Southern

community. Unpublished dissertation, New York University.

Avis, W. 1961. The New England short V: a recessive phoneme. Language

37:544-58.

Babbitt, E. H. 1896. The English of the lower classes in New York City and
vicinity. Dialect Notes 1:457-64.

Bailey, C.-J. N. 1969a. The integration of linguistic theory: internal recon-

struction and the comparative method in descriptive linguistics, with
an appendix of 107 pan-dialectal ordered rules. Paper given before
Conference on Historical Linguistics in the Light of Generative The-
ory, Los Angeles. Printed in part in R. Stockwell and R. Macaulay
(eds.), Historical Linguistics and Generative Theory. Bloomington,
Ind.: Indiana University Press, 1972.

.
1969b. Introduction to Southern states phonetics. University of
Hawaii Working Papers in Linguistics 4-5.

Baird, S. J.
1969. Employment interview speech: a social dialect study in
Austin, Texas. Unpublished dissertation, University of Texas.

Baratz, J. C. 1969. A bi-dialectal task for determining language proficiency

in economically disadvantaged Negro children. Child Development
40.

Barber, Bernard. 1957. Social stratification. New York: Harcourt, Brace.

Bernstein, B. 1964. Elaborated and restricted codes. In Gumperz and Hymes

1964.

Bickerton, D. 1971. On the nature of the Creole continuum. Mimeo.

Bloch, B. 1948. A set of postulates for phonemic analysis. Language 24:3-46.

Bloomfield, L. 1933. Language. New York: Henry Holt.

. 1944. Secondary and tertiary responses to language. Language

20:45-55.

327
328 BIBLIOGRAPHY

Bright, W., ed. 1966. Sociolinguistics. The Hague: Mouton.

, and A. K. Ramanujan. and language

1964. Socio-linguistic variation
change. Jn H. G. Lunt, Proceedings of the Ninth International
ed.,

Congress of Linguists. The Hague: Mouton.

Bronstein, A. 1962. Let's take another look at New York City speech. Ameri-
can Speech 37:13-26.

Brown, L. 1969. The social psychology of variations in French Canadian

speech. Unpublished dissertation, McGill University.

Brown, R., and U. Bellugi. 1964. Three processes in the child's acquisition
of syntax. Harvard Educational Review 34:133-51.

Carden, G. 1970. On post-determiner quantifiers. Linguistic Inquiry

l(4):415-28.

Cedergren, H. 1970. Patterns of free variation: the language variable. Mimeo.

, and D. Sankoff. 1972. Variable rules: performance as a statistical

reflection of competence. To appear in Language.

Chen, M. 1971. The time dimension: contribution toward a theory of sound

change. Berkeley, Cal.: University of California Project on Linguistic
Analysis Reports, Vol. 2, No. 14.

,and Hsin-I Hsieh. 1971. The time variable in phonological change.

JournaJ of Linguistics 7:1-13.

Cheng, C.-C, and W. S.-Y. Wang. 1970. Phonological change of middle

Chinese initials. Berkeley, Cal.: University of California Project on
Linguistic Analysis Reports, Vol. 2, No. 10.

Chomsky, N. 1957. Syntactic structures. The Hague: Mouton.

. 1964. Current issues in linguistic theory. Jn Fodor and Katz 1964.

. 1965. Aspects of the theory of syntax. Cambridge, Mass.: MIT Press.

1966. Topics in the theory of generative grammar. Jn T Sebeok, ed.,

.

Current trends in Jinguistics 3: Jinguistic theory. Bloomington, Ind.:

Indiana University Press.

and M.
, Halle. 1968. The sound pattern of English. New York: Harper
& Row.

Cohen, P. 1970. The tensing and raising of short a in the metropolitan area
of New York City. Unpublished Master's essay. Columbia University.

Cook, S. 1969. Language change and the emergence of an urban dialect in

Utah. Unpublished dissertation, University of Utah.
Bibliography 329

Eckert, P. 1969. Grammatical constraints in phonological change: the un-

stressed vowels of Southern France. Unpublished Master's essay,
Columbia University.

Elliot, D., S. Legum, and S. Thompson. 1969. Syntactic variation as linguistic

data. In R. Binnick et Papers from The Fifth Chicago
al. (eds.),
Regional Meeting. Chicago: University of Chicago Department of
Linguistics.

Enderlin, D. 1913. Die Mundart von Kesswil im Oberthurgau. In A. Bachman,

ed., Beitrage zur Schweizer-Deutschen Grammatik. Frauenfeld:
Huber and Co.
Ervin-Tripp, S. 1967. SocioJinguistics. Working paper No. 3, Language Be-
havior Research Laboratory, Berkeley, Cal.

Ferguson, C. A., and J. J. Gumperz. 1960. Linguistic diversity in South Asia.

Publication of the Research Center in Anthropology, Folklore, and
Linguistics, No. 13. Bloomington: Indiana University Press.

Fischer, J.
L. 1958. Social influences on the choice of a linguistic variant.
Word 14:47-56.

Fishman, J., ed. 1968. Readings in the sociology of language. The Hague:
Mouton.

. 1969. Sociolinguistics. In K. W. Back, ed., Social psychology. New

York: Wiley.

, R. L. Cooper, R. Ma, et al. 1968. Bilingualism in the barrio. Final

report on OECD-1-7-062817. Washington, D.C.: Office of Education.

Fodor, J., and J. Katz, eds. 1964. The structure of language. Englewood Cliffs,

N.J.: Prentice-Hall.

Frank, Y. A. 1948. The speech of New York City. Unpublished dissertation,

University of Michigan, Ann Arbor.

Garfinckel, H. 1967. Studies in ethnomethodology. Englewood Cliffs, N.J.:

Prentice-Hall.

Garvin, P., and M. Mathiot. 1960. The urbanization of the Guarani language.
In A. F. C. Wallace, ed., Men and cultures. Philadelphia: University
of Pennsylvania Press.

Gauchat, L. 1905. L'Unite phonetique dans le patois d'une commune, In Aus

Romanischen Sprachen und Literaturen: Festschrift Heinrich Mort,
pp. 175-232. Halle: Max Niemeyer.

,
J. Jeanjaquet,
and E. Tappolet. 1925. Tableaux phonetiques des
patois suisses romands. Neuchatel: Paul Attinger.
330 BIBLIOGRAPHY

Grant, W., and J.

Dixon. 1921. Manual of modern Scots. London; Cambridge
University Press.

Greenberg, J.
H. 1959. Language and evolution. In Evolution and anthro-
pology: a centennial appraisal. Washington, D.C.: The Anthropo-
logical Society of Washington.

, ed. 1963. Universals of language. Cambridge, Mass: MIT Press.

Grimshaw, A. D. 1968. Sociolinguistics. In W. Schramm et al., eds., Hand-

book of communication. New York: Rand McNally.

Grinder, I., and P. Postal. 1971. Missing antecedents. Linguistic Inquiry

2:209-312.

Gumperz, J. J.
1964. Linguistic and social interaction in two communities.
In Gumperz and Hymes 1964.

. 1967. On the linguistic markers of bilingual communication. In

Macnamara 1967.

. 1971. Language in social groups. Stanford: Stanford University

Press.

, and D. Hymes. 1964. The ethnography of communication. Special

issue of American Anthropologist 66(6.2).

.
}. Wilson, and R. Wilson. 1971. Convergence and creolization: a
case from the Indo-Aryan/Dravidian border in India. In Hymes
1971. Also in Gumperz 1971.

Hafner, E. M., and S. Presswood. 1965. Strong inference and weak inter-
actions. Science 149:503-9.

Halle, M. 1962. Phonology in generative grammar. Word 18:67-72.

Harris, Z. 1951. Structural linguistics. Chicago: University of Chicago Press.

Hermann, M. E. 1929. Lautveranderungen in der Individualsprache einer

Mundart. Nachrichten der Gesellschaft der Wissenschaften zu Got-
tingen, Philosophisch-historische Klasse 11:195-214.

Herzog, M. I. 1965. The Yiddish language in Northern Poland: its geography

and history. Publication of the Research Center in Anthropology,
Folklore, and Linguistics, No. 37. Bloomington: Indiana University
Press.

Hockett, C. F. 1950. Age-grading and linguistic continuity. Language

26:449-57.

. 1958. A course in modern linguistics. New York: Macmillan.

Hoenigswald, H. 1963. Are there universals in linguistic change? In Green-

berg 1963.
Bibliography 331

Homans, G. C. 1955. The human group. New York: Harcourt, Brace.

Hubbell, A. F. 1950. The pronunciation of English in New York City. New

York: Columbia University Press.

Hymes, D. 1961. Functions of speech: an evolutionary approach. In F. C.

Gruber, ed., Anthropology and education. Philadelphia: University
of Pennsylvania Press.

. 1962. The ethnography

of speaking. In T. Gladwin and W. C. Sturte-
vant, eds.,Anthropology and human behavior. Washington, D.C.
Anthropological Society of Washington.

. 1964. Introduction. In Gumperz and Hymes 1964.

, ed. 1971. Pidginization and creolization of languages. London: Cam-

bridge University Press.

Jespersen, O. 1927. A modern English grammar on historical principles, I.

London: George Allen.

. 1946. Mankind, nation, and individual from a linguistic point of

view. Bloomington: Indiana University Press.

Joos, M. 1959. The isolation of styles. Georgetown University Monograph

Series on Languages and Linguistics 12:107-13.

Kazazis, K. 1969. The relative importance of parents and peers in first-

language learning. Paper given before the Linguistic Society of
America, San Francisco, December.

Kihlbom, A. 1926. A contribution to the study of fifteenth-century English.

Uppsala: Lundequistska.

Kiparsky, P. 1968. How abstract is phonology? Mimeo. Bloomington, Ind.:

Indiana University Linguistics Club.

Klima, E. S. 1964. Negation in English. In Fodor and Katz 1964.

Kokeritz, H. 1953. Shakespeare's pronunciation. New Haven: Yale Uni-

versity Press.

Kucera, H. 1961. The phonology of Czech. The Hague: Mouton.

Kurath, H. 1939. Handbook of the linguistic geography of New England.

Providence, R. I.: American Council of Learned Societies.

. 1949. A word geography of the eastern United States. Ann Arbor:

University of Michigan Press.

, et al. 1941. Linguistic atlas of New England. Providence, R.I.:

American Council of Learned Societies.
332 BIBLIOGRAPHY

-, and R. McDavid. 1951. The pronunciation of English in the Atlantic

states. Ann Arbor: University of Michigan Press.

Kurylowicz, J.
1964. On the methods of internal reconstruction. In H. G.
Lunt, ed., Proceedings of the Ninth International Congress of Lin-
guistics. The Hague: Mouton.

Labov, W. 1964a. Phonological indices to social stratification. In Gumperz

and Hymes 1964.

1964b. The linguistic variable as a structural unit. Paper given before

.

the Linguistics Club of Washington, D.C., October. Washington Lin-

guistics Review 3:4-22 (1966).

1966a. The social stratification of English in

.
New York City. Wash-
ington, D.C.: Center for Applied Linguistics.

. 1966b. On the grammatically of everyday speech. Paper given

before the Linguistic Society of America, New York City.

. 1969. Contraction, deletion, and inherent variability of the English

copula. Language 45:715-62.

.1972a. Language in the inner city: studies in the black Engiish

vernacular. Philadelphia: University of Pennsylvania Press.

. 1972b. The recent history of some dialect markers on the island of

Martha's Vineyard. In L. Davis, ed., Studies presented to Raven
McDavid. University, Ala.: University of Alabama Press.

. The internal evolution of linguistic rules. In R. Stockwell and

1972c.
R. Macaulay, eds., Historical linguistics and generative theory.
Bloomington: Indiana University Press.

, P. Cohen, C. Robins, and J.

Lewis. 1968. A study of the non-standard
English of Negro and Puerto Rican speakers in New York City. Final
report, Cooperative Research Project 3288. 2 vols. Philadelphia, Pa.:
U.S. Regional Survey, 204 N. 35th St., Philadelphia 19104.

, and P. Pedraza. 1971. A study of the Puerto Rican speech community

in New York City. Report to the Urban Center of Columbia Univer-
sity.

, and B. Wald. 1969. Some general principles of vowel shifting. Paper

given before the Linguistic Society of America, San Francisco, '-De-
cember.

, M. Yaeger, and R. Steiner. 1972. A quantitative study of sound

change in progress. Final report on National Science Foundation
contract NSF-GS-3287. Philadelphia, Pa.: U.S. Regional Survey,
204 N. 35th St., Philadelphia 19104.
Bibliography 333

Laffal, J.
1965. Pathological and normal language. New York: Atherton Press.

Lakoff, G. To appear. Linguistics and natural logic. In J. Mehler, ed M

Handbook of cognitive psychology.

Lambert, W. E. 1967. A social psychology of bilingualism. In Macnamara

1967.

. 1969. Some current psycholinguistic research: the Tu-Vous and Le-La

studies. In Puhvel, J.,
ed., Substance and structure of language.
Berkeley: University of California Press.

, M. Anisfeld, and G. Yeni-Komshian. 1965. Evaluational reactions

of Jewish and Arab adolescents to dialect and language variations.
Journal of Personality and Social Psychology 3:313-20.

, et al. 1960. Evaluation reactions to spoken languages. Journal of

Abnormal and Social Psychology 60:44-51.

Lawton, D. 1968. Social class, language, and education. London: Routledge

and Kegan Paul.

Legum, S., C. Pfaff, G. Tinnie, and M. Nicholas. 1971. The speech of young
black children in Los Angeies. Technical Report No. 33. Los Angeles:
Southwestern Regional Laboratory.

Lehmann, W. P. 1963. Historical linguistics: an introduction. New York: Holt,

Rinehart and Winston.

, and Y. Malkiel, eds. 1968. Directions for historical linguistics. Austin:

University of Texas Press.

Levine, L., and H. J. Crockett, Jr. 1966. Speech variation in a piedmont

community: postvocalic r. In Lieberson 1966.

Levi-Strauss, C. 1963. Structural anthropology. Tr. C. Jacobson and B.

Schoepf. New York: Basic Books.

Lieberson, S., ed. 1966. Explorations in sociolinguistics. Special issue of

Sociological Inquiry 36(2).

Loman, B. 1967. Conversations in a Negro American dialect. Washington,

D.C.: Center for Applied Linguistics.

Lyons, }. 1968. Introduction to theoretical linguistics. London: Cambridge

University Press.

Macnamara, J. 1967. Problems of bilingualism. Special issue of The Journal

of Social Issues 23(2).

Mahl, G. 1972. People talking when they can't hear their voices. In A.
Siegman and B. Pope, eds., Studies in dyadic communication. New
York: Pergamon Press.
334 BIBLIOGRAPHY

Malkiel, Y. 1967. Each word has a history of its own. GJossa 1:2.

Martinet, A. 1955. Economie des changements phonetiques. Berne: Francke.

. 1964a. Elements of general linguistics. Tr. S. Palmer. Chicago:

University of Chicago Press.

. 1964b. Structural variation in language. In H. G. Lunt, ed., Proceed-

ings of the Ninth International Congress of Linguistics. The Hague:
Mouton.

Meillet, A. 1921. Linguistique historique et linguistique generale. Paris: La

Societe linguistique de Paris.

Mitchell-Kernan, C. 1969. Language behavior in a black urban community.

Working paper No. 23, Language Behavior Research Laboratory,
Berkeley, Cal.

Moulton, W. G. 1962. Dialect geography and the concept of phonological

space. Word 18:23-32.

Norman, R. 1971. An ear to New York. Unpublished manuscript.

Nottebohm, F. 1970. Vocal imitation and individual recognition of finch calls.

Science 168:480-82.

Paul, H. 1889. Principles of the history of language. Tr. H. A. Strong. New

York: Macmillan.

Peterson, G. E., and H. L. Barney. 1952. Control methods used in a study

of the vowels. Journal of the Acoustical Society of America 24:175-84.

Pilch, H. 1955. The rise of the American English vowel pattern. Word
11:57-63.

Postal, P. 1964. Constituent structure: a study of contemporary models of

syntactic description. Internal journal of American Linguistics, Pub-
lication 30.

Postal, P. 1971. Cross-over phenomenon. In Specification and utilization

of a transformational grammar. Scientific report No. 3. Yorktown
Heights, N.Y.: IBM (published by Holt, Rinehart, and Winston, New
York).

Preston, M. S. 1963. Evaluational reactions to English, Canadian French and

European French voices. Unpublished Master's thesis, McGill Uni-
versity.

Quirk, R. 1966. Acceptability in language. Proceedings of the University of

Newcastle upon Tyne Philosophical Society 1:79-92.

Reich, P. 1969. On the finiteness of natural language. Language 45.

Bibliography 335

Reichstein, R. 1960. Study of social and geographic variation of linguistic

behavior. Word 16:55.

Sacks, H. 1972. An initial investigation of the usability of conversational

data for doing sociology. In Sudnow 1972.

Sankoff, G. 1972. A quantitative paradigm for the study of communicative

competence. Paper given at Conference on the Ethnography of
Speaking, Austin, Tex.

, and H. Cedergren. 1971. Some results of a sociolinguistic study of

Montreal French. In R. Darnell, ed., Linguistic diversity in Canadian
society. Champaign, 111.: Linguistic Research, Inc.

, R. Sarrasin, and H. Cedergren. 1971. Quelques considerations sur

la distribution sociolinguistique de QUE dans le frangais
la variable
de Montreal. Paper given at the 39th Congress, Association cana-
dienne-frangaise pour Favancement des sciences.

Saussure, F. de. 1962. Cours de Jinguistique generale. Paris: Payot.

Schegloff, E. 1968. Sequencing in conversational openings. American An-

thropologist 70:1075-95.

Shewmake, E. F. 1927. English pronunciation in Virginia. Davidson, N.C.

Shuy, R., W. Wolfram, and W. K. Riley. 1967. A study of social dialects in

Detroit. Final report, Project 6-1347. Washington, D.C.: Office of Edu-
cation.

Solomon, D. 1966. The system of predication in the speech of Trinidad.

Unpublished Master's essay, Columbia University.

Sommerfelt, Alf. 1930. Sur la propagation de changements phonetiques.

Norsk Tidsskrift for Sprogvidenskap 4:76-128.

Sturtevant, E. 1947. An introduction to linguistic science. New Haven: Yale

University Press.

Sudnow, D., ed. 1972. Studies in social interaction. New York: Macmillan.

Sweet, Henry. 1900. The history of language. London: J. M. Dent.

Tarde, Gabriel. 1873. Les lois d'imitation.

Torrey, J. 1969. The learning of grammatical patterns. Journal of Verbal

Learning and Verbal Behavior 8:360-68.

Trager, G. L. 1930. The pronunciation of 'short a' in American Standard

English. American Speech 5:396-400.

. 1934. What conditions limit variants of a phoneme? American

Speech 9:313-15.
336 BIBLIOGRAPHY

1940. One phonemic entity becomes two: the case

.
of 'short a'.

American Speech 15:255-58.

Troubetzkoy, N. 1957. Principes de phonologie. Tr. J. Cantineau. 2nd ed.
Paris: Klincksieck.

Trudgill, P.J.
1971. The social differentiation of English in Norwich. Un-
published dissertation, Edinburgh University.

Tucker, G. R., and W. E. Lambert. 1966. White and Negro listeners' reactions
to various American English dialects. Mimeo.

Vendryes, J. 1951. Language: a linguistic introduction to history. Tr. Paul

Raden. New York: Barnes & Noble.

Vygotsky, L. S. 1962. Thought and language. Tr. E. Hanfmann and G. Vakar.

Cambridge, Mass.: MIT Press.

Walker, John. 1791. Principles of English pronunciation.

Wang, W. S.-Y. 1969. Competing changes as a cause of residue. Language

45:9-25.

Watson, }. D. 1969. The double helix. New York: New American Library.

Webb, Eugene J.,

et al. Unobtrusive measures: non-reactive research in the
social sciences. Chicago: Rand McNally.

Weinberg, M., and M. Najt. 1968. Los pronombres de tratamiento en el

espanol de Bahia Blanca. In Adas de la Quinta Asembiea Interuni-
versitaria de Filologia y Literaturas Hispanicas.

Weinreich, U. 1954. Is a structural dialectology possible? Word 10:388-400.

.
1959. Review of Hockett, Modern Linguistics. Romance Philology
13:329-39.

, W. Labov, and M. Herzog. 1968. Empirical foundations for a theory

of language change. In Lehmann and Malkiel 1968.

Whitney, W. D. 1901. Language and the study of language. New York:

Scribner's.

Whorf, B. L. 1943. Phonemic analysis of the English of eastern Massachu-

setts. Studies in Linguistics 2:21-40.

Wolfram, W. 1969. Linguistic correlates of social stratification in the speech

of Detroit Negroes. Unpublished thesis, Hartford Seminary Founda-
tion.

Wyld, H. C. 1920. A history of modern colloquial English. Oxford: Basil

Blackwell.

Zwicky, A. 1970. Auxiliary reduction in English. Linguistic Inquiry 1:323-36.

 INDEX

Abraham G.: stylistic array for, 102 tioned, 279. See also Centralization;
Accountability principle, 72 Diphthongs, centralized; Martha's
Adolescence: crucial period for language Vineyard
development, 138-39. See also Child;
Peer group
beliy-gut, 7
Age: and pronunciation, 17, 57, 59-60,
Bengali, 296
64-65, 105, 106, 116; and centralization
Bennie N.: stylistic array for, 106
on Martha's Vineyard, 21, 24, 30,
Bilingualism: Spanish-English, 205-6; in
166-67; and linguistic change, 134-36,
study of languages, 214-215; French-
178, 291-92; and lower middle class
English, 310-11; mentioned, 324
speech patterns, 117; and acquisition of
Biological evolution, and language. See
community speech norms, 138-40; in
Evolution, biological
subjective reaction tests, 150-56; con-
Bird song: evolutionary analogy, 324-25
cept of apparent time, 163; and socio-
Black English vernacular: characterized,
linguistic variables, 240
189-90; 223, 299; repetition tests, 212;
ah: and (oh) in New York City, 175
consonant cluster simplification, 216-
American English: Martha's Vineyard a
26;sources used in study, 216; dele-
relic area, 6-7; centralized diphthongs
tion of copula, 226-30; negative con-
in, 10-11; (r)-pronunciation, 141, 145,
cord, 234-37; 323
287n, 290-92; o-pronunciation, 292;
Blacks: and Indians on Martha's Vine-
changes western US, 309; speech
in
yard, 35; and New York City speech
stereotypes, 314-15. See also Black
patterns, 54, 118-19; 157, 318-19; and
English vernacular; Broadcasting me-
process of sound change, 172n
dia; Martha's Vineyard; New York
Bloomfield, L.: on linguistic change, xiv,
City; Southern speech patterns
22, 262, 266
American Language Survey, 80n
Borrowing, 1, 23, 308
Analogy: in language, 1, 274
Boston: speech stereotypes, 248
Anecdote, 268, 270
Breathing: as channel cue, 95
anymore, 309
Broadcasting media: pronunciation of (r),
Arrays: for stylistic variation, 99-107
138, 290; effect on American English,
Articulation, 23, 40, 164, 165
180n, 317; as source of linguistic data,
Audio-monitoring, 138, 179, 208, 213, 246
211
Austin, Texas, 68n, 238n
(aw): history of, 1, 10-11; Martha's Vine-
yard form, 9; parallelism with (ay), Canadian English, lOn
24-25; example of change from below, Canadian French, 206, 232, 241, 249, 310,
180; in Salt Lake City, 281, 309; in New 311
York City, 287-90; mentioned, 279. See Cape Cod, 10
also Centralization; Diphthongs, cen- Cape Hatteras, 248
Martha's Vineyard
tralized; Celtic, 265
(ay): of, 1, 10-11; Martha's Vine-
history Centralization: of diphthongs on Mar-
yard form, 9; parallelism with (aw), tha's Vineyard, 9-26, 39-42, 165-71. See
24-25; in New York City, 287-90; men- also Diphthongs, centralized

337
338 INDEX

Chain shift, 2, 280 Code-switching, 188-89

Change, linguistic: study of, xiii-xiv, 2-3, common-sense, 89, 118-20
161-62; data for, xiv, 3, 66; origins of, Communication: and language, 224, 271
1-2, 265, 277;neo-grammarian views, Competence, 186, 191, 267. See also
22-23; definition, 23; models for, 39, 123, Chomsky, N.
266n, 293-95, 304; role of lower middle Complementizers, 247
class, 62, 115, 134-36; and social varia- Consonant: following, 20
tion, 122, 271-72, 321-23; time concepts, Consonant cluster simplification, 190,
134-35, 275-76; direction of, 145, 286-91, 216-26, 322
295; major problems of, 160-62, 283; Constraints, variable: on linguistic
constraints upon, 220, 283, 300; effect change, 162n, 276; notation for, 218, 220;
on linguistic structure, 223-24, 231, 246; and grammatical environment, 228; in-
complexity of, 233; 19th-century views dependent, 231-33; in discourse analy-
of, 261; social context for, 3, 251-52, sis, 257; mentioned, 220, 322. See also

260; uniformitarian principle, 275; and Rules, linguistic

sex differentiation, 301-4; social evalu- Contact languages, 300
ation of, 308-17. See also Language; Contraction: in English, 227-29
Linguistics; Variable, linguistic Copula: deletion of, 226-30
Change, social, 121, 265. See also Varia- Courtroom proceedings: as source of
tion, social data, 67
Channel cues, 88, 94-97, 100, 103 Creoles, 224, 265n, 300
Charmey, Switzerland: sound change in, Czech, 205
163, 276-77, 278; role of women in lin-
guistic change, 301; mentioned, 23, 308 d: deletion of, 216-26
Chicago: sex differentiation in speech, Daily News (New York City), 46
301-2 Darwin, Charles: on language, 273
Child: rhymes and customs of, 91-92; Deletion: in English, 227-29
language development, 138-39, 263, 303, Department stores, New York City: social
304-7; monologues, 183; and "inborn stratification, 45, 47-48; advertising
theory of language," 188n; data on policies, 46-47; reasons for survey of,
speech of, 214n, 221n, 223, 268 48-49; (r)-pronunciation patterns,
Chilmark, Martha's Vineyard, 4, 14, 51-52, 60-61, 136; on Long Island, 68.
28-29, 35, 37 See aiso New York City dialect;
Chinese dialects, 321 (r)-pronunciation
Chomsky, N.: competence/performance Detroit: linguistic studies in, 206: Black
distinction, 186; on generative seman- English vernacular in, 221-22, 229-30;
tics, 198; on linguistic change, 262, 265, sex differentiation in language, 243,
266 301-2
Class, socioeconomic: stratification in (dh): in New York City, 77-78, 99-107,
New York City, 46, 150, 242; and lin- 113-14; as clue to casual speech pat-
guistic change, 63-64, 117, 134-36, 172, terns. 96; style shifting, 98; subjective
285-97; linguistic insecurity, 64-65, 133; reaction tests for, 154-55
and linguistic variables, 96n, 112-13, Dialectology, 268, 304
121, 125-32, 176, 240; lower middle class Dialects: superposed, 98; mixture of, 188,
speech habits, 100, 103, 244-45; Mobili- 223, 225; and idiolect, 192-99; syntactic,
zation for Youth index of, 112-13; 192-93; quantifier, 193-97; subordinate
(r)-pronunciation, 115, 124, 154; and and superordinate, 213-14; standard
speech norms, 63-65, 139, 249-50, 310; and nonstandard, 215, 218; borrowing,
social psychologists' view of, 258. See 286; regional, 286, 299-300; restructur-
also Hypercorrection; Index of Lin- ing of, 305-7; diversification of, 324
guistic Insecurity; Prestige; Stratifica- Diphthongs, centralized: on Martha's
tion, social Vineyard, 9-10, 18-20, 25-26, 36-38;
Climate: and language, 270n history in American English, 10-11; in-
Cockney, 285-86, 300, 316 fluences on, 20, 40
Index 339

Discourse: analysis of, 184n, 252-58; and Genesee Valley, NY, 10

language, 258 Geography: and speech patterns on
Diversification, linguistic: adaptive func- Martha's Vineyard, 25-26
tion of, 273-74, 323-25 Grammar: generative, 187, 200-201, 235,
Doris H.: stylistic array for, 102-3 252; grammatically, 188, 203; data for,
Durkheim, E.: influence on Meillet, 268 190-91, 200, 204; and sound changes,
224-25; comprehension of, 233n; and
e: merger with i, 118 speech community, 247. See also Lan-
Economic factors: on Martha's Vineyard, guage; Rules, linguistic; Structure,
27-28. See also Income linguistic
Edgartown, Martha's Vineyard, 2, 14, 28 Great Vowel Shift, 10, 143
Education, 25, 48, 112, 115 Greek dialects, 307
(eh): in New York City, 73-76, 99-107, Gumperz, J.,
205
126-27, 180; index for, 75; subjective
reaction tests, 131; changing variable,
Halle, M.: on linguistic change, 2-3, 262,
134
266. See also Chomsky, N.
Embedding, social, 286
Hermann, M. E., 23
English-descent group: on Martha's
Hillsboro, North Carolina: sex differenti-
Vineyard, 6, 30-32, 36-37. See also
ation, 243; (r)-pronunciation, 244, 291;
Yankee
mentioned, 205, 281
English language: change in, 265; histori-
Hindi, 296
cal studies of, 282, 285-87, 292; dialects,
Hockett, C. F., xiv, 22
299, 300, 316. See also American Eng-
Hymes, Dell, 95n, 184
lish
Hypercorrection: on Martha's Vineyard,
Environment, linguistic, 18, 21, 169-70
31; role in linguistic change, 39, 123,
Ethnic groups: on Martha's Vineyard, 6,
141, 178, 179-80; defined, 125; in conso-
26; in New York City, 118, 134, 153,
nant cluster simplification, 217; in verb
172-74, 177, 312n, 317, 318; and socio-
forms, 219, 221, 223; of lower middle
linguistic variables, 240; and linguistic
class, 244-45;of (r), 291; among
change, 296-98. See also Blacks;
2nd-generation speakers, 298; men-
English-descent group; Indians; Ital-
tioned, 64, 177, 208
ians; Jews; Portuguese; Puerto Ricans
"Ethnography of speaking," 184
Evaluation: and social stratification, 116, i: merger with e, 118
130; of speaker's own speech patterns, Idiolect: and dialect, 192-99; mentioned,
132, 213; and linguistic change, 162, 170, xiii, 107, 165
176-77, 283, 308-17. See also Index of Imitation: and linguistic change, 23
Linguistic Insecurity; Subjective Reac- Income, 25, 27, 47-48, 112, 115. See also
tion tests Occupation
Evolution, biological: and linguistic evo- Index of Linguistic Insecurity, 52n, 64,
lution, 161n, 273-74, 323-325 117-18, 133, 216, 245
Evolution, linguistic, 160-61 Index: phonological, 111, 123; socioeco-
Extraposition, 247 nomic, 112, 115; linguistic, 120
Indians, Gay Head, 6, 34-35, 36, 38
Family background tests, 212 Individual: as linguistic informant,
Fatigue, 82, 106n, 154-55 185-86; role in linguistic change, 277
Fischer, }. L., 205. See also ing Informant, linguistic: selection of, 57, 67,
Formant patterns, 15-17 207; and vernacular forms, 213-14;
French: cluster simplification in, 224-25; bidialectal, 215; bilingual, 215
Parisian, 278, 287; history of, 301. See ing: in New England, 238; stable socio-
also Canadian French linguistic marker, 238-40, 244
Insecurity, linguistic, 117, 133. See also
Gaelic, 315 Class, socioeconomic; Index of Lin-
Gauchat, L., 23, 323, See also Charmey guistic Insecurity
340 INDEX

Interview: constraints of, 8, 43, 109; on vidualist approach, 261-63; introspec-

Martha's Vineyard, 12-13, 40-41; casual tive method, 267; "thought-experi-
and anonymous, 49-50, 65-69; samples ments," 268-70. See aiso Change,
for, 66, 79; speech styles in context of, linguistic; Language; iangue; Neo-
79, 86-94, 95-96, 97; readings, 80-83; grammarians; individual linguists
"danger of death question," 92-94, 209; Long Island: and New York City speech
white noise and visual monitoring, patterns, 68
97-98; group, 109, 210; individual, 109, Lower East Side: sampling and interview
209; therapeutic, 253, 254-57. See aiso methods, 44, 61, 97, 110, 111, 304; com-
Observation; Observer's Paradox pared with department-store survey,
Intuition: study of, xiii, 186, 191-92; un- 60-61; (r)-pronunciation, 63, 150; acqui-
reliability of, 198-99; in discourse anal- sition of sociolinguistic norms, 139-40;
ysis, 253. See also Subjective Reaction ing in, 238-39. See aiso New York City
tests dialect
Italians: and New York City dialect, 130,
172-73, 174, 177, 297-98, 311-12 Machyn, Henry, 285, 286-87
Macy's: social status, 46-48; employees'
Jamaican dialect, 224. 227 speech norms, 51-52; occupation, and
Jespersen, O., 2, 263 pronunciation, 56. See aiso Department
Jews: in New York City dialect, 172-73, stores
177, 180, 298, 311-12, 317 Mahl, G., 97-98
Josephine P.: stylistic array, 100-101 Marathi, 297n
Martha's Vineyard: economic and social
Kannada, 296-97 background, 4-6, 27, 30-32, 34; study of,
S. Klein, 46-48. See aiso Department 4, 40-41, 163, 279; archaic speech traits,

stores 6-7; summer visitors, 6, 12, 28; in-

Kokeritz, H., 2 terview sample, 13-14. 41; local atti-
Kurylowicz, A., 264, 266 tudes, 30-32, 38-39, 298-99; subjective
reactions, 146, 308-9; mentioned, 275,
Lahu dialect, 215n 303.See aiso (aw); (ay); Centralization;
LANE. See Linguistic Atlas of New Eng- English-descent group; Indians, Gay
land Head; Portuguese
Language: and society, 11, 123, 183, Martinet, E.: on linguistic change, xiv, 2.
248-49; in urban areas, 21, 143, 298-300; 262, 265, 266; on phonological structure,
individual's development of, 138-39; 181; on linguistics, 185, 264; on social
evolution of, 123, 273, 324; study of, variation, 271
161, 184, 243, 259, 268-69; data for, 86, Meaning: language, 188n
in
191, 200-201, 205-6, 207-16; defined, on future of linguistics, 185n;
Meillet, A.:
See aiso Evolution, linguistic; Lin-
277. on linguistic change, 263, 266; on
Speech; Variables, linguistic
guistics; speech community, 318. 319
iangue: Saussure's concept of, 185-86. Minimal pairs, 83, 85, 99, 211
267; difficulties of, 191; and idiolect, Miriam L.: stylistic array, 106

mentioned, 226, 296

192; Mobilization for Youth survey, 46n, 61,
Laughter: as channel cue, 90. 95 71, 110, 111, 112, 123
Linguistic Atlas of New Engiand (LANE), Montreal, 206, 241. See aiso Canadian
4, 7, 10, 11, 12, 14, 24, 163, 166, 171 French
Linguistic Diversity in South Asia, 296 mother-wit, 119
Linguistics: historical, 2-3, 160-61, 260-61, Moulton, W. G., 181
268, 274; descriptive, 110-11; overlap
with sociology, 120-21; general, 184; Nantucket, 10
schools of, 185, 186, 260-70; methodol- Negative attraction, 193-94. 235
ogy, 201-2, 207-16; relevance of social Negative concord, 234-37, 323
context, 187, 264-70; and discourse Negative postposing, 235-36
analysis, 257; problems of, 258; indi- Negative raising, 247
Index 341

Neo-grammarians, 22-23, 163-64, 169, Past tense, 219, 225

274-75 Paul, Hermann: on linguistic change,
New England: earlyspeech charac- 261-62
teristics, 6-7; Portuguese in, 6; (aw) in, Peer group influence, 210, 304-7
11; (ay) in LANE, 10; ing in, 238; local Performance, 226. See also Competence
See also American Eng-
identity, 298. Philadelphia: sex differentiation. 304; re-
lish; Martha's Vineyard structuring of speech patterns, 305-7
New York Times, 46 Phonology: social pressure, 40; phono-
New York City dialect: linguistic varia- logical index, 111; phonological space,
bles, 11, 57, 64, 71, 72-78, 99-107, 181; recording of, 190; Troubetzkoy's
144-45, 242; social stratification, 44, views, 262; abstract, 272
112-16, 143, 294; changes in (^-pronun- Pidgins, 265n, 300
ciation, 58-59, 64-65, 138, 174; charac- Pilch, H., 2-3
teristics of, 70, 106, 124, 135; study of, Pitch: as channel cue, 95
71, 111-12, 118-19, 163;subjective reac- Portuguese: on Martha's Vineyard, 6,
tions to, 103, 117, 132-33, 138, 146-48, 32-33; centralizationamong, 33-34, 38;
311; vowel systems, 174-75, 180-81, 246, influence on Martha's Vineyard speech,
279, 287-88, 317; Black English vernac- 317-18
ular in, 221-22, 229-30; speech stereo- Postal, P., 192
types, 248, 315; influence of ethnic Probability: in variable rules, 230-32
groups, 297, 317, 318; sex differentia- Pronouns of respect, 284
tion, 301, 304. See also Department Pronunciation: and orthography, 286. See
stores; Lower East Side; Variables, lin- also Articulation
guistic Prague school of linguists, 262
Noise, white, 97-98 Psychologists: and linguistic change, 25;
Nominalization, 247 social, 258; and linguistics, 267-68
Norman French, 265 Prestige: and linguistic change, 3, 23, 179,
Norms, linguistic. See Speech norms 313-14; of New York City stores, 45-48;
Norwich, England, 206, 241-42, 243, 244, markers for, 59, 115, 141, 215, 290-92;
282, 294-95, 303 greater sensitivity of women to, 243,
302-3; definition of, 308; negative, 319.
o: New England short /o/, 7, 8; /o/ on See also Evaluation; Speech norms;
Martha's Vineyard, 8; pronunciation in Speech stereotypes; Subjective Reac-
US, 292 tion tests
Oak Bluffs, Martha's Vineyard, 41 Puerto Ricans, 118, 205. See also Spanish
Observation: casual, 13, 43, 86; systematic Pupil dilation, 245
and unsystematic, 210-11
Observer's Paradox, 61-62, 69, 209 Quantifiers, 193, 235
Occupation: on Martha's Vineyard, 25, Questionnaires, lexical, 12
26, 27-30; as index of social stratifica-
tion, 44-45, 48, 112; and linguistic vari- (r): in Martha's Vineyard, 6, 8-9, 29; social
ables in New York City, 45, 55-56, 115 stratification in New York City, 44,
(oh): in New York City dialect, 76-77, 63-64, 114-15, 117, 279; methods for
99-107, 127-28, 171-77; subjective reac- testing, 49-50; independent variables
tion tests for, 130-31; changing varia- affecting, 54-57; changing pronun-
ble, 134, 180 ciation in New York City, 58-59, 64,
Old English, 322. See also English lan- 134, 141; characteristics of in New York
guage City speech, 60-61, 72-73, 124, 312; and
(oy): and (oh), 175 casual speech, 96; stylistic arrays for,
99-107; and age level, 116, 135-36; sub-
Parent-child relationship: and language jective reaction tests for, 131, 144,
development, 268 149-58, 249; in US, 145, 244; in Hills-
paroie, 267. See also langue boro, 281; vocalization rule, 245. See
Passersby: systematic study of, 67n also Variables, linguistic
342 INDEX

Race, 54. See also Blacks; Ethnic groups Sound change: observation of, 21-22, 163,
Readings: as source of linguistic data, 13, 164; characterized, 23, 163, 164, 170;
80-83, 211-12; for New York City study of, 109, 162; embedding prob-
speech, 158-59 lems, 161, 169, 171-76; transition prob-
Recording: of Martha's Vineyard speech, lems, 161, 166, 171, 173-74; Bloomfield's
14-15, 17; of speech, 190, 204 view, 164; Hockett's view, 164; evalua-
Reichstein, R., 278 tion problem, 176-77; models for,
Rhymes, 91-92 178-80, 319-20; origins of, 178-80,
Rules, linguistic: stress, 20-21, 227-28; 323-24; and grammar, 224-25; and lan-
tensing, 73-74, 84; raising, 74, 76; social guage diversification, 272, 273-74
aspects, 192, 251, 272-73, 321-23; prob- Southern speech patterns: (ay) in, 10; reg-
lems of, 206-7, 246-47; variable, 207, ularity of, 18-19; in northern cities, 118,
225-26, 230, 237; for consonant cluster 299-300; ing in, 238
simplification, 217-18; vowel reduction, Spanish: Panamanian, 232, 241, 281-82,
227; Cedergren and Sankoff modifica- 293-94; Puerto Rican, 69, 205-6, 225, 241
schemas,
tions, 230-31; validity of rule Spectrograms, acoustic, 14-15, 75-76,
232-33; for negative attraction, 235; in- 166n, 202n
variable, 252; for sequencing, 254; in- Speech: and linguistics, xiii, 187; moni-
ternal evolution of, 276; restructuring toring of, 97-98; patterns, 124, 304-9;
of, 304-7 prestige forms, 139; subjective evalua-
tion, 144, 247-48; regional, 178; stereo-
types, 180, 248, 314-17, 320-21; difficul-
Saks Fifth Avenue: social status, 46-48; ties in study of, 187-91, 203-5;
(r)-pronunciation, by floor, 56-57. See vernacular, 208, 213-14. 215, 249; indi-
aiso Department stores cators, 237, 314, 320; social correction
Salespeople: "borrow" status, 45 of, 248, 293, 320; and language, 258;
Salt Lake City, 281, 309 markers, 314, 320. See aiso Change,
Saussure, F. de: on linguistic change, xiii, linguistic; Language; Linguistics
266-67; concept of langue, 185-86; Speech community: definition of, 120-21,
"Saussurian Paradox," 186, 267 158, 248n; and model for sound change,
Schuchardt, H., 265n, 266 178-79; and study of English grammar,
Scottish dialect, 224, 322 191, 247, 259; heterogeneous structure,
Segregation, social, 118-19 203; homogeneous model, 276, 295-96;
Semantic integration, 119-20 local identity, 298-99
Semantics, generative, 198 Speech event: study of, 49-50, 67, 69, 184,
Sex differentiation in language, 240, 243, 258n; social context, 284
244, 301-4 Speech norms: in New York City, 51,
Social change: language as indicator of, 62-63, 176-77; changes in, 63; in Austin,
111 Texas, 68n; considered invariant,
Social factors: on Martha's Vineyard, 110-11; define speech community,
24-26; effect on language, 264-65, 120-21; acquisition of, 139-40; covert,
272-73, 321-23; in sex differentiation of 177, 249-51, 313; in test situation, 213.
speech, 303-4. See aiso Change, lin- See aiso Class, socioeconomic; Evalua-
guistic; Social status; Stratification, so- tion; Speech stereotypes; Subjective
cial; Variable, linguistic; Variation, so- Reaction tests
cial Speech style: shifting, 21, 52, 97-99, 103,
Social Sciences, 201 203, 208, 213, 245, 249; variation in,

Social status: and language, 106-7, 70-71, 97; casual, 79, 80n, 85-97, 99, 108,
179-80, 284. See also Class, socioeco- 112; contexts for, 79-85; formal, 79-80,
nomic; Prestige; Stratification, social 108, 142, 190, 209, 211-12; reading,
Sociolinguistics,xiii, 72, 183-84, 296 80-83; spontaneous, 86, 91-92; stratifi-
Sociology: and linguistics, 110, 120-21, cation of, 179, 237, 241; and sociolin-
257, 267-68; of language, 183 guistic variable, 240; study of, 245-46.
Index 343

Speech style: (Cont.) Transcription, phonetic, 66, 202

See also Variable, linguistic; Variation, Trinidad dialect, 224
social Troubetzkoy, N., 262, 266
Standard English, 189 New York
uh: and (oh) in City, 174
Steve K.: stylistic array for, 103-4
Urban speech: diversity, 143; effect of
Stops: in American English, 57, 77-78,
rural migration, 298-300
124, 155, 315
Utah, 309. See also Salt Lake City
Stratification, social: defined, 44, 48,
129-30; in New York City, 47-48; and Value judgments, 13. See also Evaluation;
stylistic variation, 112; sharp, 113; fine, Subjective Reaction tests
114; detection of, 204; and sex, 243. See Variables, linguistic: selection of, 7-8, 41;
also Class, socioeconomic; Prestige; notation for, 11, 72; characterized, 71,
Social status 162; in sociolinguistic investigation, 72,
Strees. See Rules, linguistic 179; in New York City, 72-78; 99-107,
Structure, linguistic: and linguistic 118; style shifts, 96, 108, 208, 211; fre-
change, xv-xvi, 2, 223-24; analysis of, quency patterns, 108, 226n; and social
70-71, 184, 248; definitions, 107-8; vari- stratification, 115, 240; standard read-
ation in, 122, 188; and social factors, ing for, 158-59; and sound change, 171,
181-82, 251-52; "homogeneous" view, 178; persistence of stigmatized forms,
186, 203-4; and generative grammar, 249. See aiso specific variables
199-200; theoretical problems, 206. See Variables, social: on Martha's Vineyard,
aiso Language; Linguistics 24-25; in New York City, 54-57; socio-
Sturtevant, E., 3, 263, 266 linguistic, 237, 249, 322. See aiso Social
Subjective Reaction tests: in New York status; Stratification, social
City, 116-17, 130-32, 146-48, 150-56, Variation, linguistic: "free variation," xiv,
176, 311-12; "matched guise" tech- 1, 23, 188-89; in linguistic change, 1-2,

nique, 144, 145-46, 212-13, 248, 310; 271-72; on Martha's Vineyard, 9; in

control group, 156-57; for variable New York City, 70, 112, 144, 171; study
rules, 226n; reveal covert speech norms, of, 70-71, 97, 109, 122, 188; and struc-
249; mentioned, 82. See also Evaluation ture, 108; systematic nature, 111; arrays
Sweet, H., 262, 266 for, 99-107; and sociology, 120-21; syn-
Swiss French. See Charmey tactic, 193; inherent within speech
Swiss German dialects, 2n, 181, 268, 307 community, 204, 223; sociolinguistic
Syntax. See Grammar markers, 241; lexical, 246; and social
Systems, linguistic, 188 affect, 251; of individual, 321. See aiso
Speech
t, d deletion, 216-26 Variation, social: in language, 122; and
Tamil, 296 stylistic variation, 127; in lower middle
Tarde's Law, 286n, 308 class speech, 132; role in linguistic
Telephone: as source of linguistic data, change, 271-72, 308, 319-31; mentioned,
69, 89 112, 122, 262
Tempo, 81, 95 Vendryes, }., 262, 266, 269-70
Tests: two-choice and
four-choice, Verb: in Black English vernacular, 220-21.
149-58; for linguistic variables, 212-13; See also Copula; Past tense
psycholinguistic, 268. See also Subjec- Vineyard Haven, Martha's Vineyard, 4,
tive Reaction tests 14
(th): in New York City speech, 57, 77-78, Virginia, 11
115-16, 124; stylistic arrays for, 99-107; Vocabulary: on Martha's Vineyard, 7, 21;
class stratification diagram for, 112-13; changes in, 272. See also Word
subjective reaction tests for, 131-32, Volume: as channel cue, 95
154-55; stable variable, 134 Vowels, 8, 249n. See also Great Vowel
Tourism: impact on Martha's Vineyard, Shift;New York City dialect; specific
27-28 vowel forms
344 INDEX

Wages. See Income World War II: and US speech patterns,

Weinreich, U., xiv-xv, 22 116, 138, 158, 317
West Tisbury, Martha's Vineyard, 4, 14 Wyld, H. C, 2, 282, 285-86, 292
wh-attraction, 247
Whaling, 29 Yankee: on Marthas Vineyard, 29, 33, 34,
Whitney, W. D., 261, 266, 269 35, 38, 317-18
Women. See Sex differentiation Yiddish, 177, 298
Word: lists, 84-85; classes, 163, 246; as
object of social affect, 272, 321-22
Communications/ Linguistics/ Anthropology

William Labov
SOCIOLINGUISTIC PATTERNS
William Labov, an outstanding figure in the developing discipline of
sociolinguistics, has applied methods from sociology, anthropology, and
psychology, and has developed new techniques for the study of lin-
guistic variation. SOCIOLINGUISTIC PATTERNS presents for the first
time in one volume his most important work.

The book constitutes a systematic introduction to sociolinguistics, un-

matched in the clarity and forcefulness of its approach, and to the study
of language in its social setting.

CONDUCT & COMMUNICATION

A series of publications on new approaches to face-to- face
interaction, edited by Erving Goffman and Dell Hymes

Also Available:

William Labov
LANGUAGE IN THE INNER CITY
Studies in the Black Vernacular
William Labov advances conclusive arguments for the existence of the
black vernacular as a separate and independent dialect of English, with
its own internal logic and grammar. His analysis of this vernacular goes

beyond it to clarify the nature and processes of linguistic change in the

context of a changing society.

Dr. Labov's conclusions are mandatory reading for anyone concerned

with education and social change, with Afro-American culture, and with
the future of race relations in this country.

Erving Goffman
STRATEGIC INTERACTION
'We are indebted to Goffman for a new book, and to people for behaving
in ways
that continue to fascinate Like the hidden face in the picture,
. . .

it's hard not to see, once it's pointed out; Goffman not only sees it, but

makes the rest of us see it too.

T. C. Schelling

Ray L. Birdwhistell
KINESICS AND CONTEXT
Essays on Body Motion Communication
"Few pioneer workers have the opportunity that Professor Bird-
brilliant
whistellhas had to see his unique observations validated through techni-
cal innovation. So he was able to develop kinesics, which has now be-
come part of a systemic anthropological investigation."
Margaret Mead

cover design by: Cypher Associates, Inc.

Pennsylvania Paperback 52
University of Pennsylvania Press — Philadelphia.