REVIEWS

empirical data that are displayed in his work. Most of the articles are based on
data which he gained during altogether more than two years of field work in
North India, one deals with data on code-switching in Norway, one consists of a
comparative analysis of situations as culturally and linguistically distant as a
Norway and India community, one analyses data from a Mexican-American
bilingual situation in the southwestern United States. The geographical range is
as impressive as the power of theoretical and methodological analysis and as -
last but not least - the range of practical and political implications Gumperz
does not tire of pointing out: problems of language planning and construction,
communicative barriers for social change and innovation, educational policies at
the institutional level, all in linguistic situations which are considered basically as
structured in code-like ways.
One idea came to the reviewer's mind again and again while reading the
different pieces of work. One would wish that Gumperz might find time to put his
several pieces of work on India, those published in this collection and those
published elsewhere, together as a monograph. It would demonstrate even more
the truth of his conclusion that 'the analysis of speech variation should form an
integral part of the study of South Asian civilization' (91), as of course, of any
civilization. He has the data, the concepts and the methods at hand and un-
questionably disposes of the scientific competence to do so.
Reviewed by FRITZ SACK
Lehrstiihle fur Soziologie
Universitat Regensburg
(Received 8 April 1974) Regensburg, West Germany

WILLIAM LABOV, Sociolinguistic patterns. (Conduct and Communication, 4.)

Philadelphia: University of Pennsylvania Press, 1972.

INTRODUCTION
The 1960's in America saw the gradual and then increasingly widespread erosion
of the Chomskyan paradigm established by Syntactic structures and Aspects of the
theory of syntax. It is commonplace these days to identify Chomsky's contribu-
tion to linguistics as a revolution conforming to Kuhn's account of scientific
revolutions (1962). Far less agreement is to be found on the status of recent deve-
opments. Do they represent a counter-revolution (Katz & Bever forthcoming),
or the beginning of another revolution (Bailey 1971)? The answer seems to
depend partly on one's ideological alignment, and also partly on the extent of
one's willingness to find a common purpose among the disparate activities of the
'new linguistics' of the late sixties and early seventies. In many respects, this
'new linguistics' appears to be divided into two camps. On the one side there is
 LANGUAGE IN SOCIETY

the study of the concrete empirical facts of language in social context, as devel-
oped by Labov following sociolinguistics lines laid down in the late fifties and
early sixties by Bright, Ferguson, Gumperz, Hymes, Lambert, Weinreich (to
whom Labov expresses his greatest debt), and many others. On the other side is
generative semantics, searching for a grammatical theory substantially more
abstract than Chomsky's, but, for some of its proponents, arriving at a theory of
'pragmatics' ('the study of linguistic acts and the contexts in which they are
performed' [Stalnaker 1972: 395]), largely through introspection and the study
of available grammars, as in the work of Fillmore, G. Lakoff and R. Lakoff.
While Labovian linguistics and pragmatics, as developed within generative
semantics, seem methodologically and philosophically poles apart, they actually
share several fundamental principles.1 Both are concerned with 'language in use' :
we may compare Labov's focus on 'language as it is used in everyday life by
members of the social order' {Sociolinguistic patterns xiii) with claims that a
theory of generative semantics must study language in all its manifestations, in-
cluding use in social and cultural contexts (G. Lakoff forthcoming; R. Lakoff
' 1972). Therefore both are concerned with a branch of 'sociolinguistics' as op-
posed to linguistics in the strict sense of the word, which studies formal struc-
tures of grammar, usually without incorporating the concept of social meaning,
and also to the sociology of language, which is chiefly concerned with issues such
as standardization, bilingualism, education, especially curriculum reform, and
other applications of the study of language in context. Both accept a broadly
generative paradigm, although Labov appears, at least in the book under review,
Sociolinguistic patterns, to use it not so much as a theoretical base but as a
heuristic device. More importantly, both address themselves directly, though
in different ways, to the weak points of generative grammar, especially the
relationship between the abstract rules and their empirical interpretation,
between rules and actual speech (200-1). As a result, we find fundamental
questioning of the notion of absoluteness in grammar. The categorical is replaced
by the relative. Grammars no longer define the set of acceptable sentences, but
rather co-variation with linguistic and extralinguistic constraints (conversa-
tional, sociolinguistic, stylistic, as well as phonological, syntactic, or semantic).
In the work of some generative semanticists, even time-honored categories like
Noun, Verb, Adjective are shown to be so constrained by linguistic context as to
be arbitrary, and are replaced by relatively non-discrete 'squishes', represented
by matrices of syntactic interaction (Ross 1972). The common ground between
abstract logical and philosophical introspection on the one hand and massive
interviewing and experimenting on the other can perhaps best be found in
Fillmore's program for the 'working grammarian' and Hymes' delineation of the

[1] It is informative in this connection to compare points made about generative semantics
in Katz & Bever (forthcoming) and those made about 'empirical' linguistics in Bailey
(1971). There are, of course, other ways of testing such hypotheses.
90
 REVIEWS

scope of sociolinguistics at the 1972 Georgetown University Round Table

(Fillmore 1973; Hymes 1973a). It can be inferred, too, from the mutuality of
endeavour, despite sharp disagreement on particulars, at the 1972 conference on
'New ways of analyzing variation in English' (Bailey & Shuy 1973), and the
1973 LSA Summer Institute on 'Language in Time and Space'.
This is not to say that there are no important theoretical, as opposed to
methodological, differences between the two sides. On the contrary, as Socio-
linguistic patterns so forcefully demonstrates, for Labov language is not only
heterogeneous and constrained by non-linguistic factors, but furthermore, it is
constantly changing; therefore, a theory of language must be a dynamic one, a
position largely not held by the generative semanticists, who are primarily
concerned with a static model of language.
The Labovian paradigm, then, is essentially a radical departure from the
transformational framework of the sixties, though obviously as much dependent
on that model as on earlier sociolinguistic studies. In the modesty so character-
teristic of this book, Labov, however, claims that we do not necessarily need a
'new theory of language'; rather, we need a 'new method of work' (207). The
facts presumably speak for themselves, once approached in the right way, that is,
by taking up 'a number of problems where progress has been blocked' (259).
Not revolution, but quiet, thoughtful acceptance of the implications of empirical
study, in addition to the other aspects of linguistics, is what Labov claims to seek
(259). But just as a new way of thinking may result in a new way of doing, as in
Chomsky's case, so a new way of doing may result in a new way of thinking, as in
Labov's. His theory necessarily takes on a new shape, and demands an essentially
different view of the scope of linguistic study.
Labov's exploration of 'a new method of work' over the last ten years has been
so extensive that it would fill many volumes. Most of his results, other than his
dissertation (1966) have appeared in the form of articles and reports. Socio-
linguistic patterns is one of two books that have recently been published by
Pennsylvania University Press, gathering together in more or less modified form
materials published in scattered, though by no means inaccessible places. This
volume focuses on the methods and theory of studying language in a dynamic
framework. The examples in the first seven chapters are mainly from phonology;
in fact, the first seven chapters, which present extensive materials from his MA
thesis on linguistic patterns in Martha's Vineyard (1963), and his dissertation on
New York (1966) might be called 'Variations on the variables ay, aw, r, eh, oh,
th, dh: method and theory.' The last two chapters, 'The study of language in its
social context' and 'The social setting of linguistic change', which together form
nearly half of the book, are more recent, and more theoretically oriented; they
range over the whole field of grammar, but their main emphasis, too, is phono-
logical. (The companion volume, Language in the inner city, provides in depth
analysis of Black English Vernacular as spoken in large urban areas, based on the

91
 LANGUAGE IN SOCIETY

method and theory developed in Sociolinguisticpatterns; see review by Wright in

next issue.)
Given the diversity (and familiarity from previous publication) of parts of the
book, it seems best to concentrate on the whole, for gathered together, the parts
take on a deeper significance in the larger scheme of theory-making; moreover,
the book is proclaimed on the back cover as a 'systematic introduction to socio-
linguistics'. I shall take stock of the methodology and theory presented in this
book, while recognizing that the full Labovian paradigm encompasses more.

METHODOLOGY
It is not surprising that the book is essentially a series of studies in empirical
methodology. The format, however, somewhat obscures demonstration of the
need for such methodology and of its theoretical implications. The reader would
do well to turn to Labov 1972 for a concise, but fuller view. In brief, in the
1940's and 1950's analyses were justified by formal models couched in terms of
induction and idealized 'discovery procedures', but practical work was left
informal. (Pike's concern to integrate theory and practice was marginally
heeded.) Transformationalists have been little interested in field work, arguing
that one must know a language natively to do analysis in depth, and ruling out
study of variability through focus on idealized homogeneous systems. Concern
for empirical methodology has been associated in the eyes of most linguists with
models of grammar unable to rise above observational accuracy. Against this
background, one of Labov's major contributions has been explicit development
of sophisticated methods of data-collection, ranging over contextual spectra,
within a broadly generative framework, methods designed to help 'hold up as the
highest goal of linguistics, the possibility of being right' (Labov 1972: 98; see also,
with less emphasis, 259) - right not only as to observation, but as to structure
and theory. One may evaluate and verify hypotheses in ways that do not depend
upon conflicting judgements, and that provide truly cumulative knowledge. Such
hypotheses may be formal, as are the bracketing conventions of transformational
grammar, which Labov verifies by showing that many contexts function inde-
pendently of each other, not incrementally, as constraints on the probability of
co-occurrence of a rule (231);2 or they may be axiomatic, such as the neo-gram-
marian hypothesis that sound change precedes fluctuation, which Labov rejects
on the basis of evidence that fluctuation and temporary dissolution of word-
classes precede sound change and regularity (165, n. 6, 246).
Sociolinguistic patterns explores mainly, but by no means exclusively, socio-

[2] Kiparsky, it should be noted, argues for the 'psychological reality', hence validity, of
just such rule bracketings, on the grounds that rules 'collapsed by braces form units of
a kind which can undergo systematic change' (1968:179).
92
 REVIEWS

economic, stylistic, and age-related parameters of variability.3 The crucial

question, discussed at greatest length in Chapter III, on 'The isolation of con-
textual styles', is how to obtain relatively unskewed data on the vernacular: 'the
style in which the minimum attention is given to the monitoring of speech' (208).
Linguists themselves cannot be a source of such data since their particular set to
the problem in question inevitably structures their responses; in fact, linguists
have often been known to develop a language style peculiarly their own. How-
ever, interviews by no means ensure unbiased data either. Labov dwells at
length on the 'Observer's Paradox' that the interviewer's presence itself changes
the structure under investigation (209). This is of course not a problem just for
linguistics, but for psychology and sociology and all interview- and experiment-
dependent study. While formal styles are relatively easy to elicit through reading
of minimal pairs, the vernacular is almost impossible to tap, except by sneak
recordings as an anonymous bystander. Opposed on ethical grounds to such
recordings, Labov's most-used strategy, as described in this book, was to ask
his interviewees to read dialogue passages especially designed to encourage
unconscious use of style-shifting, such as the simulated conversation of teenagers,
and to answer questions about personal subjects like fear of death (93). Data
obtained in such ways has the advantage of being closely comparable to data
obtained from other interviewees, but it has its limitations, beyond the obvious
one of the Observer's Paradox. The written materials presuppose a considerable
degree of literacy and are therefore not uniformly applicable. Furthermore, to
be useful, the written passages have to contain key phonological items. Too many
too close together may cause difficulties (105 n. 13); furthermore, the self-
consciousness of the passages surely cannot escape intelligent readers when they
come across such obviously forced narrative as: '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' (82: dark vs. dock); 'she had
to have a chocolate milk at Chock Full O'Nuts. Chalk this up as a total loss, I
told myself (83: chocolate, chock, chalk). Interview techniques also have the
disadvantage of being to a large degree dependent on the investigator's own
social rapport with the interviewees, and on the availability of good (i.e. expensive)
recording equipment. Recently Labov has largely abandoned the reading materi-
als, and has turned from interviewing individuals, who necessarily present an
'idiolect', if variable, to group sessions (as mentioned in 109, and detailed in
Labov 1973). Groups highlight the importance of social interaction in language
use, and contribute somewhat to solution of the problem of how to deal with
large communities (247), but still suffer from dependency on the investigator's
personality and equipment.

[3] See Pride (1973) for a useful summary of the different views of speech act theory in
sociolinguistics, as presented in Gumperz and Hymes.

93
 LANGUAGE IN SOCIETY

THE CONCEPT OF 'RULE*

Sociolinguistic patterns insists on the importance of regarding the data collected

not as the ad hoc product of interviews which will merely provide information on
the frequencies with which certain speakers use certain patterns, whether studied
as individuals or as groups, but rather as material from which the linguist can
derive rules that define the potential for use (231). The rules themselves are
largely adopted from generative grammars (252) and then redefined in terms of
their probability of occurrence, given certain specifiable contexts.
On first glance, these rules look deceptively like transformational rules. The
'consonant cluster simplification rule' (whereby bold, fast, find, walked, digged
and so forth are frequently pronounced with a single final consonant, especially
if there is no preceding morpheme boundary) is presented informally, that is,
without quantification, as:
[-cont]-><0>/[ + cons]<0> # # <syl>
where (y indicates variability (220). The formalism, and the claim that we
need no new theory of language, must not, however, mislead us into thinking
that Labov is presenting us with a transformational grammar, modified by
introduction of quantitative probability, so that the 'free variation' of optional
rules is redefined as 'constrained variation'. He is clearly working in a generative
tradition that requires us to 'generate', that is, explicitly assign structural
descriptions to possible structures. However, the rules account not for an ideal-
ized homogeneous structure, but rather for heterogeneity where it exists - rules
may therefore be categorical or variable (252). Heterogeneity is not only intra-
lingual, but also extralingual, a function not only of phonological, syntactic, or
semantic conditioning, but also of other factors found to be relevant in a com-
munity, such as age, style (formal, vernacular and so forth), socio-economic
status and education (115), ethnic background (118), social mobility (286 n. 9),
sex (141, 301-4), hearer and setting (237), and presumably also geographic
location. The boundaries between linguistics and non-linguistics are partially
dissolved in the statement of constraints. The dissolution is not complete, since
non-linguistic factors play a role only as contexts, not as input to structural
changes in a rule; furthermore, linguistic contexts are quantifiable, not so, at
least as yet, stylistic and other contexts (241).
The rules further differ from their counterparts in transformational grammar
in being 'speakers' rules', or producer rules, not language-users' rules (225).
Labov insists that grammars should predict probabilities (which are fixed), hence
be models of potential behavior, as opposed to statistical frequencies (which are
randomly variable, and hence samples of actual behavior); he thereby suggests
that he is dealing with linguistic 'competence'. Indeed Cedergren & Sankoff
(forthcoming), to whom Labov acknowledges a great debt for quantitative analy-

94
 REVIEWS

sis, explicitly and repeatedly speak of competence with reference to speaker's

potential, and Labov himself uses this term when he says the ability 'to accept,
preserve, and interpret rules with variable constraints' is an important aspect of
human beings' 'linguistic competence' (226). However, his rules are not 'com-
petence rules' in any sense understood by a transformationalist, who defines
competence as what the speaker-hearer knows. Instead, Labov's rules reflect
what a speaker may do (potential speech). The separation of speaker and hearer,
producer and interpreter, is followed through with evidence that variability is
largely a function of speaking; perception, however, tends to be categorical and
stereotypic (226). Labov's evidence for the separation of production and per-
ception is based on interview studies of speakers' subjective evaluation of their
own and others' speech. It is widely supported by work in other fields, most
notably developmental psycholinguistics, where enormous differences can be
shown to exist between imitation (non-spontaneous production based on per-
ception), and spontaneous production. Developmental studies have, however,
also demonstrated that comprehension, judged from action in response to per-
ceived utterances and action correlating with spontaneous production, may be
different and more comprehensive than either production or perception, and that
therefore a system of comprehension somewhat analogous to speaker-hearer
competence may be presumed to exist, functioning as a link between the com-
petences for perception and production (for some discussion see Ferguson &
Slobin 1973, especially 465-521; similar conclusions have also been reached in
studies of pidginization and creolization processes, Hymes 1971 Part III). Such
a language-user's competence is not denied by Labov, but seems to be left as a
nebulous possibility in the enigma of what it is that constitutes a theory of
language.
A rule, then, appears to be essentially a statement of the probability that a
speaker will apply a certain mapping of more abstract rules onto a less abstract
set. Labov tells us nothing about his view of the psychological reality or unreality
of rules. Since they structure variability and since certain variables, such as
r-constriction, become norms in certain communities, and may be subject to
more or less conscious manipulation (178-80), rules presumably are intended to
model some sort of psychological reality. They certainly have social reality in
that they may be object of evaluation. However, 'social value', that is, identifica-
tion with a group, is entirely dependent according to Labov on variability, and
is therefore 'parasitic on language' (322) and by no means co-extensive with
linguistic structure. Rules apparently also have some biological reality in the
sense that the function of variability is correlated with the function of diversifica-
tion of the species in the Darwinian sense - linguistic diversification is regarded
as necessary for the survival and evolution of language, culture and society: 'I
am inclined to believe that the development of linguistic differences has positive
value in human cultural evolution - and that cultural pluralism may even be a

95
 LANGUAGE IN SOCIETY

necessary element in the human extension of biological evolution'(324). This

view is admittedly speculative, but it is an important extension of the new para-
digm, suggesting that a theory of language must admit that language as well as
human society and biological species are subject to evolutionary adaptation. The
potential equality of all languages to fit the needs of the community has been a
fundamental point of agreement among linguists in this century; denial of this
premise seems to strike at something far more fundamental than denial of the
value of synchronic, homogeneous models - Greenberg (1966) and Sapir (1921),
and many others, especially of the Prague school have, after all, propounded
dynamic models of language and Sapir certainly saw language as variable and
socially constrained in something close to Labov's sense, but all regarded lan-
guages and dialects as potentially equal. However discomforting the argument
that languages are adaptive, everyone should read Hymes' article, 'Speech and
language: on the origins and foundation of inequality among speakers' (1973),
which strikes at the very roots of the problem: 'A perspective which treats
language only as an attribute of Man' (i.e. potentially equal for all) 'leaves
language as an attribute of men unintelligible'; the human species and human
culture are obviously adaptive; how can it be that language is not 'in large part
what users have made it'? (Hymes 1973b: 60). Hymes' philosophical question
strikes at a corner-stone of current linguistics; relatively unanswerable in terms
of the current limits of Labovian methodology, the same issue, raised not so much
as a philosophical question but as expression of personal belief (273-4, 323—5)
may elicit a less thoughtful reaction from Labov's readers, but it is not any the
less to be ignored.
The function of variable rules in language, therefore, is viewed as provision of
the means by which language diversification and adaptation can take place. There
is a kind of cyclic effect; variability may give rise to social evaluation, and evalua-
tion may give rise to social affect and ultimately new rules. It is this view of the
function of rules and of social evaluation of those rules that makes comprehen-
sible the statement 'as far as the synchronic aspect of langugage structure is
concerned, it would be an error to put much emphasis on social factors' (251).
The 'sociolinguistics' of this book is a study of variability: methods of analysis,
the structure and nature of variable rules, the implications of covariation between
language and social structure for linguistic change. It is not a study of language
as a social tool.
In this respect the book seems relatively far afield from the concern of many
sociolinguists who study the function of language for people as social beings
(see, for example, Halliday 1973, and work based in large part on speech act
theory and ethnography of speaking, as represented in Giglioli 1972; Gumperz
& Hymes 1972; and Sudnow 1973). Labov in no way denies the importance of
developing a theory of the function of language, and hence of linguistic rules in
general as opposed to variability in particular. Indeed, there is tentative discussion

96
 REVIEWS

of the function of language in a brief outline of discourse rules (252-8); more

importantly, much of Language in the inner city is devoted to developing a theory
of discourse, and in the first volume of this journal he clearly states that 'the
basic linguistic conundrum' is ' "Why does anyone say anything?" ' (1972: 114).
In light of the claim that Sociolinguistic patterns is an introduction to socio-
linguistics, however, a fuller discussion of the role of language in society would
clarify the ways in which language is correlatable with social context, and would
make the argument as to the adaptive function of variability more convincing.
Given that Labbv's theory of linguistic change (see below) requires placing
change in its social matrix (162, 283), his focus on co-variation between linguistic
and social structure leaves unforged the essential link in the argument - why
does a language exist for a society and the individuals that constitute it?

THEORY OF LINGUISTIC CHANGE

As we have seen, 'sociolinguistics' in its most meaningful sense is conceived as

a branch of historical linguistics. Indeed, the major theoretical contribution of
Sociolinguistic patterns as a whole is to the study of linguistic change. This term
is used advisedly. It is important to distinguish two conceptions of historical
linguistics, summarized as 'diachrony' and 'linguistic change'. The Saussurean
concept of diachrony, as opposed to synchrony, has implied a strict separation
between dynamic and static aspects of language. Although the dichotomy has
been subject to criticism almost since it was made, and work^has been done that
transcends it (especially in the Prague School), certain associated ideas have
persisted, namely, that (1) a change occurs only when the system changes (the
problem being, whose system? idiolect-based grammars seem to suggest that of
the individual, rather than of the community (Weinreich, Labov & Herzog
1968)); (2) the only way to study change is to compare synchronic accounts of
successive systems; (3) change cannot be observed in progress. In contrast to
'diachrony', the concept of 'linguistic change' studies change as a special case
of variability within a system that is by nature variable and dynamic, and has
potential directionality. Recent studies of historical linguistics in relation to
typology (e.g. Greenberg 1966); to language acquisition, whether by children or
adults (e.g. Baron 1972, Traugott forthcoming a, b); to articulatory processes
(e.g. Ohala forthcoming; Stampe 1969); to perceptual processes (e.g. Bever &
Langendoen 1972), and to variability, whether social, stylistic, geographical, age-
or sex-related, and so forth (e.g. Bailey 1973; Bickerton forthcoming; Hymes
1964, Part VIII, and 1971; Weinreich 1953; and above all, the work of Labov)
have increasingly undermined the Saussurean dichotomy. In particular, it is
recognized that change can at least be inferred from contemporary data (thus,
currently observable stages of acquisition can be projected onto the earlier
history of the language, or even forward into the future, or be used to develop

97
 LANGUAGE IN SOCIETY

hypotheses about processes such as are involved in pidginization and creoliza-

tion). Labov makes the even stronger claim that change can actually be observed
in progress, either in real time (comparison of the language of the community at
ten or twenty year intervals) or, more feasibly, in apparent time (comparison of
the language used by speakers of different ages at the same period of the language)
(163). Labov's theoretical claims are formulated chiefly in Chapters VII, 'On
the mechanism of linguistic change', and IX, 'The social setting of linguistic
change'. Those wishing to explore the data and arguments in more detail
should not fail to read Labov, Yaeger & Steiner (1972), and Labov (forth-
coming).
The most interesting part of Labov's discussion of linguistic change involves
answering the question 'What is change?'.4 Labov's unequivocal answer in
Chapter IX is that 'the language has changed only when a group of speakers use
a different pattern to communicate with each other' (277). The traditional
dichotomy between origin and propagation of a change is, he claims, useless:
'the origin of a change is its "propagation" or acceptance by others' (277).
'Change occurs when a certain variable moves in a predictable direction' (162),
that is, comes to have a certain significance in the community.
Fundamental to Labov's view of language change is what he means when he
asks, with reference to the increased centralization of (ay) and (aw)5 in Martha's
Vineyard among younger speakers, 'Is this an example of sound change, or is it
merely evidence for a regular change in speaking patterns which is correlated
with age?' (21). The wording, using 'change' in two different ways, gives us a
foretaste of the difficulties there are in answering the question. The clearest
indication of change, according to Labov, in when there is a marked lack of
parallelism in the stratification of variability. Patterns like:

B C D

show only difference,* but are not 'in change', even when correlated with age
(apparent time), provided the parallelism remains; such patterns are found for
(th), (dh), (ing). However, patterns like the following:

[4] For exploration of various answers to this question, with emphasis on the importance
of studying variability, see Householder (1972).
[5] Parentheses indicate that the linguistic form within them is variable.
[6] The lines indicate social stratification; A indicates casual speech, B careful speech,
C reading style, and D word lists.
 REVIEWS

A B C D
are 'in change'; such patterns are found for (r), (eh), (oh) (127, 245). The
cross-over principle, where one class uses a variable far more extensively than
the next higher one on the social scale, is a key indicator of change. This parti-
cular kind of change, demonstrating movement from lower to higher social
status directly proves fake the idea of a uniform standard language surviving and
changing in self-contained uniform ways - this is, of course, not a new point (cf.
Sapir said, 'the uncontrolled speech of the folk' provides the information for the
future, not the 'controlled', standardized form (1921: 156)) ;7 nevertheless, much
historical work perforce has compared standard language (or at least controlled,
literary language), because that is often all that remains to us. The kind of cross-
over pattern under discussion is a case of 'hypercorrection', or generalization
beyond the probability patterns of some model group. Hypercorrection is thus
promoted to the status of a regular and essential ingredient in socially-correlated
linguistic change, rather than a concept used merely to define such random
phenomena as use of between you and I. This integration of'hypercorrection' into
structured processes of change is somewhat similar in import to Kiparsky's
integration of analogy into the category of systematic linguistic generalization or
simplification (1968). It also promotes the importance of stylistic correlates to
variables, since the cross-over principle is observed primarily in connection with
speech that is careful to formal, not in vernacular varieties (cf. especially Chapter
V: 'Hypercorrection as a factor in linguistic change').
Not all sound-change, however, is to be identified with hypercorrection and
cross-over patterns - for one, stigmatization may lead to suppression of variables,
or to the redevelopment of parallel patterns (295). Labov identifies thirteen stages
of change (178-80), starting with change below the level of consciousness, and
ending with change subject to conscious awareness, as in the case of stereotyping
and stigmatization. The first stage is origination within a restricted subgroup as
an 'undefined linguistic variable' at a time when the separate identity of the
group is weakened (hardly the spread which is change!). At this first stage, before
the variable is identifiable and therefore before it has acquired any real status as
the marker of a group, one is surely hard put to identify the change in progress.
This stage cannot be subject to questions about real change or not, and seems

[7] Indeed, similarities throughout to Sapir, in thought and word, are striking; it is
surprising not to find him mentioned either in Sociolinguistic patterns or in Weinreich
Labov and Herzog (1968), which takes a considerably more comprehensive view of the
history of historical linguistics.

99
 LANGUAGE IN SOCIETY

dangerously close to renewing the origination-propagation dichotomy (indeed,

this dichotomy is still made on pages 23 and 123). True change must be identified
at least with group membership (stage 2). (aw) and (ay) on Martha's Vineyard
were not studied with sufficient depth for plotting on either the parallel or the
cross-over grid; nevertheless, it is argued that they do reflect change in progress,
since they are clearly identified with the groups associated with the island, as
opposed to the mainland, though largely at the unconscious level, and pattern in
a curvilinear wedge-like way, with dramatic increase of centralization among
speakers aged 31-45. If so, another model of change apart from the cross-over
pattern seems to be utilized:

I II III

where I indicates the oldest generation and III the youngest. Relatively uncurved
lines would represent mere difference, not change.
Labov suggests that among the important questions to ask in developing a
theory of linguistic change is how one stage evolved into another (the 'transition'
problem), how the change is embedded in the social and linguistic matrix (the
'embedding' problem), and what the subjective reaction to change is (the
'evaluation' problem) (162). The latter question is so atypical of most historical
linguistics that it seems at first a totally unequal and surprising member of the
triad (although Sapir had already pointed to the implications of the dichotomy
between 'maintaining caste' (conscious evaluation) and actual use (1921: 157)).
The importance of the evaluation problem to Labov becomes immediately clear
in the light of what he means by sociolinguistics, and his view that social factors
have significance to the extent that they may bring about social affect. That
speaker and hearer rules may differ implies that they may have different social
affect; any theory of language change concerned only with speaker rules would
therefore be unjustly skewed. Since Labov concentrates on speaker rules, the
evaluation problem is not made as central to theory as the transition and embed-
ding problems, but some very important findings on the subject are discussed
throughout the book and especially in Chapter VI: 'The subjective dimension of
a sound change in progress.' It is argued, from extensive empirical evidence,
that hypercorrection is correlated with hypersensitivity; speakers have positive
reactions to prestige forms (135, 144), or negative ones to stigmatized forms
(130), that bear little relation to their own use (131). Since there can be no
hypersensitivity where there is no change, hypersensitivity can be a clue to
whether a change is in progress or not. However, if Labov's thirteen stages are

100
 REVIEWS

indeed all stages of change, hypersensitivity can only relate to later stages of
change when there is consciousness of certain variables, and it is not a necessary
condition for detecting language change in progress.
Since hypercorrection is typical of linguistic change except in its earliest stages,
and since people are sensitive to stigmatized forms, why do stigmatized forms
persist? The answer, it is shown, cannot lie in laziness and so forth; rather, the
answer lies in the neurological fact that people cannot accurately monitor them-
selves on the one hand, and in the social fact of need for group-solidarity on the
other. A distant target such as a nonstigmatized form, may be an object for
attack, a close one for amicable compliance (250); furthermore, as we have seen,
language must be variable for survival of the culture. Variation within groups is
minimal (106-8; cf. also Sapir 1921: 148-50); that is, though individuals may
vary, their overall structured behavior is like that of their peers; between groups,
however, distinction can arise as a function of social space; hence mixing in urban
communities may lead to levelling within groups, but diversification between
them (143, 300), and the supposed paradox of geographically-defined levelling
but socially-defined diversification is resolved in the axiom that vertical, social
stratification is more distinct than horizontal, geographical stratification (cf. also
Bailey 1973). The reverse problem, why language remains as uniform as it does,
is barely touched on, but relative uniformity can presumably also largely be
explained in terms of group-loyalty (Sapir's 'maintaining caste'); it is, of course,
not just stigmatized forms that are suppressed, but a vast number of innovations
that barely reach the level of social let alone personal consciousness.8
It is difficult to criticize Labov's work since he so assiduously points himself
to problems and gaps in the theory (e.g. 162 n. 3, 245-7). However, three
particular difficulties with the view of linguistic change as presented in Socio-
linguistic patterns can be identified here, all pertaining to generalization of the
theory.
(i) As Labov says, a 'uniformitarian principle' is essential. The study of change
in progress is valid only if the results can be projected onto known historical
change. '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' (275). Even though the studies, with the exception of
those in Martha's Vineyard, are largely of urban communities of a size and
structure undreamed of before the nineteenth century, it does seem reasonable
for us to reconstruct stages of change on the principle that changes are embedded
in certain specifiable ways within the social as well as linguistic matrix, and that
group solidarity is a major factor in the diversification as well as levelling of
language. We must, then, recognize that the view that a homogenous language

[8] For recent discussion of this and other important issues of 'actuation', see Householder
(1971 Chapter XVI).
IOI
 LANGUAGE IN SOCIETY

either can or should be reconstructed is inconsistent with the principles of

linguistic change. This has long been recognized; however, the comparative
method of reconstruction has always needed to assume relative uniformity in the
proto-language, because it cannot be applied without regarding 'all inter-con-
sistent reconstructions as inter-relevant for the purpose of comparative analysis'
(Dyen 1969: 517). No other method of reconstruction, however much influenced
by consideration of universals, language acquisition, or sociolinguistic theory, has
yet been developed which can assume otherwise, although some provocative
general theoretical comments on the possibility of integrating the various per-
spectives of historical linguistics have been made by Bailey (1972). We certainly
find no clue to how this integration is to be effected in Sociolinguistic patterns;
however, a significant beginning is to be found in Labov et al. (1972) with
reference to the Great Vowel Shift. Unfortunately, since the texts on which we
base our historical reconstruction are so limited and are largely preserved only
because they were considered of some cultural value, largely historical, moral, or
pedagogical, we cannot, in fact, reconstruct many, or any, accurate accounts of
socially-correlated change. Plays and some novels provide the best source of
evidence for variability ;9 however, unless the author was aware of his or her own
variability in ways atypical of most language users, and perceived the variability
of others in non-categorical ways, such texts should provide us with stereotypes
rather than patterns of speech (201), so even such texts will be of limited help.
Furthermore, most texts represent the written style(s) of language, and therefore
are subject to constraints not under investigation in Sociolinguistic patterns; most
obviously, written language is on the whole more conscious and less variable than
spoken language. The limits imposed by the documents available for study of
past history bring up the all-important question: what are the patterns of
variability that covary with 'register' - that is, with discourse patterns used for
various functions of language (cf. Ellis & Ure 1967; Ferguson 1959), whether
baby-talk, the language of buying and selling, the format of the short story, of
the picaresque novel, of 'rapping', etc. The list is enormously long, and many
interesting studies on the subject are forthcoming; however, register as correlated
with change has regretfully drawn little attention since the early twentieth
century, when studies of subjects such as the change from the Old English
paratactic (basically co-ordinating) to the Middle English hypotactic (basically
sub-ordinating) system abounded. The spoken language presumably did not
change significantly from Old to Middle English times with respect to co-
ordination and subordination, but the choice of the structures suitable for certain
subjects did as literary and other forms of discourse developed in the vernacular.
In the contemporary situation, the influence of sermon registers on Black Eng-
lish Vernacular, and of political rhetoric on teenagers seem obvious even to the

[9] See Salmon (1965) for a pioneering study of 'colloquial language' in Shakespeare.
102
 REVIEWS

layman; their effect can be seen at all levels - syntactic, lexical, and phonological,
and suggests that register stratification may be an important factor in change.
(ii) If Labov is perhaps over-confident about the possibility of projecting his
findings onto the past, or at least.the remote past, he is overly doubtful about the
possibility of projecting them onto levels of grammar higher than the phonetic.
Led by his failure, at the time when Sociolinguistic patterns was written, to
identify significant evidence for regular social distribution, including assignment
of social evaluation of variables at higher levels of the grammar, Labov concluded
that such rules are largely not subject to patterned variation and therefore do not
have social value, at least not until the variable rule becomes nearly categorical
(251, 321-3). Recent studies have in fact shown that higher level rules, phono-
logical and syntactic, are indeed subject to patterned variation (e.g. Laberge 1972;
Sankoff 1973; and Labov's own Language in the inner city). Although these
studies of syntactic structures are almost all concerned with grammatical cate-
gories such as quantifiers, negatives, pronouns, and auxiliary verbs, all of which
are normally realized as function words, and therefore often subject to phonetic
constraints, they nevertheless clearly demonstrate that a dichotomy between
lower and higher rules is not to be made. This distinction, had it existed, would
have caused grave difficulties for Labov's theory. For one, it would have called
deeply into question the definition of change as propagation socially defined, and
would have denied the claim that 'one cannot understand the development of a
language change apart from the social life of the community in which it occurs'
(3), for it would have shown that a large part of variation was 'free' after all. It
would certainly have severely limited the findings of patterned variation. It
would also have suggested that there is no clear relation between syntactic rules
that are significantly affected by phonetic conditioning (e.g. inflectional rules)
and those that are not, such as word order. It is interesting to see how deeply
this possibility affects Labov's discussion of syntax in Sociolinguistic patterns, and
how problematic it permits treatment of interconnections between different but
related syntactic rules to become; an example is given in the next section, on the
problems of generalizing findings on the directionality of change.
(iii) It is claimed that variable rules will help us explain the directionality of
change in ways that Halle's description of change (1962) does not, when it argues
for rule addition by adults, normally introduced at points of break in the gram-
mar, and rule simplification by children (162 n. 2). Labov's system can explain
variation within a rule once it has begun to take on a direction, but we must not
conclude that we can therefore explain directionality in its wider sense. While it
is recognized that only 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' (175) and a sketch of interrelated covariation among the tense-
vowel variables is given, it is unclear how directionality would be handled where
relationships between rules at different levels of the grammar are at stake,

103
 LANGUAGE IN SOCIETY
particularly if they are not subject to the same social constraints as phonetic rules.
In his all-important rider to variability theory, Labov says: '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 structural change
to compensate for the loss of information involved' (223); such structural change
is 'dramatic'. One example that he cites is the 'phonological and grammatical
simplification' that in some English-based Creoles 'has effectively reduced clusters
so that final inflections are typically eliminated altogether' and past tense -ed
has been replaced by do (224). The wording suggests one rule takes over when
another is lost. A host of questions arises. How, without some overall theory of
'psychological space' (to borrow Sapir's term) or of the function of language, can
one account for the fundamental principle in language change that structures
which are lost usually reappear, often manifested elsewhere in the surface
structure (cf. Kurylowicz 1972). Such an explanation would presumably require
that certain semantic primitives and relations must be given some surface
manifestation for language to be language and to be an adequate tool for linguistic
use.
The concept of psychological space suggests that obscuring of a semantic
relation may be possible only if some other manifestation is in competition, hence
that compensation precedes rather than follows invariance. Such a sequence of
events is indeed postulated for the loss of English inflections, though with great
hesitation: 'we can hazard the guess that some of the compensating changes in
word order had already occurred' (322).to But it is only for the English inflections
that gradual change is considered. Other morpho-syntactic changes are considered
dramatic. The issue of the Creoles is complicated by the fact that Labov appar-
ently holds, in the passage cited, that Creoles derive from superstrate speakers
simplifying their system, rather than from substrate speakers with a different
system hypothesizing about the superstrate (cf. Bickerton forthcoming; Whin-
nom 1971). But, influenced on the one hand by the Morpheme Structure Rules
of their own languages and on the other by perceptual strategies which may have
interpreted the word-final consonant cluster reduction of English speakers as
invariant (in keeping with Labov's comments on perceptual strategies), it may
be that Creole speakers never had any 'consonant cluster reduction rule'. Instead,
it is more likely, considering evidence from languages like West African Pidgin
English (Schneider 1966), that they used adverbs to express semantic tense
relations and periphrastic do developed as a tense marker in competition with
the past tense adverbs; if this is so, -ed was derived by decreolization processes
and came to be in competition with do and the adverbs. The directionality, then,

[10] Labov's unwillingness to do more than hazard a guess on the grounds that systematic
patterning at the syntactic level had not yet been possible, presents the danger of a
return to mere physicalism and observationalism, and of projecting onto our study of the
past only that which is observable today.
IO4
 REVIEWS
was from analytic to synthetic, and the change was gradual, variable, and certainly
not dramatic.
The direction of change postulated for the Creole do parallels to some extent
the direction of acquisition of language. While the particular stages of acquisition
a particular child undergoes cannot constitute change in Labov's sense, since
the child's restructurings are internal and do not have meaning for the group, it
is nevertheless important for a theory of language change to seek to account for
directionality in its global sense with reference to broad typologies of acquisition
as well as social context, since it is only through acquisition that structures can
come to have social meaning and spread (that is, acquisition is a prerequisite for
change, even if it is not change). It is through acquisition that pidgins and Creoles
come into being (pidgins are learned, and were presumably first developed,
largely by adults, and Creoles are developed by children, as can currently be
witnessed in the development of first-generation Tok Pisin Creole (Sankoff &
Laberge 1973)). Furthermore, although language acquisition is not isomorphic
with change, and indeed often is the mirror-image of change, the frequency of
either parallelism or mirror-imaging is so great as to suggest a fruitful source of
explanation for change. To this extent, even though his identification of the
adult's role in language change as rule elaboration and of the child's as rule
simplification is open to question, Halle does in fact build potential directionality
into his theory of language change, despite Labov's claim to the contrary,
provided one can presuppose a metatheory including markedness hierarchies.11
In both theories, Halle's and Labov's, limited directionality can then be ex-
plained, but at entirely different levels of particularity.

CONCLUSION

The doors that Labov has unlocked open up on a world of variability so struc-
tured and so extensive as to bring into question many of our familiar modes of
thinking on fundamental issues and to call for new exploration. If, like other
major original theoretical or methodological statements such as Chomsky's
Syntactic structures, Sociolinguistic patterns focuses too much on some aspects
of the problem without sufficient attention to others, and if, like other major
'introductions' such as Lyons' Introduction to theoretical linguistics, it some-
times assumes too much for the beginning students, like them it also presents a
challenge that cannot be ignored, and like them has enormously enriched the
field of linguistics. It must be counted among the dozen or so works that
make up the core of any linguist's library, whatever his or her theoretical
alliances.

[11] See suggestions concerning the content of such a metatheory in Chen (1973).
 LANGUAGE IN SOCIETY

REFERENCES
Bailey, C.-J. N. (1971). Trying to talk in the new paradigm. Papers in linguistics 4. 312-38.
(1972). The integration of linguistic theory: internal reconstruction and the com-
parative method in descriptive analysis. In R. P. Stockwell and R. K. S. Macaulay (eds),
Linguistic change and generative theory. Bloomington: Indiana University Press. 22-31.
(1973). The patterning of language variation. In R. W. Bailey and J. L. Robinson
(eds), Varieties of present-day English. New York: Macmillan. 158-86.
Bailey, C. J. & Shuy, R. W. (eds) (1973). New ways of analyzing variation in English.
Washington, D.C.: Georgetown University Press.
Baron, N. (1972). The evolution of English periphrastic causatives: contributions to a general
theory of linguistic variation and change. Ph.D. diss., Stanford University.
Bever, T. G. & Langendoen, D. T. (1972). The interaction of speech perception and
grammatical structure in the evolution of language. In R. P. Stockwell and R. K. S.
Macaulay (eds), Lingustic change and generative theory. Bloomington: Indiana University
Press. 32-95.
Bickerton, D. (forthcoming). On the nature of a Creole continuum. Lg.
Cedergren, H. J. & Sankoff, D. (forthcoming). Variable rules: performance as a statistical
reflection of competence. Lg.
Chen, M. (1973). On the formal expression of natural rules in phonology. JL 9. 223-49.
Dyen, I. (1969). Reconstruction, the comparative method, and the proto-language uni-
formity assumption. Lg. 45. 499-518.
Ellis, J. & Ure, J. (1967). Language varieties: register. Encyclopedia of linguistics, informa-
tion and control. Oxford: Pergamon.
Ferguson, C. A. (1959). Diglossia. Word 15. 325-40.
Ferguson, C. A. & Slobin, D. I. (eds) (1973). Studies in child language development. New
York: Holt, Rinehart & Winston.
Fillmore, C. J. (1973). A grammarian looks to sociolinguistics. In R. W. Shuy (ed.),
Georgetown University monograph series on languages and linguistics 25. 273—87.
Giglioli, P. P. (ed.) (1972). Language and social context. Harmondsworth: Penguin.
Greenberg, J. H. (1966). Language universals. The Hague: Mouton.
Gumperz, J. J. & Hymes, D. (eds) (1972). Directions in sociolinguistics: the ethnography of
communication. New York: Holt, Rinehart & Winston.
Halle, M. (1962). Phonology in generative grammar. Word 18. 54—72.
Halliday, M. A. K. (1973). Explorations in the function of language. London: Arnold.
Householder, F. W. (1971). Linguistic speculations. New York: Cambridge University
Press.
(1972). The principal step in linguistic change. Language sciences 20. 1-5.
Hymes, D. (ed.) (1964). Language in culture and society. New York: Harper & Row.
(ed.) (1971). Pidgiv.ization and creolization of languages. London: Cambridge Uni-
versity Press.
(1973a). The scope of sociolinguistics. In R. W. Shuy (ed.), Georgetown University
monograph series on languages and linguistics 25. 313-33.
(1973b). Speech and language: on the origins and foundations of inequality among
speakers. Daedalus Summer 1973. 59-85.
Katz, J. J. & Bever, T. G. (forthcoming). The fall and rise of empiricism.
Kiparsky, P. (1968). Linguistic universals and linguistic change. In E. Bach and R. T.
Harms (eds), Universals in linguistic theory. New York: Holt, Rinehard & Winston.
Kuhn, T. S. (1962). The structure of scientific revolutions. Chicago: The University of
Chicago Press.
Kuryiowicz, J. (1972). The role of deictic elements in linguistic evolution. Semiotica 5.
I74-83-
Laberge, S. (1972). Changement dans I'usage des pronoms inde'finis chez les Montrealais.
M.A. Thesis, University of Montreal.
Labov, W. (1966) The social stratification of English in New York City. Washington D.C.:
Center for Applied Linguistics.
106
 REVIEWS
(1972). Some principles of linguistic methodology. LinS. 1. 97-120.
(!973). The linguistic consequences of being a lame. LinS. 2. 81-115.
(forthcoming). On the use of the present to explain the past. Proceedings of the Xlth
international congress of linguists, Bologna, August-Sept. 1972.
Labov, W., Yaeger, M. & Steiner, R. (1972). A quantitative study of sound change in progress.
NSF Report, Contract NSF-GS-3287. University of Pennsylvania.
Lakoff, G. (forthcoming). Interview with Herman Parret. In H. Parret (ed.), Discussing
language. The Hague: Mouton.
Lakoff, R. (1972). Language in context. Lg. 48. 907—27.
Ohala, J. J. (forthcoming). Experimental historical phonology. In J. M. Anderson and
C. Jones (eds), Proceedings of the First international congress on historical linguistics,
Edinburgh, Sept. 1973.
Pride, J. B. (1973). Review of J. J. Gumperz and D. Hymes (eds), Directions in sociolin-
guistics. LinS. 2. 245-63.
Ross, J. R. (1972). The category squish: endstation Hauptwort. Papers from the eighth
regional meeting of the Chicago Linguistic Society. 316-28.
Salmon, V. (1965). Sentence structures in colloquial Shakespearian English. Transactions
of the Philological Society. 1965. 105-40.
Sankoff, G. (1973). Above and beyond phonology in variable rules. In C.-J. N. Bailey and
R. W. Shuy (eds), New ways of analyzing variation in English. Washington, D.C.:
Georgetown University Press. 44-61.
Sankoff, G. & Laberge, S. (1973). On the acquisition of native speakers by a language.
Kivung 6. 32—47 (Journal of the Linguistic Society of Papua).
Sapir, E. (1921). Language: an introduction to the study of speech. New York: Harcourt,
Brace & World.
Schneider, G. D. (1966). West African Pidgin-English: a descriptive linguistic analysis with
texts and glossary from the Camaroon Area. Athens, Ohio.
Stalnaker, R. C. (1972). Pragmatics. In D. Davidson and G. Harmon (eds), Semantics of
natural language. Dordrecht: Reidel. 380-97.
Stampe, D. (1969). The acquisition of phonetic representation. Papers from thefifthregional
meeting of the Chicago Linguistic Society. 443-54.
Sudnow, D. (ed.), (1972). Studies in social interaction. New York: Macmillan, The Free
Press.
Traugott, E. C. (forthcoming a). Language change, language acquisition, and the genesis
of spatio-temporal terms. In J. M. Anderson and C. Jones (eds), Proceedings of the First
international conference on historical linguistics, Edinburgh, Sept. 1973.
(forthcoming b). Changement linguistique; sa relation avec l'acquisition de la
langage maternelle. Langages.
Weinreich, U. (1953). Languages in contact. New York: Publications of the Linguistic
Society of America.
Weinreich, U., Labov, W. & Herzog, M. I. (1968). Empirical foundations for a theory of
language change. In W. P. Lehmann and Y. Malkiel (eds), Directions for historical
linguistics. Austin: University of Austin Press.
Whinnom, K. (1971)- Linguistic hybridization and the 'special case' of pidgins and Creoles.
In D. Hymes (ed.), Pidginization and creolization of languages. New York: Cambridge
University Press. 91-115.
Reviewed by ELIZABETH CLOSS TRAUGOTT
Committee on Linguistics and Department of English
(Received 16 March 1974) Stanford University, Stanford, California, USA

A. SCHWEIZER, Problems in the sociology of language in contemporary American

linguistics. Leningrad, 1971.
The scope of this book is indeed as restricted as the title suggests. Only four topics
107