Social Network and Social Class: Toward an Integrated Sociolinguistic Model
Author(s): Lesley Milroy and James Milroy
Source: Language in Society, Vol. 21, No. 1 (Mar., 1992), pp. 1-26
Published by: Cambridge University Press
Stable URL: http://www.jstor.org/stable/4168309
Accessed: 07-08-2017 21:17 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide

range of content in a trusted digital archive. We use information technology and tools to increase productivity and

facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

Terms and Conditions of Use

Cambridge University Press is collaborating with JSTOR to digitize, preserve and extend
access to Language in Society

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

Language in Society 21, I-26. Printed in the United States of America

Social network and social class:
Toward an integrated sociolinguistic model’

LESLEY MILROY AND JAMES MILROY

Departments of Speech (L.M.) and English Language (J.M.)

University of Newcastle upon Tyne
Queen Victoria Road at St. Thomas’ Street

Newcastle upon Tyne NEi 7RU, United Kingdom

ABSTRACT

In sociolinguistics, approaches that use the variables of socioeconomic

class and social network have often been thought to be irreconcilable.
In this article, we explore the connection between these variables and
suggest the outlines of a model that can integrate them in a coherent

way. This depends on linking a consensus-based microlevel of network

with a conflict-based macrolevel of social class. We suggest interpreta-

tions of certain sociolinguistic findings, citing detailed evidence from re-
search in Northern Ireland and Philadelphia, which emphasize the need
for acknowledging the importance of looseknit network ties in facilitat-
ing linguistic innovations. We then propose that the link between net-
work and class can be made via the notion of weak network ties using

the process-based model of the macrolevel suggested by Thomas
H0jrup’s theory of life-modes. (Sociolinguistics, sociology, quantitative
social dialectology, anthropological linguistics)

One of the most important contributions of Labov’s quantitative paradigm
has been to allow us to examine systematically and accountably the relation-

ship between language variation and speaker variables such as sex, ethnic-
ity, social network, and – most importantly perhaps – social class. Language
variation in large and linguistically heterogeneous cities as well as in smaller
communities has been revealed not as chaotic but as socially regular, and
Labov and others have shown how investigating this socially patterned vari-
ation can illuminate mechanisms of linguistic change. In this article, we fo-
cus on the variables of social class and social network, both of which have
appeared in some form in a large number sociolinguistic studies of variation
and change. Our principal interest lies not in the complex sociological issues
associated with class and network, some of which we discuss here, but in un-
derstanding the role of class and network in patterns of linguistic variation
and mechanisms of linguistic change.

? 1992 Cambridge University Press 0047-4045/92 $5.00 + .00

1

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

Social class is fundamentally a concept designed to elucidate large-scale so-
cial, political, and economic structures and processes, whereas social network
relates to the community and interpersonal level of social organization. Be-

ginning with Bott in 1958 (revised in 197I), a number of British anthropol-

ogists developed network-analytic procedures because they were dissatisfied
with what they saw as an overreliance on highly abstract social, political, and

economic frameworks in accounting for forms of behavior of individuals.

Personal social networks were generally seen as contextualized within this

broader framework, which was bracketed off to allow attention to be con-

centrated on developing less abstract modes of analysis capable of account-
ing for the variable behavior of individuals more immediately. However, it
is important to remember that such bracketing off is wholly methodologi-

cal and does not reflect an ontological reality; no one claims that personal
social network structure is independent of the broader social framework that
constantly constrains individual behavior. While acknowledging these con-
straints, a fundamental postulate of network analysts is that individuals cre-
ate personal communities that provide them with a meaningful framework
for solving the problems of their day-to-day existence (Mitchell 1986:74).

Our own work in Belfast has concentrated chiefly on detailed empirical
analysis of linguistic and social variation at this interpersonal and commu-
nity level, and in this article we want to propose a means of integrating re-
search at this level with research that relates language variation to social class.
Like the British sociologist Anthony Giddens, who insisted that “the study
of day to day life is integral to the reproduction of institutional practices”
(I984:282), we prefer to view the so-called micro- and macrolevels of anal-

ysis, to which network and class respectively may be thought to correspond,
as embodying complementary rather than conflicting perspectives.

A different question, however, is the adequacy of the conceptualization

of class that is current in much contemporary sociolinguistics, several schol-
ars having remarked that the social theory implicitly adopted by sociolin-
guists is in need of explicit formulation and critique. “Sociolinguists have
often borrowed social concepts in an ad hoc and unreflecting fashion, not
usually considering critically the implicit theoretical frameworks that are im-
ported wholesale along with such convenient constructs as three-, four- or
nine-sector scalings of socioeconomic status” (Woolard I985:738).

What Woolard is criticizing here is the procedure whereby a particular so-
cial class model is imported as an initial ad hoc means of organizing data,
not because of its theoretical suitability, but for the purely pragmatic reason
that it has been widely used in sociological surveys and so is readily opera-
tionalizable. Thus, although many impressively consistent patterns of vari-
ation have emerged from urban sociolinguistic work, an adequate social
framework within which to interpret their results is still lacking. In attempt-
ing now to develop such a framework, it seems best to start with the rich so-

2

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

SOCIAL NETWORK AND SOCIAL CLASS

ciolinguistic evidence that has been gathered over the last 25 years, only then
looking for a social theory that can account for it coherently. Working in this

order will allow principled decisions to be made about the kind of framework
required.

We do not claim yet to have found the ideal social class model; in this ar-

ticle, we do no more than try to integrate existing findings and suggest the
kind of model that seems to be required. A number of sociolinguists have
remarked that the conception of social class underlying Labov’s work in New

York City and Philadelphia is not particularly appropriate (Rickford 1986;
Sankoff, Cedergren, Kemp, Thibault, & Vincent I989). His key sociolinguis-
tic notion of speech community emphasizes shared norms of evaluation
throughout the community, where speakers are said to agree on the evalua-

tion of these very linguistic norms that symbolize the divisions between them.

This sociolinguistic model seems to reflect a consensus view of society of the
type associated with the sociologist Talcott Parsons, whereby the community
is envisaged as fundamentally cohesive and self-regulating. Yet, the vitality
and persistence of nonstandard vernacular communities uncovered by many
researchers (including Labov) are more readily interpretable as evidence of

conflict and sharp divisions in society than as evidence of consensus.
Although we certainly need to assume some kind of consensus to account

for data such as the cross-class agreement on the phonolexical rules for
raising and tensing of (a) in Philadelphia (Labov I98I), scholars such as

Rickford (I986), working on Guyanese creole, have concluded that conflict
models of social class have been unduly neglected by sociolinguists. Indeed,
support for a conflict model of society is provided by Labov’s own recent
work in Philadelphia, where he found progressive segregation and linguis-

tic differentiation between black and white networks (Labov & Harris I986).
Furthermore, a conflict model is essential if we are to account for the

phenomenon of linguistic change, with which some kind of social conflict is

generally associated. Labov himself has acknowledged that “a thorough-
going structural-functional approach to language could be applied only if lin-
guistic systems did not undergo internal change and development” (Labov
I986:283).

Although acknowledging that the question here is one of the relative
weight given to conflict and consensus perspectives, rather than an absolute

opposition between the two (cf. Giddens I989:705), we suggest that a social
class model based on conflict, division, and inequality best accounts for
many of the patterns of language variation uncovered by the detailed work
of sociolinguists, generally on phonological or morphological variables. The
Marxist notion of the linguistic market has been used in urban sociolinguis-
tics (see Sankoff et al., I989, for a recent example), the general contention
being that language represents a form of social and cultural capital that is
convertible into economic capital. Dittmar, Schlobinski, and Wachs (I988)

3

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

provided a particularly useful exposition of the linguistic market concept in
relation to their analysis of Berlin vernacular. However, Woolard (I985) sug-

gested that standard/vernacular opposition emerging from so much research
needs to be discussed in terms of alternative linguistic markets. This is con-

trary to Bourdieu’s (I977, I984) view of a single dominant linguistic market
where the rule of the legitimate language is merely suspended, its domina-
tion temporarily absent, when the vernacular is used.

Our own work as well as that of others supports Woolard’s analysis. Just
as there is strong institutional pressure to use varieties approximating to the
standard in formal situations, effective sanctions are in force in nonstandard
domains also. For example, in Belfast, New York City, and (no doubt) else-
where young men are ridiculed by their peers if they use middle-class forms.
Woolard suggested that much recent sociolinguistic work that has concen-
trated on competing social values using contrastive status/solidarity concepts
(or something similar) offers a particularly promising bridge between socio-
linguistic and social theory (see Brown & Levinson I987, for a discussion of
such work). A framework that emphasizes competing social values rather
than consensus offers a plausible interpretation of the mass of variable
linguistic and social detail from inner-city Belfast reported in Milroy and Mil-

roy (1978), L. Milroy (I987a), J. Milroy (I98I), and elsewhere. The phono-
logical structure of Belfast vernacular can be coherently described only if it
is analyzed as an internally consistent (but systematically variable) vernac-
ular, rather than an unsuccessful approximation to middle-class Belfast or
standard English varieties (for a discussion see J. Milroy 1992, Ch. 3). We
interpreted close-knit social networks as mechanisms enabling speakers to
maintain such vernacular codes, which themselves constitute an actively con-
structed, symbolic opposition to dominant, legitimized codes.

An analysis in these terms takes us part of the way, but it does not account
for wider social structures, and so it needs to be supplemented by an appro-
priate social class model. The success, persistence, and precise form of the
symbolic opposition enacted by small-scale networks will depend not upon
community-internal linguistic or interactional factors, but upon the relation
of the resisting group to the national economy and to like groups in other
cities or states (see Gal I988). The level of integration of any given group into
the wider society is likely to be inversely related to the extent to which it
maintains a distinctive vernacular. This is why the outcome in terms of lan-
guage survival or shift in Belfast may be different from that in Paris or Co-
penhagen; in Catalonia different from Gascony. It will be constrained by
variations in political, economic, and social structures that are specific to
these different localities. Furthermore, close study of networks and the lan-
guage pattems associated with them can give us some idea of the mechanisms
that give rise to correlations between language and class.

4

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

SOCIAL NETWORK AND SOCIAL CLASS

So far, we have tried to outline some general prerequisites for an integrated
and socially coherent sociolinguistic theory, constructed to take account of
well-established linguistically detailed findings of urban vernacular research.

In the following sections, we summarize the chief principles underlying a net-
work analysis of language variation, looking first at close-knit communities
and then at more loose-knit types of network structure of a kind generally
associated with mobile individuals. We argue that the structure and social
function of what might be described as both “strong” and “weak” network
types needs to be considered in order to integrate a network model with a so-
ciolinguistically plausible and socially adequate model of class.

NETWORK STRUCTURE IN CLOSE-KNIT COMMUNITIES

A social network may be seen as a boundless web of ties that reaches out
through a whole society, linking people to one another, however remotely.
But for practical reasons the analyst studies social networks as “anchored”
to individuals, and interest has most often focused on relatively strong first-
order network ties – that is, those persons with whom ego directly and reg-
ularly interacts. This principle of anchorage effectively limits the field of

study, generally to something between 30 and 5o individuals, although it is
assumed that second-order ties to whom ego is linked through others are also
influential (see Milroy i987a).

Two types of personal network characteristics are generally distinguished

by anthropologists: structural, which pertains to the shape and pattern of
the network, and interactional, which pertains to the content of the ties.
Both structural and interactional characteristics are important in constrain-
ing social action. Investigators from several disciplines who have developed
formal methods of analyzing the properties of networks have tended to con-
centrate on structural properties such as density, whereas social investigators –
who want to account for the observable behavior of individuals tend to give
equal weight to interactional features such as the multiplexity, history, du-
rability, frequency, and intensity of ties (see, e.g., Cochran, Larner, Riley,
Gunnarsson, & Henderson I99o; Surra I988). Some important structural and
interactional features are conveniently listed by Mitchell (I986).

Our analysis of the relationship between language variation and personal
network structure in three Belfast inner-city communities attempted to dem-
onstrate that a close-knit, territorially based network functions as a con-
servative force, resisting pressures for change originating from outside the
network. By close-knit we mean relatively dense and multiplex, these two
concepts being of critical importance in a comparative analysis of social net-
works. In a maximally dense and multiplex network, everyone would know

5

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

everyone else (density), and the actors would know one another in a range

of capacities (multiplexity). Close-knit networks, which vary in the extent to

which they approximate to an idealized maximally dense and multiplex net-

work, have the capacity to maintain and even enforce local conventions and

norms – including linguistic norms – and can provide a means of opposing

dominant institutional values and standardized linguistic norms. Their capac-
ity to do this, however, seems to be dependent on their territorial restriction

to specific neighborhoods, the day-to-day behavior of individuals being less

constrained by geographically dispersed networks. Network analysis thus of-

fers a basis for understanding the community-level mechanisms that under-
lie processes of language maintenance, and the persistence over centuries of

stigmatized linguistic forms and low-status vernaculars in the face of pow-

erful national policies of diffusing and imposing standard languages is indeed
remarkable.

Apart from its theoretical value, a network approach has been found use-

ful in providing a suitable methodology for studying ethnic or other sub-
groups in the population in situations where a social class model (particularly

one that focuses on consensus) is less practical. Quite apart from any theo-
retical problems, an initial approach in terms of class is difficult if subgroups
are distributed unequally with respect to class. A network approach is more
feasible with groups who are economically marginal, or powerless, or resident
in homogeneous and territorially well-defined neighborhoods. Moreover, a

strong sense of ethnicity or of local identity often creates and maintains lo-
calized cultural and linguistic norms and value systems that are presented and
perceived as sharply opposed to the mainstream values of outsiders. Ap-
proaching such communities initially in network rather than class terms can

allow the researcher to get a grip on the relation between linguistic variabil-
ity and social structure. Examples of sociolinguistic applications of network

analysis are: Schmidt (I985: Australian aboriginal adolescents), Bortoni-
Ricardo (I985: rural immigrants to a Brazilian city), Gal (1979: bilingual
peasant workers), Lippi-Green (I989: an Alpine rural community in Austria),
V. Edwards (I986: British black adolescents in the Midlands), and W. Ed-
wards (i99o: black Detroit speakers). Labov and his colleagues in their Phil-
adelphia neighborhood studies also used the network concept at the fieldwork
stage (Labov & Harris I986). So there is little disagreement on the practical
usefulness of a network-based methodology.

It has sometimes been suggested that close-knit types of community net-
work tend to be rural and that they are nowadays marginal to urban life. This

is suggested by, for example, the large sociological literature on “the stran-

ger,” the marginal individual who is often seen as typical of the modern city
dweller. Harman (I988) reviewed and evaluated much of this work. In the
same vein, Wirth, an influential member of the Chicago school of urban so-

6

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

SOCIAL NETWORK AND SOCIAL CLASS

ciologists, proposed that urban conditions give rise to impersonality and so-
cial distance (Wirth 1938). This fits in with the arguments we present shortly
on the role of weak ties in urban communities, but it does not tell the whole
story about urban life. For example, recent comparative research in Europe
and the United States has suggested that although the personal networks of
socially and geographically mobile persons are more geographically dispersed
and less kin-based than the traditional type of close-knit network, they are
also larger, more supportive and more affectively satisfying (Cochran et al.
i99o). Furthermore, as Fischer (I982) also emphasized, highly educated and
mobile individuals are more able to be selective in their choice of contacts
than the individual embedded in the localized solidary network, which can
be oppressive as well as supportive. Classic examples of such localized soli-
dary networks are the Italian American “urban villagers” described by Gans
(I962) or the close-knit Yorkshire mining communities described by Dennis,
Henriques, and Slaughter (1957).

These traditional close-knit (often indigenous) urban communities are less
salient in American and British cities than they once were, but they are ap-
parently being replaced by similar types of community created by newer
immigrants. More importantly perhaps, as Giddens (I989) pointed out,
neighborhoods involving close kinship and personal ties seem to be actually
created by city life, and Fischer (I984) suggested that whereas small towns
do not permit cultural diversity, cities do. For example, those who form part
of urban ethnic communities gravitate to form ties with, and often to live
with, others from a similar linguistic or ethnic background. These ethnic
groups seem to use the close-knit network as a means of protecting their in-
terests while the community develops the resources to integrate more fully
into urban life. For example, differences in the network structure of mem-
bers of the Chinese community in Newcastle upon Tyne correlate both with
different patterns of language choice and with different levels of integration
into non-Chinese domains of urban life (Milroy & Li I99I). Bortoni-Ricardo
(I985) made a similar point with regard to rural migrants to Brasilia. There-
fore, the type of close-knit community most easily conceptualized in network
terms is likely to be a product of modern city life rather than a residue of
an earlier type of social organization.

Such groups are important in providing a focal point for stigmatized ur-
ban vernaculars and other nonlegitimized linguistic norms, and so need to
be accounted for in any sociolinguistic theory. That is why some form of net-
work analysis that examines the relationship between the individual and the
primary group is so important. But the observable indicators of network
strength, a measure of integration into a close-knit group, will vary in kind
with community organization. For example, membership in a religious group
might be irrelevant in a contemporary northern English coal-mining commu-

7

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

nity, but highly relevant in an English Midlands black community (Edwards
I986).

The Philadelphia neighborhood studies are relevant to this point. Labov

and Harris (I986:21) suggested that although social network is useful as a
methodological tool, it has little explanatory value in itself and must be sup-

plemented at the interpretative stage by what they call the “social history”

of speakers: “the kinds of social experience they have had in dealing with
members of other groups, the way they have used language in their life.” At
many points in their article they appeared to represent social network and

social history as alternative and possibly contradictory modes of analysis,
with social history being a more powerful explanatory category than network.
But there is in fact no contradiction here. The wider social interactions (out-

side of the ingroup) that they describe as social history are themselves describ-
able and intepretable in terms of a network model. In that Labov and his

colleagues are examining the nature of the tie between group and individual,
they are carrying out no more and no less than a network analysis, employ-
ing indicators of network structure that relate to duration and content of tie

(cf. Surra I988; see also Cross I990, for an illuminating account of the re-
lationship between network structure and ethnicity). Still, they reject social
network at the interpretative phase of their work and use explanations that
fall back on a rather vague notion of “prestige,” which is related ultimately
to a primarily consensual concept of social class. However, much of the Phil-
adelphia data, as reported by Ash and Myhill (I986), is also open to an ex-
plicitly network-based interpretation. This interpretation has the advantage
of avoiding prior modeling of the behavior of individuals in terms of social
class, power, or dominance, and we comment further on this later. In gen-
eral, it seems preferable to carry out an initial analysis of small-scale com-
munities in terms of a notion such as network, which is designed for this level

of social organization, prior to working out an appropriate model of class
that relates to the macrolevel of analysis.

An important finding of the Philadelphia studies (Labov & Harris I986)
is that the speech of black and white ethnic groups is diverging in certain
ways, and it is this divergence that the investigators chose to emphasize. They
expressed this finding in terms of participation in linguistic changes: Certain
changes in progress in the white community are not in progress in the black
community. However, it can also be expressed in terms of the model we are
developing here, which would focus on the conflict between the two groups:
The two communities do not agree on norms of usage, and so in this respect
we can say that the sociolinguistic situation is one of conflicting norms rather

than consensus. It is also likely that the links between black and white com-
munities are on the whole relatively weak, in the sense that cross-ethnic net-
works tend not to be dense, multiplex, or territorially based (Cross i99o). It

8

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

SOCIAL NETWORK AND SOCIAL CLASS

is this notion of weak ties to which we now turn, before examining its ap-

plication to the Philadelphia communities.

STRONG AND WEAK NETWORK TIES

One important corollary to the link between language maintenance and a

close-knit network structure is that outside innovation and influence will be
associated with the weakening of such a structure. This accounts for our find-
ing in inner-city Belfast that speakers whose ties to the localized network are

weakest are those who approximate least closely to vernacular norms. Such

speakers are most exposed to external, often standardizing, pressures (Mil-

roy & Milroy I985). There is, however, a general methodological problem

associated with network analysis. Although it can be readily operationalized
to study speakers whose networks are of a relatively close-knit type, it can-
not so readily handle socially and geographically mobile speakers whose per-

sonal network ties are not predominantly dense or multiplex. So we cannot

easily demonstrate the effects of weak ties by the quantitative methods that
are used to demonstrate the effects of strong ties, as in inner-city Belfast, for

example. Fortunately, however, a large amount of linguistic evidence is avail-
able that enables us to follow the sociolinguistic implications of the line of
reasoning developed by Granovetter (1973, I982) in his examination of the
social function of weak network links. Granovetter argued that although
strong ties (of the sort associated with close-knit networks) facilitate local co-
hesion, they lead to overall social fragmentation. This seems to be the kind

of interclass and intercommunity fragmentation that we described in Belfast
and that Labov pointed to in Philadelphia with respect to black and white
groups. However, Granovetter argued that it is the (often numerous) weak
ties between relatively close-knit groups through which innovation and in-
fluence flow and that lead to an overall social cohesion capable of balanc-
ing the fragmentation and conflict associated with strong ties. It is important
here to keep in mind Granovetter’s insistence on the paradox that weak ties
can be described as strong in that they “provide links to a community beyond
the immediate social circle, information about education and employment
opportunities . . . and access to diverse ideas and perspectives” (Cochran
1990:289). Persons who contract mainly strong, localized, and often kin-
based ties are denied parallel access to these resources, and, as we have noted,
strong ties of this kind can be norm-enforcing and oppressive.

Following Granovetter’s closely argued article, we have proposed not only
that groups linked internally mainly by relatively weak ties are susceptible to
innovation, but also that innovations between groups are generally trans-
mitted by means of weak rather than strong network ties (e.g., through cas-
ual acquaintances rather than kin, close friends, or workmates). Weak ties

9

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

are, of course, likely to be much more numerous than strong ties. This ar-
gument runs somewhat counter to the general assumption that diffusion of
linguistic change is encouraged by relatively open channels of communica-

tion and discouraged by boundaries or weaknesses in lines of communica-

tion. However, as we argued in some detail (Milroy & Milroy I985), there

are many well-known patterns of change that are difficult to explain in this
apparently common-sense way. Some of these involve large-scale and long-

term changes over considerable distances, as discussed by Trudgill (1983,

I986). Examples are the spread of uvular [r] across national boundaries to
affect many northern European cities, the spread of certain London features
to Norwich, and the appearance of similar developments in unrelated or dis-

tantly related languages (e.g., preaspiration of voiceless stops in Icelandic and
Scottish Gaelic). It is very hard to see how the relevant populations in such

cases could be linked by strong ties. Other examples are at a more detailed

community level, such as the social configuration of the spread of [a] back-
ing from protestant East Belfast into the Clonard – a West Belfast Catholic

community. This spread, which we look at in a little detail, needs to be de-
scribed within a wider historical, social, and linguistic context.

We studied the variables (a) and (e) very intensively both in the inner-city
communities of Clonard, Hammer, and Ballymacarrett and in the slightly
higher status communities of Andersonstown and Braniel. We also estab-

lished a broad social class distribution by means of a doorstep survey car-
ried out on randomly sampled households in Belfast (J. Milroy I99I, 1992;
L. Milroy i987b:82). The vowel /a/ (as in man, grass) shows variation across
a wide phonetic continuum between long, backed, rounded realizations and
shorter front and front-raised realizations. The vowel /e/ (as in went, ques-
tion) varies between long, mid realizations and short, lower realizations (see
J. Milroy I98I, for an analysis of the phonological complexities). Realiza-
tions of both variables are strongly affected by sex, network structure, and
social class of the speaker. The extensive quantitative analysis reported in de-
tail elsewhere (most relevantly for this argument: J. Milroy 1992; L. Milroy

I987a; Milroy & Milroy I985) shows that raised, lengthened variants of /e/
are associated principally with women and middle-class speakers, and backed
variants of /a/ with men and working-class speakers.

It is clear from the historical and dialectological data presented by Patter-
son (i86o), Staples (I898), Williams (1903), and Gregg (1972) that /a/ back-
ing and /e/ raising are both relatively recent phenomena in Belfast but are

characteristic of modern Scots and originate in the Ulster-Scots-speaking di-
alect area of Down and Antrim (as distinct from the Mid- and West-Ulster
non-Scots hinterland). As Figure I shows, East Belfast adjoins the Ulster-
Scots region of North Down, whereas West Belfast points southwest down
the Lagan Valley, the speech of which is Mid-Ulster, with less Scots influ-
ence. Furthermore, immigration to West Belfast is recent and is largely from

10

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

SOCIAL NETWORK AND SOCIAL CLASS

Rathlin

Island

mBallymena -\_
Larne

Lurgan * LBELFAST

FIGURE I: Map showing the Ulster Scots area (shaded) in relation to East and
West Belfast.

a Mid- and West-Ulster hinterland. Present-day sociolinguistic evidence sug-
gests that the incoming variants of (e) and (a) are diffusing from east to west
of the city; scores for /a/ backing are higher for East Belfast working-class
men than for any other group studied, whereas East Belfast working-class
women use the low, conservative variants of (e) less than any other inner-city
group. The higher status Andersonstown and Braniel speakers exhibit a sim-
ilar pattern of sex differentiation but, as one might predict from the infor-
mation presented so far, use the incoming variants of (e) more frequently,
and the incoming variants of (a) less frequently, than inner-city speakers.

11

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

TABLE I. Contrasting patterns of distribution of two vowels

involved in change, according to social class and sex of
speaker, relative frequency of innovatory variants,

and level of correlation with network strength

High correlation

Change led by with network strength

(a) Males (working-class variant) Females

(e) Females (middle-class variant) Males

In summary (and this general distribution is confirmed by a doorstep sur-
vey), raised variants of (e) are in the inner-city associated particularly with
women (and, we might add, with careful speech styles). They are also asso-

ciated generally with slightly more prestigious outer-city speech. Incoming
variants of (a) show an almost perfectly converse pattern of social distribu-
tion, being associated with male, vernacular inner-city speech. Taking this
evidence together with the historical and geographical data outlined earlier,
we note that although incoming variants of both vowels appear to have orig-
inated in the same hinterland Scots dialect, each has assumed a diametrically
opposed social value in its relatively new urban setting.

The relationship between speaker choice of variant and individual network
structure adds a further complexity to this pattern, and it is the overall re-
lationship among social class, sex, and network structures of speaker that is
of particular relevance here. Extensive statistical analysis of the relationship
between language variation and social network has shown that whereas
choice of variant correlates with network structure among some inner-city
subgroups, these sociolinguistic patterns are quite different for each vowel.
Although (a) is generally sensitive to network structure, choice of variant is
more closely correlated with network structure for women than for men; this
is despite the fact that women (like middle-class speakers generally) use in-
coming backed variants much less frequently than men. The converse is true
of (e); whereas men use incoming raised variants much less than women, the
correlation between choice of variant and network structure is higher for
men. We argued on the basis of these data that (e) functions particularly
clearly for men and (a) for women as a network marker and noted that in
each case it is the group for whom the vowel has less significance as a net-
work marker that seems to be leading the linguistic change. The complex re-
lationship among class and sex of speaker, network structure, and language
use is summarized in Table i, and the data upon which this discussion is
based are reported in Milroy and Milroy (I985), L. Milroy (I987a), and J.
Milroy (1992).

12

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

SOCIAL NETWORK AND SOCIAL CLASS

We are now in a position to relate these patterns to the general argument

outlined earlier, namely that a close-knit network functions as a conserva-

tive force, resisting pressures for change originating from outside the net-

work. Those whose ties are weakest approximate least closely to vernacular

norms and are most exposed to external pressures for change. The analysis
presented here suggests that the vernacular speakers associated most strongly

with the innovation are in each case those for whom the vowel functions least

prominently as a network marker. It is as if a strong relationship between

the network structure of a given group and choice of phonetic realization of

a particular vowel disqualifies that group from fulfilling the role of innova-

tors with respect to that vowel. Conversely, the weakening of the language/

network relationship with respect to a group of speakers may be a necessary
precondition of that group fulfilling the role of linguistic innovators.

These observations provide further evidence to support the contention that

a weakening of network links is implicated in social processes of linguistic

change. Furthermore, some innovations seem to have crossed the sectarian

boundary in working-class Belfast to produce an intercommunity consensus

on norms among the generation of speakers who were most rigidly segregated
from each other. The problem of explaining how a linguistic change such as

(a) backing could possibly diffuse under such conditions dissolves if we ac-
cept Granovetter’s principle that it is the multiple weak ties of casual inter-
action (example for these speakers might be ties contracted in shops and
social security offices) through which innovation is routinely transmitted
rather than strong neighborhood ties of close association.

The Philadelphia data presented by Labov and Harris (I986) and Ash and

Myhill (I986) appear also to be amenable to an interpretation in terms of
weak ties, even though the researchers have preferred to account for the pat-

terns revealed by their analysis in terms of prestige and dominance. Labov
and Harris (I986:20-2I) mentioned the prestige of the localized innovator
and spoke of the dominant dialect as opposed to the dominated. This con-
trast of dominance was used by Ash and Myhill in interpreting their findings
with respect to four groups of speakers: a core white group, a core black
group, and two marginal groups – a group of blacks who have considerable
contact with-whites (henceforth WBs) and a group of whites who have con-
siderable contact with blacks (BWs). It is these marginal groups that inter-
est us here.

Ash and Myhill interpreted the pattern in Figure 2 as evidence that WBs
converge toward white morphosyntactic norms more markedly than BWs

converge toward black norms, and this is explained as the result of the dom-
inance of the white dialect. However, we do not need to invoke a macro no-
tion such as dominance in interpreting data at this community level. Two
patterns are particularly noticeable in the language of these two contact
groups (white-oriented blacks and black-oriented whites). First, on morpho-

13

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

100

~80 –

2 40

20

MA4 aaen 003

3rd sg copula poss ain’t/
-s -s didn’t

?Blacks who have little contat with whites (ndex of contact < 7) _ - - _ _ - - Blacks who have considerble conact with whites (Index of contact >8)
* > > > ” ” | | | “Whites who have caWiderable contact with blacs

………… ………White who have litle contact with blaks

FIGURE 2: Average percentage use of black English vernacular-marked mor-
phosyntactic variants by four groups of Philadelphia speakers
(adapted from Ash & Myhill I986:39).

syntactic variation their scores average about the same, and on copula de-
letion and ain’t for didn’t, the whites actually outperform the blacks on
“black” variants. Second, whereas the core black group uses these features
quite variably (presumably also using the “white” variants), the core white
group does not use the black variants at all. So in their convergence pattern,
the BWs and the WBs have different starting points. In ethnic group terms,
the white group starts much further back on the black English dimension
than the black group on the white dimension. And it follows from Figure 2
that the core black vernacular incorporates a resource not available to main-

stream white speakers – the capacity to alternate between black and white
morphosyntactic variants according to occasion of use. To this extent, these
speakers resemble the inner-city Belfast speakers, who also have at their dis-
posal alternating forms that carry different symbolic functions according to

14

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

SOCIAL NETWORK AND SOCIAL CLASS

occasion of use. Typically, one of these alternants has in-group functions and

belongs to the vernacular, whereas the other has out-group functions and is

more standardlike. We have discussed them in various places, most recently
in J. Milroy (i99i, 1992).

From this perspective, the convergence of WBs toward white norms is not

so remarkable – as these white norms are already available to them within
an existing pattern of core black vernacular variation. The reason why the
white-oriented blacks use the white norms more often than other blacks is

accessible through a theory of weak ties, as it is clear from the authors’ de-

scriptions of these speakers that their contacts with whites are of a classic
weak-tie type. They are described as con men, hustlers, and political activ-

ists, and it is hardly plausible that con men (for example) could successfully

practice on persons with whom they had contracted strong (dense and multi-

plex) ties. The degree to which these speakers use the white norms is increased
by the range and number of situations in which they have weak-tie contacts
outside their core community, and for them the adoption of more white us-
age is functional in their weak-tie contacts. The suggestion by Ash and My-
hill (I986:41) that prestige is the explanation for this shift toward white
norms seems to be quite a weak explanation, which simply begs the question
of what is meant by prestige in such a context (for a discussion of prestige-

based arguments, see J. Milroy I987).
The convergence of black-oriented white speakers to black norms is in a

sense more remarkable, as the core white dialect does not possess the new
variants (copula deletion, etc.) that they adopt (to a certain extent) in car-
rying out what must presumably be an act of linguistic accommodation.
These outside variants have to be acquired, and so some affirmatory effort
is involved. Although the researchers do not give precise information as to
the strength of these speakers’ participation in black culture, the model of
linguistic diffusion and change we have outlined in this section would pre-
dict that their ties with both communities are likely to be relatively weak. On
the basis of the information provided by Ash and Myhill, we assume that it
is this group, and not the WB group, who most resemble the peripheral char-
acters who Rogers and Shoemaker (197I) argued are typical of the innovat-
ing individual. Cross’s (i99o) comparison of the friendship networks of black
and white families offers further interesting perspectives on interethnic sit-
uations such as this.

It seems therefore that despite the fact that Labov and his colleagues ex-
plicitly rejected social network as an interpretative concept, some of their
findings are open to interpretations based on a network analysis that distin-
guishes between the social functions of strong and weak ties. Furthermore,
it also seems that this type of interpretation begs fewer questions than one
that appeals to assumptions about the wider social structure, as implied in
concepts such as prestige and dominance, and that we can best lay the

15

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

groundwork for an integrated theory by following through the implications
of the weak-tie model that we have outlined in this section.

THEORETICAL IMPLICATIONS OF A WEAK-TIES MODEL

Speakers whose ties to a localized network are weakest, who approximate
least closely to the norms of their local community, and who are most ex-
posed to external pressures are frequently found in the middle-class or up-
per-working-class areas of cities. However, in the previous section we alluded

to the practical difficulties in carrying out empirical investigations of loose-
knit network structures, which characterize residents of Andersonstown and
Braniel. Others have encountered the same problems, for example, in the
prosperous Berlin suburb of Zehlendorf (Labrie I988; see also L. Milroy

I987b: I98). But as many people (particularly city dwellers, as Wirth sug-
gested) contract weak ties, we need to take such ties into account in our de-
scription of sociolinguistic structure. And despite the empirical difficulties
in handling weak ties, an extension of network analysis that focuses on their
properties provides a crucial link with more abstract social theories of class.
It is clear that class-specific network structures are not arbitrarily constituted
but emerge from large-scale social and economic processes that themselves
give rise to (for example) the social and residential mobility associated with
loose-knit networks.

The relationship between the variables of class and network have been con-

sidered in some depth by Fischer (I982) in San Francisco and by Cochran
et al. (I990) in Germany, Sweden, Wales, and the United States. Investiga-
tors have generally emphasized the effects of education and affluence in af-
fording access to a socially and geographically wider range of contacts and
in enhancing the ability to maintain those contacts. Generally speaking, mid-
dle-class networks (consisting largely of weak ties in Granovetter’s sense) are
larger, less kin- and territory-oriented and perceived as more supportive.
Mewett (I982) examined the relationship between class and network from a
different perspective, arguing that class differences in small communities be-
gin to emerge over time as the proportion of multiplex relationships declines.
Observations such as this suggest a route for constructing a two-level socio-
linguistic theory, linking small-scale structures such as networks, in which in-
dividuals are embedded and act purposively in their daily lives, with larger
scale and more abstract social structures (classes) that determine relationships
of power at the institutional level.

From the point of view of the sociolinguist, it is smaller scale close-knit
networks that renew and maintain local systems of norms and values within
which discourse processes of the kind analyzed by Gumperz (I982) are un-
derstood and enacted. And it is network structures that link the interactional
level with the political and economic, where diverse local responses of lin-

16

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

SOCIAL NETWORK AND SOCIAL CLASS

guistic groups are constructed “to material and cultural domination” (Gal
I988). We need such a dual level of analysis if we are to understand the fre-

quently negative self-evaluations of speakers of urban vernaculars, who nev-

ertheless continue to use them in their daily lives. In this section, we use the

weak-tie concept to link systematically a network and a class-based analysis.

We have argued that weak ties between groups regularly provide bridges

through which information and influence flow and are more likely than

strong ties, which are by definition concentrated within groups, to fulfill this

function. Thus, whereas strong ties give rise to a local cohesion of the kind
described in inner-city Belfast, they lead, as we have noted, to overall frag-
mentation. Indeed, it is this potential for explaining both patterns – local sta-

bility and cohesion versus overall fragmentation and conflict – that allows
us to relate a network analysis to a model of social structure at the macro-
level. This is an important point, as some of the comments made in recent

years about network models in sociolinguistics by, for example, Labov (I986)

and Guy (I988) assume that their application is limited to strong ties in
close-knit communities; and indeed they have been used chiefly in such com-
munities (but see Bortoni-Ricardo I985). Guy’s remark that network is a mic-
rosociological concept, whereas class is macroscopic, seems reasonable if we
limit network analysis to close-knit networks. But an analysis that takes into
account the function of weak ties allows us not only to link the two levels
in a principled way, but to develop a clearer idea of which type of social class
model is appropriate. The analysis so far suggests an urban community that
consists of clusters of individuals connected internally by differing propor-
tions of weak and strong ties, which in turn are connected to other clusters
by predominantly weak ties. Middle-class groups will tend to be internally
connected with a higher proportion of weak ties than working-class groups.

This conclusion is entirely consistent with Labov’s finding that innovat-
ing groups are located centrally in the class structure, characterized by him
as upper-working or lower-middle class (Kroch 1978; Labov I980:254). For,
in British and American society at least, close-knit, territorially based, kin-
oriented networks are located most clearly in the lowest classes, but upper-
class networks are in some respects structurally similar, being relatively dense,
small, close-knit and kin-oriented. Consider Mills’s (1956) description of the
American power elite, and the close ties among British upper-class speakers
acquired at a limited number of private schools and universities and subse-
quently maintained for life. The majority of social and geographically mo-
bile speakers fall between these two points. Thus, if we extend a network
analysis to include an examination of loose-knit network types, which are
susceptible to outside (frequently standardizing) influences, it is evident that
network-based and class-based analyses are not contradictory as is sometimes
suggested; rather, they complement each other. Moreover, a network anal-
ysis can give us an idea of the interpersonal mechanisms giving rise to the ob-

17

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

servable language/class correlations that are such a prominent feature of
research in the quantitative paradigm.

AN INTEGRATED MODEL?

At this point, we have a picture of various ethnic and class groups as both

internally structured and connected to each other with varying proportions
and numbers of strong and weak ties. For example, ethnic sub-groups in Brit-
ain such as the black speakers studied by V. Edwards (I986) have a predom-
inantly strong-tie internal structure but seem to be linked by relatively few
weak ties to white working-class groups. These white groups in turn might
have a similar internal network structure but have more weak tie links with

other white working-class groups. Vertical links to middle-class groups might
be fewer (this seemed to be the case in Belfast) and moreover to be frequently
institutional to such persons as doctors, lawyers, teachers, welfare person-
nel, and the like. Middle-class groups for their part – professional, neighbor-

hood, and friendship groups – are characterized by a higher proportion of
weak ties internally than working-class groups; hence the problems of study-
ing them systematically in network terms in Zehlendorf and in outer-city Bel-
fast. But however we interpret the concept of class and however we model
these localized networks, Granovetter’s concept of the weak tie can be used
to link close-knit community level groupings to more abstract institutional
structures.

Such an analysis attributes the behavior of speakers to the constraining ef-
fects of the network or to the diminution of those effects that enables the le-
gitimized language to permeate networks, rather than by any direct effect of
prestige as defined by the perceived attributes of speakers who are seen to
“belong” to different status groups. Social class is not conceived of here as
a graded series of pigeonholes within which individuals may be placed. Fol-
lowing the analysis of the Danish Marxist anthropologist Thomas H0jrup
(I983), a view of social class more consistent with network analysis conceives
of it as a large-scale and ultimately economically driven process that splits
populations into subgroups. The groups sharing certain social and economic
characteristics and lifestyles that emerge from this split may loosely be de-
scribed as classes, but as we shall see H0jrup offered a more explicitly mo-
tivated description in terms of life-mode. The attraction of this analysis from
our point of view is that different types of network structure emerge from
the conditions associated with the life-modes of these subgroups, and local
and individual social behavior is seen as mediated through these smaller scale
structures rather than directly related to class.

Whereas network analysis of the kind we have outlined so far can delin-
eate various economic, political, and subcultural groupings in society, it can-
not say anything about the varying potentials of such networks to exercise

18

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

SOCIAL NETWORK AND SOCIAL CLASS

the economic and political power that is the source of conflict and inequal-

ity in society. In linguistic terms, this means that powerful networks have the

capacity to impose their linguistic and cultural norms on others, whereas

powerless ones do not but can merely use the resources of the network to

maintain and at best renew their own linguistic and cultural norms. There-

fore, to supplement network analysis we need a social theory such as

H0jrup’s, which can explicitly link a network analysis of subgroups within

society to an analysis of social structure at the political, institutional, and eco-
nomic levels.

H0jrup’s analysis is based on ethnographic work in Denmark and exten-

sive analysis of social and economic structure in other western European
countries. Although it begs as many questions as it offers solutions (see, e.g.,

Pedersen I99I, who questioned its applicability to women), it is particularly

suggestive in helping to construct a model of sociolinguistic structure that in-
tegrates the variables of social class and social network. With specific ref-
erence to western Europe, he proposed a division of populations into
subgroups that are described in terms of three life-modes. These life-modes
are seen as necessary and inevitable constituents of the social structure as a
whole. His conception of this larger social structure is Marxist, and the ini-

tial analysis is in terms of modes of production and consumption. Thus, cru-
cially, these subgroups are not seen as socially or culturally arbitrary but as
the effect of “fundamental societal structures which split the population into
fundamentally different life-modes” (H0jrup I983:47). Class is thus seen as
a dynamic process that gives rise to these life-modes. H0jrup’s analysis is par-
ticularly helpful in suggesting a further integration of the concepts of net-
work and class, because the different types of network structure that we
distinguished in the previous section can be seen to a considerable extent as
springing from differences in the life-modes of different individuals. Al-
though the argumentation supporting his analysis is lengthy and complex,
H0jrup used a limited number of straightforward concepts to distinguish the
three life-modes. Life-mode i is the life-mode of the self-employed, Life-

modes 2 and 3 of two different types of wage earner. Of critical importance
is the ideological orientation of the three groups to work, leisure, and fam-
ily. We focus a brief description of each of them on evident points of con-
tact with our network analysis.

Life-mode i

This is the life-mode of the self-employed, of which a close-knit family-
centered network with little distinction between work and leisure activities
and a strong solidarity ideology is particularly characteristic. The family-
owned business might be in agriculture or fishing, a corner shop, or a res-
taurant. In this life-mode, social relationships in the form of family ties or
cooperative relations among colleagues bind the producers into a cohesive

19

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

production unit. The primary concern is to keep the production rolling, and
all the family and other affiliated producers are involved in this. The pur-
pose of the enterprise is to be able to remain self-employed, a means that is
its own end. The concept of “free time” has little meaning in this life-mode,
because the producer is not put to work but puts himself or herself to work
to gain independence. Thus, the concepts of “leisure” and “work” have a to-

tally different meaning from that which they assume for wage earners, and
it is clear why a close-knit type of network structure and a solidarity ethic
will be associated with this life-mode, which itself follows from the type of
economic activity in which the producers engage. H0jrup did not see this
kind of life-mode as a relic of an earlier period (cf. the pervasiveness of close-
knit networks in modern cities noted earlier) but as highly efficient and com-
petitive, given its flexibility of operation and the commitment of the
producers.

Life-mode 2

Wage earners are different from Life-mode I commodity producers in that

they are incorporated in a long and complex process of production that they
do not own or control. Life-mode 2 is that of the ordinary wage earner, the
purpose of whose work according to H0jrup, is to provide an income that

enables a meaningful life during the worker’s free time. The family differs
from Life-mode I families in being separate from the wage earner’s work ac-
tivities and is the framework within which nonproductive leisure activity
takes place. The Life-mode 2 worker lacks the commitment to work char-
acteristic of Life-mode I, being prepared to sell his or her labor thereby be-
coming mobile and severing existing close-knit network ties where there is an
adequate inducement to do so. If wages are low however, the wage earner
has to demand enough to survive. Hence, the solidarity that arises among
workers who earn little – a solidarity reflected at the institutional level in the
establishment of trade unions. At a neighborhood level, this solidarity is em-
bodied in the close-knit networks of the traditional working-class society
of the kind investigated in Belfast. Following H0jrup’s analysis, we surmise
that the solidarity ethic would collapse and network ties become weaker if
economic and political conditions allowed workers to feel secure in their
future prospects, if they earned enough to become mobile, to buy better
houses and cars, to take holidays abroad, and so on. There do in fact ap-
pear to be differences of this kind in behavior between different groups of
wage earner, as we noted in our analysis of the outer-city versus inner-city
areas in Belfast. Moreover, as nation states vary with respect to the wages
and conditions offered to workers, close-knit networks will be associated with
Life-mode 2 workers in some countries more than in others. Cochran et al.
(I990) discussed in considerable detail in several chapters the effect of a

20
This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

SOCIAL NETWORK AND SOCIAL CLASS

whole range of political, economic, social, and cultural factors on social net-
work structure.

Lockwood’s (I989) classic investigation of class consciousness and images

of class structure in Britain fits in broadly with such an analysis, particularly
in its distinction between the outlooks described respectively as proletarian

traditionalism and privatized worker (see also Giddens I989:224). Privatized
workers, exemplified by the Luton car workers studied by Goldthorpe et al.
(I968-69), live apart from traditional working-class areas in the suburbs and

see work as a way of achieving a satisfactory lifestyle for themselves and their
families. They apparently reject the traditional working-class solidarity ethic,
but Lockwood stressed that a certain level of grievance tends to recreate this
ethic as does the us/them, insider/outsider imagery characteristic of close-
knit communities and of the traditional proletarian ideology. The persistence
and renewal of this imagery (and its associated network structures) seem to
spring fairly directly from changes in economic and power structures in
society.

Life-mode 3

Whereas the Life-mode 2 wage earner performs the routine tasks of the work
force at a given daily or hourly rate, the Life-mode 3 wage earner is a higher
professional or managerial employee with a high level of skill. This skill is
itself a saleable commodity, and the wage earner is paid to arrange, moni-
tor, and control the production process. Typically, the concept of work and
leisure and the role of the family are in sharp contrast to those of Life-
mode 2. This is because the Life-mode 3 goal is to rise up through the hier-

archy, obtaining control through managerial and professional roles so as to
exercise progressively more power and ultimately to escape from the control
of others so as to control resources and exercise power on one’s own account.
This process demands an immersion of the individual in work, a competi-
tive attitude to colleagues, and a blurring of the boundaries between work
and leisure. The family and its way of life fulfills a supportive role in rela-
tion to the career. Work therefore is life to a high degree, and the concept
of freedom is not one of free time but is associated with the work situation
and the career perspective.

Just as different types of network structure emerge from the economic con-
ditions associated with Life-modes i and 2, so a certain type of personal net-
work structure is likely to follow from Life-mode 3. These wage earners will
be socially and geographically mobile as they pursue their careers, forming
many loose ties, particularly of a professional kind, through which innova-
tions and influence may be transmitted. However, they will also form rela-
tively close-knit clusters and coalitions within their personal networks
through which they may control considerable resources. This seems to fit in

21

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

Maintenance of non-legitimized Dominance of legitimized
linguistic code linguistic code

Mainly strong Community-based ties Mainly weak

A\A/
INon-community-I
x based strong ies

Ufe-modes 1 2 3 (coaltions, power
elites etc.)

Macro level social, political and
I economic structure l

FIGURE 3: Macro- and microlevels of sociolinguistic structure.

with our general characterization of the differing role of loose-knit and close-

knit network ties. The primarily loose-knit network of the Life-mode 3 in-
dividual ensures that the dominant linguistic market – as embodied in some
form of legitimized or standard language – holds sway without hindrance
from (in Woolard’s terms) alternative vernacular markets. Figure 3 is a sche-
matic representation of the relation of social network structure to these three
life-modes.

It is important to emphasize that the concept of life-mode, like that of net-
work, is a structural one, in that the ideological and cultural characteristics
of a particular life-mode are determined by its contrast to the other life-
modes in the social formation. The interrelationships among the three life-
modes and the cultural practices associated with each one will therefore take
different forms in, for example, Denmark, Ireland, England, and Germany.
In each of these countries, the three fundamental modes of production that
the life-modes reflect “will appear in different variants and in different com-
binations of opposition and independence” (H0jrup I983:47). One conse-
quence of this chain of dependence running from political and socioeconomic
structures through life-modes to network structure and ultimately to socio-
linguistic structure (see Figure 3) is, as we have already suggested, that close-
knit networks will be associated with Life-mode 2 individuals in some nation
states more than in others. This seems to be the case if, for example, we com-

22

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

SOCIAL NETWORK AND SOCIAL CLASS

pare Belfast with Copenhagen. In Copenhagen, these wage earners are ap-
parently more mobile and prosperous and less inclined to live and work
together in close-knit groups of the kind described in Belfast (Gregersen &
Pedersen i99i). This in turn will give rise to sociolinguistic patterns that de-
pend on varying local contingencies and hence to urban vernaculars varying
in their degree of focusing and vitality.

CONCLUSION

The purpose of this article has been to work toward an integrated model of
sociolinguistic structure that links in an explicit way the social variables of
socioeconomic class and social network. Although these variables are often
presented as unrelated or even contradictory, we have tried to demonstrate
that, although they are at different orders of generality, it is useful to pro-
pose an interpretation of sociolinguistic space that conceives of them as
interrelated. They are, of course, related in reality also. Particular configu-
rations of network structure do not emerge accidentally for no particular rea-
son – the form they take is dependent on the large-scale social, political, and
economic structures that sociolinguists generally access in terms of socioeco-
nomic class. Thus, an attempt to integrate class and network as interpreta-
tive categories is well motivated.

However, the model of social class to which we have appealed here is not
the stratificational consensus-based model that has been generally favored
by Western sociolinguists. Whereas Labov’s view of speech community has
emphasized shared norms throughout the community and is thus related to
a consensus model of social class, we have preferred to emphasize the con-
flicts and inequalities in society that are symbolized by opposing linguistic
norms. This analysis emphasizes the basis of personal social networks in con-
sensus, whereas class differences involve not consensus but conflict. The
weak-tie model of Granovetter suggests a means of linking network and class,
as strong-tie situations predict agreement on norms, whereas weak-tie situ-
ations favor change and hence conflict of norms.

The analysis of higher level social structure that we have found most use-
ful here treats social and cultural divisions as emerging ultimately from the
economic inequalities produced by social class. H0jrup (I983) assumed such
a framework in his process-based model of life-mode, which we have used
in conjunction with network analysis to develop an integrated model of so-
ciolinguistic structure. By emphasizing in particular the importance of dis-
tinguishing between relatively strong and relatively weak network ties, we
have suggested how these economically determined life-modes give rise not
only to the social and cultural differences described by H0jrup, but to dif-
ferent kinds of network structure. This will further enable us to specify the

23

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

conditions in which the linguistic norms of the groups are likely to be focused
or diffuse, and the conditions in which they are open to, or resistant to,
change.

NOTE

I. Versions of this article were given at the Sociolinguistics Symposium at Roehampton in
April I99o and at the International Colloquium “Des Langues et des Villes,” organized by CLAD,
at L’Universite Cheik Anta Diop, Dakar, Senegal, in December I990. Our thanks to colleagues
who commented and gave advice at these meetings. Particular thanks to Padraig 6 Riagain, Sally
Boyd, and Dell Hymes for their useful comments; to Inge-Lise Pedersen, University of Copen-
hagen, who drew our attention to H0jrup’s work on life-modes; and to Marina Beale for as-
sistance with the background literature in sociology.

REFERENCES

Ash, S., & Myhill, J. (I986). Linguistic correlates of inter-ethnic contact. In Sankoff (I986).

33-44.
Bortoni-Ricardo, S. M. (I985). The urbanisation of rural dialect speakers: A sociolinguistic study

in Brazil. Cambridge: Cambridge University Press.
Bott, E. (197I). Family and social network (2nd ed.). London: Tavistock.
Bourdieu, P. (I977). The economics of linguistic exchanges. Social Science Information

I6(6):645-68.

(1984). Capital et marche linguistique. Linguistische Berichte 90:3-24.
Brown, P., & Levinson, S. (1987). Politeness. Cambridge: Cambridge University Press.
Cochran, M. (i99o). Environmental factors constraining network development. In M. Cochran

et al. (I990). 277-96.
Cochran, M., Larner, M., Riley, D., Gunnarsson, L., & Henderson, C. R. (eds.) (i99o). Ex-

tending families: The social networks of parents and their children. Cambridge: Cambridge
University Press.

Cohen, A. (ed.) (I982). Belonging. Manchester: Manchester University Press.
Cross, W. E. (I99o). Race and ethnicity: The effects on social networks. In M. Cochran et al.

(i0o). 67-87.
Dennis, N., Henriques, F. M., & Slaughter, C. (1957). Coal is our life. London: Eyre and

Spottiswoode.
Dittmar, N., & Schlobinski, P. (I988). The sociolinguistics of urban vernaculars. Berlin: de

Gruyter.
Dittmar, N., Schlobinski, P., & Wachs, I. (I988). The social significance of the Berlin urban

vernacular. In N. Dittmar & P. Schlobinski (I988). 19-43.
Edwards, V. (i986). Language in a black community. Clevedon: Multilingual Matters.
Edwards, W. (i99o). Social network theory and language variation in Detroit. Paper presented

at the Eighth Sociolinguistic Symposium, Roehampton, London.
Fischer, C. (i982). To dwell among friends: Personal networks in town and city. Chicago: Uni-

versity of Chicago Press.
(I984). The urban experience (2nd ed.). New York: Harcourt Brace Jovanovich.

Gal, S. (I979). Language shift: social determinants of linguistic change in bilingual Austria. New
York: Academic.

(I988). The political economy of code choice. In Heller (I988). 245-63.
Gans, H. J. (I962). The urban villagers: Group and class in the life of Italian-Americans (2nd

ed.). New York: Free Press.
Giddens, A. (I984). The constitution of society. Cambridge: Polity.

(I989). Sociology. Cambridge: Polity.

24

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

SOCIAL NETWORK AND SOCIAL CLASS

Goldthorpe, J. H. et al. (I968-69). The affluent worker in the class structure. Cambridge: Cam-
bridge University Press.

Granovetter, M. (I973). The strength of weak ties. American Journal of Sociology 78:1360-80.
(I982). The strength of weak ties: A network theory revisited. In P. V. Marsden & N.

Lin (eds.), Social structure and network analysis. London: Sage. I05-30.
Gregersen, F., & Pedersen, I. L. (i99i). The Copenhagen study in urban sociolinguistics. (Univer-

sitetsjubilaeets Danske Samfund, Serie A). Copenhagen: C. A. Reitzel.
Gregg, R. J. (1972). The Scotch-Irish dialect boundaries in Ulster. In M. F. Wakelin (ed.), Pat-

terns in the folk speech of the British Isles. London: Athlone Press. 109-39.
Gumperz, J. (I982). Discourse strategies. Cambridge: Cambridge University Press.
Guy, G. R. (I988). Language and social class. In Newmeyer (I988). 37-63.
Harman, L. D. (I988). The modern stranger: On language and membership. Berlin: Mouton

de Gruyter.
Heller, M. (I988). Code-switching. Berlin: Mouton de Gruyter.
H0jrup, T. (I983). The concept of life-mode: A form-specifying mode of analysis applied to

contemporary western Europe. Ethnologia Scandinavica I-50.
Kroch, A. S. (I978). Toward a theory of social dialect variation. Language in Society 7:17-36.
Labov, W. (1972). Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press.

(ed.) (I980). Locating language in time and space. New York: Academic.
(I98I). Resolving the neogrammarian controversy. Language 57:267-309.
(I986). Language structure and social structure. In S. Lindenberg et al. (eds.), Ap-

proaches to social theory. New York: Russell Sage.
Labov, W., & Harris, W. (i986). De facto segregation of black and white vemaculars. In Sankoff

(I986). 1-24.
Labrie, N. (I988). Comments on Berlin urban vernacular studies. In Dittmar & Schlobinski

(I988). 191-206.
Lippi-Green, R. L. (I989). Social network integration and language change in progress in a rural

alpine village. Language in Society I8:2I3-34.
Lockwood, D. (I989). The black-coated worker: A study in class-consciousness. Oxford: Claren-

don. (Original work published i966)
Mewett, P. (I982). Associational categories and the social location of relationships in a Lewis

crofting community. In Cohen (I982). IOI-30.
Mills, C. W. (I956). The power elite. Oxford: Oxford University Press.
Milroy, J. (I98I). Regional accents of English: Belfast. Belfast: Blackstaff.

(I987). The concept of prestige in sociolinguistic argumentation. York Papers in Lin-
guistics I3:2I5-26.

(I99I). The interpretation of social constraints on variation in Belfast English. In J.
Cheshire (ed.), English around the world: Sociolinguistic perspectives. Cambridge: Cambridge
University Press. 75-85.

(I992). Linguistic variation and change. Oxford: Blackwell.
Milroy, J., & Milroy, L. (I978). Belfast: Change and Variation in an urban vernacular. In P.

Trudgill (ed.), Sociolinguistic patterns in British English. London: Arnold. 19-36.
(I985). Linguistic change, social network, and speaker innovation. Journal of Linguis-

tics 21:339-84.
Milroy, L. (I987a). Language and social networks (2nd ed.). Oxford: Blackwell.

(I987b). Observing and analysing natural language. Oxford: Blackwell.
Milroy, L., & Li, W. (I99I). A social network perspective on code-switching and language choice:

The example of the Tyneside community. In Papers for the Symposium on Code-switching
in Bilingual Studies: Theory, significance and perspectives. (Vol. i). Strasbourg: European
Science Foundation.

Mitchell, J. C. (i986). Network procedures. In D. Frick (ed.), The quality of urban life. Ber-
lin: de Gruyter. 73-92.

Newmeyer, F. (ed.). (i988). Linguistics: The Cambridge survey. Vol. 4. Cambridge: Cambridge
University Press.

Patterson, D. (i860). Provincialisms of Belfast. Belfast: Mayne Boyd.
Pedersen, I.-L. (I99I). Sociolinguistic classification in a gender perspective. Unpublished manu-

script. University of Copenhagen, Institut for dansk dialektologi.

25

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

LESLEY MILROY AND JAMES MILROY

Rickford, J. (I986). The need for new approaches to social class analysis in linguistics. Lan-
guage and Communication 6(3):215-2 1.

Rogers, E. M., & Shoemaker, F. F. (i97I). Communication of innovations (2nd ed.). New York:
Academic Press.

Sankoff, D. (I986). Diversity and diachrony. Amsterdam: John Benjamins.
Sankoff, D., Cedergren, H., Kemp, W., Thibault, P., & Vincent, D. (I989). Montreal French:

Language, class, and ideology. In W. Fasold & D. Schiffrin (eds.), Language change and vari-
ation. Amsterdam: John Benjamins. I07-I8.

Schmidt, A. (I985). Young people’s Djirbal. Cambridge: Cambridge University Press.
Staples, J.H. (I898). Notes on Ulster English dialect. Transactions of the Philological Society

357-87.
Surra, C. A. (I988). The influence of the interactive network on developing relationships. In

R. M. Milardo (ed.), Family and social networks. Newbury, CA: Sage. 48-82.
Trudgill, P. (I983). On dialect. Oxford: Blackwell.

(I986). Dialects in contact. Oxford: Blackwell.
Williams, R. A. (I903). Remarks on Northern Irish pronunciation of English. Modern English

Quarterly 6:129-35.
Wirth, L. (1938). Urbanism as a way of life. American Journal of Sociology 44(I): 1-24.
Woolard, K. (I985). Language variation and cultural hegemony: Toward an integration of lin-

guistic and sociolinguistic theory. American Ethnologist 12:738-48.

26

This content downloaded from 130.126.162.126 on Mon, 07 Aug 2017 21:17:23 UTC
All use subject to http://about.jstor.org/terms

• Contents

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

• Issue Table of Contents

Language in Society, Vol. 21, No. 1 (Mar., 1992), pp. 1-188
Volume Information
Front Matter
Social Network and Social Class: Toward an Integrated Sociolinguistic Model [pp. 1-26]
Complimenting and Involvement in Peer Reviews: Gender Variation [pp. 27-57]
Women in Charge: Politeness and Directives in the Speech of Japanese Women [pp. 59-82]
Naive Linguistic Explanation [pp. 83-91]
Sociolinguistic Behavior in a Detroit Inner-City Black Neighborhood [pp. 93-115]
Reviews
Directions in Sociolinguistics
Review: untitled [pp. 117-123]
Discourse
Review: untitled [pp. 123-128]
Review: untitled [pp. 128-132]
Review: untitled [pp. 132-136]
Review: untitled [pp. 136-142]
Language Descriptions
Review: untitled [pp. 142-145]
Language Origin
Review: untitled [pp. 145-147]
Language Varieties and Situations
Review: untitled [pp. 147-152]
Review: untitled [pp. 152-154]
Review: untitled [pp. 154-157]
Review: untitled [pp. 158-162]
Review: untitled [pp. 162-164]
Review: untitled [pp. 164-167]
Review: untitled [pp. 167-169]
Review: untitled [pp. 169-173]
Festschrifts
Review: untitled [pp. 173-175]

Brief Notices
Newsletter on Interlinguistics [pp. 177-178]
Corrections: Publications Received [p. 178]
Review: Corrections: The Sociolinguistics of Urban Vernaculars: Case Studies and Their Evaluation [p. 178]
Review: Corrections: Edward Sapir: Linguist, Anthropologist, Humanist [p. 178]
Review: untitled [pp. 178-179]
Review: untitled [p. 179]
Review: untitled [pp. 179-180]
Review: untitled [p. 180]
Review: untitled [pp. 180-181]
Publications Received [pp. 181-188]
Back Matter

“Why be normal?”: Language and identity practices
in a community of nerd girls

M A R Y B U C H O L T Z

Department of English
Texas A&M University

College Station, TX 77843–4227
bucholtz@tamu.edu

A B S T R A C T

The introduction of practice theory into sociolinguistics is an important re-
cent development in the field. The community of practice provides a useful
alternative to the speech-community model, which has limitations for lan-
guage and gender researchers in particular. As an ethnographic, activity-
based approach, the community of practice is of special value to researchers
in language and gender because of its compatibility with current theories of
identity. An extension of the community of practice allows identities to be
explained as the result of positive and negative identity practices rather than
as fixed social categories, as in the speech-community model. The frame-
work is used here to analyze the linguistic practices associated with an un-
examined social identity, the nerd, and to illustrate how members of a local
community of female nerds at a US high school negotiate gender and other
aspects of their identities through practice. (Community of practice, gender,
discourse analysis, identity, social construction, social practice, speech com-
munity, adolescents, nerds)*

In sociolinguistics, social theory is rooted in the concept of the speech commu-
nity. As a language-based unit of social analysis, the speech community has al-
lowed sociolinguists to demonstrate that many linguistic phenomena previously
relegated to the realm of free variation are in fact socially structured. Thus Labov
1966 showed that the linguistic heterogeneity of New York City can be quanti-
tatively analyzed as the patterning of a single speech community, despite differ-
ences in New Yorkers’ language use based on sociological variables such as age,
social class, and gender.

Nonetheless, because the concept of speech community is indigenous to so-
ciolinguistics, it is not connected to any larger social theory. This theoretical
isolation, along with the fact that the speech community defines the social world
in strictly (socio)linguistic terms, has meant that sociolinguistic theory has largely
stood apart from theoretical advances in related disciplines. Meanwhile, within
sociolinguistics, the concept of the speech community has been hotly contested

Language in Society28, 203–223. Printed in the United States of America

© 1999 Cambridge University Press 0047-4045/99 $9.50 203

cola
Highlight

cola
Highlight

cola
Highlight

and continually revised as researchers have uncovered the limitations of previous
definitions.

The speech community presents special difficulties for researchers in the so-
ciolinguistic subfield of language and gender. The disciplinary autonomy of theory
based on the speech community is unproblematic for traditional sociolinguistic
research, which uses social information to account for linguistic phenomena such
as sound change. But when sociolinguists reverse the direction of analysis – ask-
ing instead how linguistic data can illuminate the social world, as language and
gender researchers seek to do – then connections to social theory beyond linguis-
tics become imperative. Moreover, the speech community model, which was de-
signed to analyze sociolinguistic phenomena at a macro level, is often inappropriate
and inadequate for the kinds of questions currently being asked in language and
gender scholarship. Central among these is the question of identity: How do speak-
ers use language to project their identities as gendered beings? And how are
gender identities interwoven with other social parameters?

This article draws on a theory of community and identity that avoids the prob-
lems associated with the speech community model. The new framework, the
community of practice, emerges frompractice theory, an approach that has
currency in such disciplines as sociology, anthropology, and education. The con-
nections of the community of practice to these recent developments in other fields
allow sociolinguists to offer more fully theorized social explanations than were
possible with the earlier model. In addition, the community of practice over-
comes many of the faults that sociolinguists have found with the speech commu-
nity, and it therefore has wide applicability to the field’s central questions. The
theory’s broad range of use is especially evident in language and gender studies –
because, unlike the speech community, the community of practice was intro-
duced into sociolinguistics specifically to address issues of gender.

In this article, I build on the theory of the community of practice to develop its
potential as an analytic tool for the sociolinguistic investigation of gendered iden-
tities. The framework is applied to a social identity, that of the nerd, which has
remained out of bounds in traditional sociolinguistic research based on the speech
community. This identity is analyzed within the community of practice frame-
work because only this concept permits us to draw on the linguistic and social
information necessary to understand the production of nerd identity. I argue that
nerd identity, contrary to popular perceptions, is not a stigma imposed by others,
but a purposefully chosen alternative to mainstream gender identities which is
achieved and maintained through language and other social practices.

L A N G U A G E A N D P R A C T I C E T H E O R Y

The idea that the social world is best viewed as a set of practices is not new.
Praxis is a foundational concept of Marxism, and more recently Giddens 1979

M A R Y B U C H O L T Z

204 Language in Society28:2 (1999)

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

has offered a practice-based account as a way out of the impasse created by
social structure, on the one hand, and personal agency, on the other. Given the
focus of practice theory on enduring social activity, it was perhaps inevitable
that it should soon come to view language as a central object of social analysis.
Outside linguistics, this perspective has been most fully articulated by the French
sociologists Pierre Bourdieu (1978, 1991) and Michel de Certeau (1984). Both
Bourdieu and Certeau understand language in relation to other social practices,
and both scholars view language as a social phenomenon, rather than merely as
an abstract formal system. As a consequence, they explicitly align their work
with the sociolinguistic enterprise, broadly conceived; the litany of familiar
names they invoke includes Joshua Fishman, Erving Goffman, William Labov,
and Emanuel Schegloff.

For Bourdieu, the starting point of practice ishabitus, the set of dispositions
to act (e.g. speak, walk, read, or eat) in particular ways which are inculcated in each
individual through implicit and explicit socialization. These dispositions are linked
to particular social dimensions such as class and gender. Habitus is also tied to the
body viahexis, the individual’s habitual and socially meaningful embodied stances
and gestures, and through other aspects of physical self-presentation. Language is
merely one practice in which habitus is embedded, and through which the indi-
vidual becomes socially locatable to observers. Thus non-linguistic social prac-
tices and language should be approached in analogous ways. As Bourdieu observes
(1991:89),

Not only are linguistic features never clearly separated from the speaker’s whole
set of social properties (bodily hexis, physiognomy, cosmetics, clothing), but
phonological (or lexical, or any other) features are never clearly separated
from other levels of language; and the judgement which classifies a speech
form as “popular” or a person as “vulgar” is based, like all practical predica-
tion, on sets of indices which never impinge on consciousness in that form.

Bourdieu here offers two important methodological insights to sociolinguists:
first, that non-linguistic practices may carry important linguistic information (and
vice versa); and second, that a complete sociolinguistic analysis must examine
multiple levels of language simultaneously. Yet, as a theorist, Bourdieu is less
useful to sociolinguists, and especially to language and gender scholars. His in-
sistence on the unconsciousness of practice reflects a general attenuation of agency
within his theory. Although speakers are not bound by their habitus, which is
inflected by the particular context in which it occurs, the tendency is to act in
accordance with what has been naturalized as appropriate. Bourdieu sees the
individual, then, more as a product of social structure than as a free agent. Prac-
tice at the local level – especially linguistic practice, which is embedded in the
class habitus of the standard and the non-standard – is primarily in the business of
reproducing existing social arrangements.

“ W H Y B E N O R M A L ? ” : L A N G U A G E A N D I D E N T I T Y P R A C T I C E S

Language in Society28:2 (1999) 205

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

For Certeau, by contrast, the individual is much more agentive, because the
focus of investigation is subversion as well as reproduction of the social order.
But like Bourdieu, Certeau finds inspiration for his project in linguistics:

Our investigation . . . can use as its theoretical model theconstruction of
individual sentences with anestablished vocabulary and syntax. In linguis-
tics, “performance” and “competence” are different: the act of speaking (with
all the enunciative strategies that implies) is not reducible to a knowledge of
the language. By adopting the point of view of enunciation – which is the
subject of our study – we privilege the act of speaking; according to that point
of view, speaking operates within the field of a linguistic system; it effects an
appropriation, or reappropriation, of language by its speakers; it establishes a
present relative to a time and place; and it posits acontract with the
other (the interlocutor) in a network of places and relations. These four char-
acteristics of the speech act can be found in many other practices (walking,
cooking, etc.). (1984:xiii; original emphasis)

Certeau here makes the link between language and other social practices even
more explicit than did Bourdieu before him. Certeau sees all social practices,
both linguistic and non-linguistic, as similar in their social effects. But where
Bourdieu considers practice to be a reproduction of social structure, Certeau views
it as an appropriation, an act of agency. The point, then, is to understand how
culturally shared resources (such as language) are made to serve the specific
social needs of individuals. These needs may enforce the social status quo, but
they may just as easily challenge or revise it.

A third theory of practice has been developed within anthropology by Ortner
1996, who criticizes earlier scholarship on the grounds that it fails to take seri-
ously the practices of women. Making the female agent central in the project of
practice theory, Ortner constructs a framework that has room for both structure
and agency. Although language is not a guiding concept in Ortner’s work as it is
for Bourdieu and Certeau, she views structure itself as textual in nature – the
“field of a linguistic system”, in Certeau’s words – within which an individual act
of speaking operates. Thus a complete analysis of gender, and especially of lan-
guage and gender, cannot focus on texts alone. As Ortner argues (1996:2),

Studies of the ways in which some set of “texts” – media productions, literary
creations, medical writings, religious discourses, and so on – “constructs” cat-
egories, identities, or subject positions, are incomplete and misleading unless
they ask to what degree those texts successfully impose themselves on real
people (and which people) in real time. Similarly, studies of the ways in which
people resist, negotiate, or appropriate some feature of their world are also
inadequate and misleading without careful analysis of the cultural meanings
and structural arrangements that construct and constrain their “agency”, and
that limit the transformative potential of all such intentionalized activity.

M A R Y B U C H O L T Z

206 Language in Society28:2 (1999)

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

The possibility – and the reality – of such unified analyses within language and
gender studies is offered by the community of practice framework. More than any
previous approach in sociolinguistics, the community of practice allows research-
ers to examine, in a theoretically adequate way, both the actions of individuals
and the structures that are thereby produced and reproduced, resisted and subverted.

G E N D E R , T H E S P E E C H C O M M U N I T Y ,

A N D T H E C O M M U N I T Y O F P R A C T I C E

Ortner’s introduction of a feminist perspective was a relatively late development
in practice theory in anthropology and sociology. Likewise, the theory of the
community of practice, which emerged from education (Lave 1988, Lave &
Wenger 1991, Wenger 1998), was not applied to gender until it was imported into
linguistics by Eckert & McConnell-Ginet in a highly influential survey article
(1992). As an alternative to the speech community – a central analytic tool of
sociolinguistics – the community of practice requires language and gender schol-
ars to rethink traditional notions of community, identity, and gender. However,
Eckert & McConnell-Ginet do not offer an explicit critique of the speech com-
munity; although that concept has been widely debated (see Hudson 1980, Wil-
liams 1992), its particular limitations for language and gender research have not
been systematically addressed. I suggest six ways in which the speech commu-
nity has been an inadequate model for work on language and gender:

(a) Its tendency to take language as central.
(b) Its emphasis on consensus as the organizing principle of community.
(c) Its preference for studying central members of the community over those

at the margins.
(d) Its focus on the group at the expense of individuals.
(e) Its view of identity as a set of static categories.
(f ) Its valorization of researchers’ interpretations over participants’ own un-

derstandings of their practices.

Language vs. social practice

The speech community has been defined in many ways, but every definition
posits language as a primary criterion of community. What is taken as shared may
be the linguistic system (Bloomfield 1933:42–56); or shared linguistic norms
(Labov 1972, Guy 1988); the pattern of variation (Milroy 1992); or only a set of
sociolinguistic norms (Romaine 1982). The emphasis may be less on the linguis-
tic system, and more on shared interactional settings and norms (Hymes 1974,
Dorian 1982, Silverstein 1996). But in every case, the focus remains on language.
Even many scholars who advocate a more interactional approach understandin-
teraction to be a preeminently linguistic concept. Other forms of mutual en-
gagement – that is, all non-linguistic aspects of social activity – are marginalized
or ignored.

“ W H Y B E N O R M A L ? ” : L A N G U A G E A N D I D E N T I T Y P R A C T I C E S

Language in Society28:2 (1999) 207

cola
Highlight

cola
Highlight

cola
Highlight

By recognizing practice – the social projects of participants – as the motivat-
ing context for linguistic interaction, the theory of the community of practice
makes activity much more central to sociolinguistic analysis. Just as importantly,
whereas the speech community model understands language as fundamentally
disembodied – as detachable from the physicality of speakers – the community of
practice quite literally reincorporates language into the physical self. In this re-
gard, it echoes Bourdieu’s concept of hexis – a crucial connection for feminist
researchers, for whom the specificity of the gendered body is a theoretical start-
ing point.

Consensus vs. conflict

Another aspect of the traditional model that has received a great deal of criticism
is the idea that the speech community is constituted around shared sociolinguistic
norms. This definition was first proposed by Labov 1972, and was taken up by
many subsequent researchers. The postulate that speakers agree on and uphold
certain linguistic forms as normative, regardless of differences in social back-
ground, assumes a consensus model of society that is at odds with a long-standing
tradition of social theory. Several sociolinguists have critiqued the Labovian def-
inition of the speech community on these grounds (e.g. Rickford 1986, Milroy
1992). Moreover, the invocation of “norms” obscures the fact that these are suc-
cessfully imposed ideologies favoring the interests of the powerful (Bourdieu
1991). This arrangement has long been recognized by scholars of language and
gender who have worked to combat views of women’s language as deficient in
comparison to men’s (see Cameron 1992:42 ff.)

Central vs. marginal members

The language of norms also presumes that some members of the speech commu-
nity are central and others are marginal, and that it is the central members who are
of interest. To be sure, the structured heterogeneity of the speech community
improves on earlier models by recognizing the existence and systematicity of
heterogeneity; however, speakers who do not share the same norms (for example,
because they are recent immigrants or transplants from other regions) are ex-
cluded from the community.1 Thus, despite the model’s emphasis on heteroge-
neity, the focus is in fact on what speakers share. Marginal members rarely enter
the analysis, and when they do, they remain at the margins; their linguistic prac-
tices are used primarily to demonstrate how they fall short of central member-
ship. Language and gender researchers are acutely aware of the problems with
this approach. Because women may be defined, implicitly or explicitly, as mar-
ginal to the vernacular speech community, they may be underrepresented or sim-
ply misrepresented (cf. Morgan 1999). Speakers whose identities differ from those
of the wider community – especially those whose gender identities do not con-
form to community norms – are likewise omitted or obscured in research within
this paradigm.

M A R Y B U C H O L T Z

208 Language in Society28:2 (1999)

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

The expectation of consensus in speech community norms also requires that
the system be closed to outside influence. The possibility of interaction between
speech communities is not important in the model.2 Hence researchers seek same-
ness, not difference; difference (e.g. in language use) is contained by interpreting
it as sameness at an underlying level (e.g. in shared sociolinguistic norms). With
this emphasis on analysis of the group as an autonomous system, phenomena
resulting from linguistic and cultural contact (Pratt 1987) may be overlooked.
For example, the focus on the internal workings of the speech community does
not accommodate investigations of gendered interaction across cultural groups.

Groups vs. individuals

Related to the problem of homogeneity in the speech community model is its
privileging of the group over the individual as the unit of analysis. In such an
approach, the role of the individual is merely to instantiate the practices of the
group. Individual actions result less from choice and agency than from a social
order that impinges on individuals from above. The traditional model’s strong
preference for structure over agency means that individual variation, or style, is
interpreted as the mechanical outcome of structural forces such as situational
norms. A more agentive view locates style in personal choices concerning self-
presentation (Johnstone 1995, 1996, Johnstone & Bean 1997). This perspective,
which also admits structural constraints on the individual, is well suited for gen-
der studies, given the field’s longtime recognition that individuals make purpose-
ful choices in the face of the limitations imposed on them by social structures. As
Ortner notes above, one of the benefits of practice theory is its ability to cope with
both aspects of women’s (and men’s) lives.

Identity categories vs. identity practices

The structural perspective is a static perspective, one in which the social order
remains largely unaltered. Changes in the practices of its inhabitants have the
effect of keeping the system in equilibrium. Nowhere is this more evident than in
the speech community model’s implicit theory of identity: Individuals are viewed
as occupying particular social identities throughout their lives by virtue of their
position in the social structure. Such an analysis is particularly problematic for
researchers of language and gender. The concept of identity is central to gender-
oriented research, but the version offered by the speech community framework
contradicts basic insights of recent feminist theory. Contemporary feminists view
identities as fluid, not frozen; they note that, although identities link individuals
to particular social groups, such links are not predetermined. Instead, identities
emerge in practice, through the combined effects of structure and agency. Indi-
viduals engage in multiple identity practices simultaneously, and they are able to
move from one identity to another. This process is not entirely unconstrained;
speakers may end up reproducing hegemonic identities more often than resisting

“ W H Y B E N O R M A L ? ” : L A N G U A G E A N D I D E N T I T Y P R A C T I C E S

Language in Society28:2 (1999) 209

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

them, as suggested by Holmes 1997. It is also important, however, to call atten-
tion to the previously unacknowledged flexibility of identity formation.

Top-down vs. bottom-up

For the specificity of identity to become visible, it must be examined from the
point of view of the individuals who enact it. Such a vantage point is not avail-
able within the speech community model, which privileges the analyst’s inter-
pretations over those of participants. Indeed, the speech community itself is an
analytic construct which may fail to correspond to its putative members’ own
perceptions. Nonetheless, many analyses are carried out under the belief that
the linguist has access to elements of speakers’ reality that are not available to
the speakers themselves.

An alternative to this top-down paradigm isethnography, an approach that
is participant- rather than analyst-driven. Where the speech community frame-
work is skeptical of speakers’ perspectives on their own practices, ethnography
makes local interpretations central to the analysis. Gender does not have the same
meanings across space and time, but is instead a local production, realized dif-
ferently by different members of a community; thus an ethnographic orientation
yields particularly fruitful results for language and gender research.

N E R D S , G E N D E R , A N D T H E C O M M U N I T Y O F P R A C T I C E

The inadequacies of the speech community model for scholars of language and
gender are overcome in the theory of the community of practice as articulated by
Eckert & McConnell-Ginet 1992, 1995.3 Rather than investing language with a
special analytic status, the community of practice framework considers language
as one of many social practices in which participants engage. By defining the
community as a group of people oriented to the same practice, though not nec-
essarily in the same way, the community of practice model treats difference and
conflict, not uniformity and consensus, as the ordinary state of affairs. The in-
herent heterogeneity of the community of practice also brings marginal members
to the forefront of analysis. One reason for this shift to the margins is that some
peripheral members are recognized as novices, as in Lave & Wenger’s original
formulation (1991). More importantly, however, the community of practice, un-
like the speech community, may be constituted around any social or linguistic
practice, no matter how marginal from the perspective of the traditional speech
community. Likewise, by focusing on individuals as well as groups, the theory of
the community of practice integrates structure with agency. And because identi-
ties are rooted in actions rather than categories, the community of practice model
can capture the multiplicity of identities at work in specific speech situations
more fully than is possible within the speech community framework. Such nu-
anced description is also facilitated by Eckert & McConnell-Ginet’s intrinsically
ethnographic approach to language and gender research. The remainder of this
article draws on the above characteristics of the community of practice to dem-

M A R Y B U C H O L T Z

210 Language in Society28:2 (1999)

cola
Highlight

cola
Highlight

cola
Highlight

onstrate the theory’s utility in the investigation of an understudied social identity
as it emerges locally in a high-school setting.

Eckert 1989a offers an account of the social organization of a typical suburban
US high school. She found that students’ social worlds and identities were de-
fined by two polar opposites: the Jocks (overachieving students who oriented to
middle-class values) and the Burnouts (underachieving students who were bound
for work, rather than college, at the end of their high-school careers). Yet the
dichotomy that separated these students also united them in what can be under-
stood as a single community of practice, since the ultimate goal of members of
both groups was to becool. The difference lay in how each group defined coolness.

Not all high-school students, however, share the Jocks’ and Burnouts’ pre-
occupation with coolness. A third group, the nerds, defines itself largely in op-
position to “cool” students – whether Jocks, Burnouts, or any other social identity.
Nerds stand as the antithesis of all these groups, a situation that Eckert succinctly
captures in her observation, “If a Jock is the opposite of a Burnout, a nerd is the
opposite of both” (1989a:48). But despite the structural significance of the nerd
in the organization of youth identities, few researchers have examined its impli-
cations, and those who have tried have fallen far short of the mark in their analy-
ses. Thus the sociologist David Kinney, in a rare study of nerds (1993), argues
that, in order to succeed socially, nerds must undergo a process of “recovery of
identity” that involves broadening one’s friendship network, participating in ex-
tracurricular activities, and heterosexual dating: In short, they must become Jocks.
Another scholarly treatment (Tolone & Tieman 1990) investigates the drug use of
nerds in an article subtitled “Are loners deviant?” – in other words, are nerds
really Burnouts?

What both studies overlook is that being a nerd is not about being a failed
Burnout or an inadequate Jock. It is about rejecting both Jockness and Burnout-
ness, and all the other forms of coolness that youth identities take. Although
previous researchers maintain that nerd identity is invalid or deficient, in fact
nerds, like Jocks and Burnouts, to a great extent consciously choose and display
their identities through language and other social practices. And where other
scholars tend to equate nerdiness with social death, I propose that nerds in US
high schools are not socially isolated misfits, but competent members of a dis-
tinctive and oppositionally defined community of practice. Nerdiness is an espe-
cially valuable resource for girls in the gendered world of the US high school.

Elsewhere (Bucholtz 1998) I describe the social identity of the nerd and detail
the phonological, syntactic, lexical, and discourse practices through which nerd
identity is linguistically indexed. Here I propose a framework for the classifica-
tion of such practices. These linguistic indices are of two kinds:Negative iden-
tity practices are those that individuals employ to distance themselves from a
rejected identity, whilepositive identity practices are those in which indi-
viduals engage in order actively to construct a chosen identity. In other words,
negative identity practices define what their users arenot, and hence emphasize

“ W H Y B E N O R M A L ? ” : L A N G U A G E A N D I D E N T I T Y P R A C T I C E S

Language in Society28:2 (1999) 211

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

identity as an intergroup phenomenon; positive identity practices define what
their usersare, and thus emphasize the intragroup aspects of social identity. The
linguistic identity practices of nerds in the present study are shown in Table 1.

The negative identity practices listed here work to disassociate nerds from
non-nerds, and especially from cool teenagers. Each of these practices, which
mark nerdy teenagers as avowedly uncool, constitutes a refusal to engage in the
pursuit of coolness that consumes other students. Meanwhile, all the positive
identity practices listed contribute to the speaker’s construction of an intelligent
self – a primary value of nerd identity. These linguistic practices also have non-
linguistic counterparts in positive and negative identity practices of other kinds
(see below).

But linguistic practices can often reveal important social information that is
not available from the examination of other community practices alone. For ex-
ample, Eckert & McConnell-Ginet 1995 apply the theory of the community of
practice to Eckert’s study of Jocks and Burnouts. Linguistic analysis revealed
that the two groups were participating at different rates in the Northern Cities
Vowel Shift, with the most innovative vowels being those used by the “Burned-
Out Burnout girls”, the most extreme adherents to this social identity. Eckert &

TABLE 1. Linguistic identity practices of nerds at Bay City High School.

Linguistic Level Negative Identity Practices Positive Identity Practices

Phonology Lesser fronting of (uw) and
(ow)a

Phonology Resistance to colloquial
phonological processes
such as vowel reduction,
consonant-cluster simplifi-
cation, and contraction

Employment of superstan-
dard and hypercorrect
phonological forms (e.g.
spelling pronunciations)

Syntax Avoidance of nonstandard
syntactic forms

Adherence to standard and
superstandard syntactic
forms

Lexicon Avoidance of current slang Employment of lexical items
associated with the formal
register (e.g. Greco-
Latinate forms)

Discourse Orientation to language form
(e.g. punning, parody,
word coinage)

aIn Bucholtz 1998 I offer a fuller discussion of the phonological and syntactic patterns of nerds. The
present article focuses primarily on lexicon and on discursive identity practices. The variables (uw)
and (ow) are part of a vowel shift that is characteristic of California teenagers (Hinton et al. 1987,
Luthin 1987). It is stereotypically associated with trendy and cool youth identities.

M A R Y B U C H O L T Z

212 Language in Society28:2 (1999)

cola
Highlight

cola
Highlight

McConnell-Ginet’s finding runs counter to the sociolinguistic tenet that “in sta-
ble variables, women use fewer non-standard variants than men of the same so-
cial class and age under the same circumstances” (Chambers 1995:112).4 The
researchers argue that the vowels employed by the Burned-Out Burnout girls are
resources through which they construct their identities as tough and streetwise;
unlike the boys, who can display their toughness through physical confronta-
tions, female Burnouts must index their identities semiotically, because fighting
is viewed as inappropriate for girls. Thus Burnout girls and boys share an orien-
tation toward toughness in their community of practice, but the practice of tough-
ness is achieved in different ways by each gender. By viewing language as
equivalent to other social practices like fighting, Eckert & McConnell-Ginet are
able to explain the ethnographic meaning of the Burnout girls’ vowel systems,
and to show how, as symbolic capital (Bourdieu 1978), language can acquire the
empowering authority of physical force itself.

Nerds, of course, attain empowerment in very different ways than either Burn-
outs or Jocks. One of the primary ways they differ from these other, more trend-
conscious groups is through the high value they place on individuality. Compared
to both Jocks and Burnouts – who must toe the subcultural line in dress, language,
friendship choices, and other social practices – nerds are somewhat less con-
strained by peer-group sanctions.

For girls, nerd identity also offers an alternative to the pressures of hegemonic
femininity – an ideological construct that is at best incompatible with, and at
worst hostile to, female intellectual ability. Nerd girls’ conscious opposition to
this ideology is evident in every aspect of their lives, from language to hexis to
other aspects of self-presentation. Where cool girls aim for either cuteness or
sophistication in their personal style, nerd girls aim for silliness. Cool girls play
soccer or basketball; nerd girls play badminton. Cool girls read fashion maga-
zines; nerd girls read novels. Cool girls wear tight T-shirts, and either very tight
or very baggy jeans; nerd girls wear shirts and jeans that are neither tight nor
extremely baggy. Cool girls wear pastels or dark tones; nerd girls wear bright
primary colors. But these practices are specific to individuals; they are engaged
in by particular nerd girls, not all of them.

The community of practice model accommodates the individuality that is par-
amount in the nerd social identity, without overlooking the strong community ties
that unify the nerd girls in this study. The community of practice also allows us to
look at nerd girls in the same way that Eckert & McConnell-Ginet 1999 view the
Burnout girls: as speakersand social actors, as individualsand members of
communities, and as both resisting and responding to cultural ideologies of gender.

I D E N T I T Y P R A C T I C E S I N A L O C A L N E R D C O M M U N I T Y

To illustrate the value of the community of practice framework, I will focus on a
single social group that displays the nerd social identity. Nerds at the high school

“ W H Y B E N O R M A L ? ” : L A N G U A G E A N D I D E N T I T Y P R A C T I C E S

Language in Society28:2 (1999) 213

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

cola
Highlight

in my study constitute a single community insofar as they engage in shared prac-
tices, but this identity is divided into particular social groups whose members
associate primarily with one another, and these groups form their own commu-
nities of practice. In communities of practice, unlike speech communities, the
boundaries are determined not externally by linguists, but internally through eth-
nographically specific social meanings of language use. As suggested above,
ethnographic methods therefore become crucial to the investigation of commu-
nities of practice.

The ethnographic fieldwork from which the data are taken was carried out
during the 1994 –95 academic year at a California high school that I call Bay City
High. The social group of nerd girls that is the focus of this discussion is a small,
cohesive friendship group that comprises four central members – Fred, Bob, Kate,
and Loden – and two peripheral members, Carrie and Ada. (Ada does not appear
in the data that follow.) All the girls are European American except Ada, who is
Asian American. The same group also formed a club, which I will call the Ran-
dom Reigns Supreme Club.5

Random Reigns Supreme is more properly described as an anti-club, which is
in keeping with the counter-hegemonic orientation of nerd identity. It was created
by members in order to celebrate their own preferences, from Sesame Street to
cows to Mr. Salty the pretzel man. Members emphasize the “randomness” of the
club’s structure. It is not organized around shared preferences; instead, any indi-
vidual’s preferences can be part of the club’s de facto charter, and all six members
are co-presidents. This structure contrasts with the corporate focus and hierar-
chical structure of most school clubs, which bring together people who are other-
wise unconnected to perform a shared activity (Eckert 1989a). The Random Reigns
Supreme Club centers around members’ daily practices, not specialized activi-
ties. It has no goals, no ongoing projects, and no official meetings. Nevertheless,
members proudly take their place among the corporate clubs in the pages of the
school’s yearbook. The girls’ insistence on being photographed for the yearbook
has a subversive quality: The photo publicly documents the existence of this
otherwise little-recognized friendship group, and demands its institutional legit-
imacy on par with the French Club, the Backpacking Club, and other activity-
based organizations. Like their yearbook photograph, the language used by the
girls not only marks their nerd identity but also expresses their separation from
outsiders. As shown by the following examples (taken from a single interaction),
the details of interaction are important and contested resources in defining a
shared oppositional nerd identity within the club’s community of practice.

Positive identity practices

As indicated above, many positive identity practices in which nerds engage con-
tribute to the display of intelligence. The community value placed on intelligence
is reflected in non-linguistic identity practices oriented to the world of school,
books, and knowledge. This orientation is amply illustrated in the following.6

M A R Y B U C H O L T Z

214 Language in Society28:2 (1999)

cola
Highlight

cola
Highlight

cola
Highlight

(1) 1 Carrie: Where where do those seeds come from?
2 ^points to her bagel&
3 ^laughter&
4 Bob: [Poppies. ]
5 Fred: [Sesame plants.]
6 Carrie: {But what do they look like?}̂high pitch&
7 Fred: I have no idea. hh
8 Bob: Sesame:.
9 Carrie: [Is anybody– h]

10 Fred: Ask me (.) [tomorrow. ]
11 I’ll look it up for you. h
12 Carrie: h Is anybody here knowledgeable about (.)
13 the seeds on top of bagels?/
14 Fred: /Sesame.
15 Bob: They’re sesame?
16 They’re not sunfl– ?
17 No,
18 of course they’re not sunflower.
19 Loden: Yeah,
20 [What kind of seeds are– ]
21 Carrie: [Because sunflower are those whopping ones?]
22 Bob: [Yeah.
23 Yeah.
24 I know. ]
25 ^laughter&

Carrie’s question in line 1 creates the conditions for intellectual display. Although
the humor of the question is acknowledged through laughter (line 2), it receives
immediate, serious uptake from two participants, Bob and Fred (lines 4 –5). Car-
rie’s subsequent question (line 6), however, forces an admission of ignorance
from Fred (line 7).

Because knowledge is symbolic capital within the nerd community of prac-
tice, Fred’s admission results in some loss of face. She recovers from this (minor)
social setback by invoking the authority of a reference book (I’ll look it up for
you, line 11). In this way Fred can safely assure her interlocutor that, although she
does not yet know the answer, she soon will. She is also able to one-up Bob, who
has misidentified the bagel seeds (line 4) and continues to show some skepticism
about Fred’s classification of them (Sesame:, line 8). Fred tracks this indirect
challenge for five lines, through her own turn and Carrie’s next question; rather
than continuing to participate in the series of adjacency pairs that Carrie has
initiated (lines 12–13), she responds to Bob (line 14). Fred thus succeeds in dis-
playing both actual knowledge, about the type of seeds under discussion, and
potential knowledge, about the appearance of sesame plants.

Claims to knowledge are, however, often disputed in this community of prac-
tice. After Bob provides an incorrect answer to Carrie and receives a correction
from Fred, she continues to exhibit doubt about Fred’s knowledge (line 15). She
offers a second incorrect identification of the seeds in line 16, but this time she
interrupts herself and self-corrects (lines 17–18), in an effort to prevent further
other-correction. She does not succeed, however; and when Carrie explains why

“ W H Y B E N O R M A L ? ” : L A N G U A G E A N D I D E N T I T Y P R A C T I C E S

Language in Society28:2 (1999) 215

cola
Highlight

Bob is mistaken, the latter overlaps with her, offering three quick acknowledg-
ments that are designed to cut off Carrie’s turn (lines 22–24).

This passage shows several deviations from the preference organization of
repair in conversation (Schegloff et al. 1977), according to which self-initiation
and self-repair are preferred over initiation and repair by another. Bob twice
initiates dispreferred repairs of Fred’s turns (lines 8, 15), and she even begins to
carry out the repair itself in line 16. When Bob initiates a repair of her own
utterance through self-interruption in the same line, Carrie performs the repair
despite Bob’s efforts to prevent her from doing so (lines 21–24). The frequent
apparent violations of repair organization suggest that, in this community of prac-
tice, self-repair is preferred only by the speaker; the listener’s positive face (the
desire to be viewed as intelligent) wars against and often overrides consideration
of the speaker’s negative face (the desire not to be viewed as unintelligent).

Bob’s loss of face in ex. 1 leads her, in ex. 2, to initiate a new conversational
direction:

(2) 26 Bob: They come from trees.
27 They have big trees and they just
28 [ra:in down seeds]
29 [̂ laughter& ]
30 Carrie: [No they don’t. ]
31 Uh uh.
32 Why would little tiny seeds [come from– ]
33 Fred: [{into baskets.}]̂ smiling quality&
34 Ye:p,
35 [({I’ve been there.})]^smiling quality&
36 Carrie: [No:. ]
37 Loden: [No:. ]
38 Bob: [[Little tiny leaves come from trees, ]]
39 Fred: [[And the whole culture’s built around it,]]
40 like in: some countries,
41 All they do is like the women come out and they have ba(h)skets on
42 th(h)eir h(h)eads and they st(h)and under a [tree,]

Bob jokingly provides an authoritative answer to Carrie’s question (lines 26–28)
and thereby skillfully shifts attention from her own lack of knowledge to Carrie’s.
Fred eagerly joins in with the parody of scientific discourse, amplifying on the
theme while supplying invented anthropological details that invoke the didactic
style of a typical high-school classroom or public television documentary (33–
35, 39– 42). Such teasing episodes are frequent in this friendship group. But more
importantly, this exchange is a collaborative performance of nerd identity: The
participants collude in sustaining the frame of an intellectual debate, even as
laughter keys the talk as play. Nerd identities are here jointly constructed and
displayed.

In ex. 3, Carrie – who up to this point has mostly provided opportunities for
others to display their nerd identities, rather than participating herself (but see
below) – shifts the topic, which she sustains for the rest of the interaction:

M A R Y B U C H O L T Z

216 Language in Society28:2 (1999)

cola
Highlight

(3) 43 Carrie: [My– ]
44 You sound like my crusty king,
45 I’m writing this (.) poem because I have to like incorporate these
46 words into a poem, and it’s all about–
47 ^interruption, lines omitted&
48 Fred: So what about this king?

Carrie’s discussion of a class assignment returns to a central value of nerdiness:
school. The topic is sustained for 56 lines and 26 turns; and although it is inter-
rupted immediately after Carrie introduces it (line 47), Fred prompts her to return
to the subject several minutes later (line 48). Carrie’s enthusiastic description of
her poem – and the eager participation of others in this topic – is rare among
students with cool social identities, but it is quite common among nerds, for
whom academic pursuits are a central resource for identity practices.

At the same time, however, Carrie’s selection of subject matter for her poem,
with its mildly scatological – or at least “gross” theme (line 80) – is playfully
subversive of school values and emphatically counter to traditional feminine top-
ics, as ex. 4 illustrates:

(4) 49 Carrie: He’s like (.) has this (.) castle,
50 (xxx: Is he xxx king?)
51 Carrie: No–
52 Yeah,
53 he is.
54 Loden: hh
55 Carrie: He has this–
56 {He has this castle right?
57 except it’s all crusty,}
58 ^rustling of lunch bag, clanging of aluminum can&
59 (Fred: Uh huh.)
60 Carrie: And so he lives on a boat [in the moat.]
61 Bob: [A crusty– ]
62 ^Fred crushes her aluminum can&
63 Kate: Who:a!
64 ^quiet laughter&
65 Bob: Is it really [crusty?]
66 Carrie: [He’s– ]
67 And so like the– like because– the people are trying to convince
68 him that like he should stay in the castle and he’s all,
69 {“No, it’s crusty!”} ^high pitch, tensed vocal cords&
70 [̂ laughter& ]
71 Carrie: [{“I’m in the moat!”}] ^high pitch, quiet&
72 right,
73 Bob: What’s wrong with [crusty castles?]
74 Carrie: [And so– ]
75 Well,
76 Would [you want to live ]5
77 Kate: [Crusty (castles). ]
78 Carrie: 5in a castle full of crust?
79 {[ é@é]} ^noise of disgust and disapproval&
80 Kate: [How gross.]
81 Bob: [I mi:ght. ]
82 Carrie: Huh?

“ W H Y B E N O R M A L ? ” : L A N G U A G E A N D I D E N T I T Y P R A C T I C E S

Language in Society28:2 (1999) 217

Bob here enters into the unfeminine spirit of Carrie’s narrative, even outdoing
Carrie with her repeated insistence on her own immunity from “gross” subjects
like crustiness (lines 73, 81). A competitive tone is also evident in the multiple
challenges she issues to Carrie throughout the latter’s narrative (lines 65, 73). As
questions, these challenges echo Carrie’s earlier questions (lines 1, 6, 12–13); but
whereas Carrie’s appeared to be genuine information-seeking questions, Bob’s
are not. Carrie’s recognition of this fact is shown by her failure to respond at all
to the first question, and by her answering the second question with an equally
challenging question of her own (Would you want to live in a castle full of crust?,
lines 76, 78). Bob’s face-threatening response (I mi:ght, line 81) perpetuates the
jocular-combative tone. In ex. 5, however, this combativeness becomes not a
shared resource for joint identity construction, but a marker of social division.
The positive identity practices that dominate in the earlier part of the interaction
are replaced by negative identity practices, as community members experience a
threat not only to their face but also to their identities.

Negative identity practices

Example 5 is a continuation of Bob’s face-threatening questions to Carrie. This
final series of questions is unified through a shared template (like1 adj 1 crust);
their syntactic similarity emphasizes that they are designed as a series, and it thus
produces an effect of unremitting interrogation.

(5) 83 Bob: What kind of crust?
84 Like,
85 bread crust?
86 Carrie: Like
87 Bob: Like [eye crust? ]
88 Carrie: [crusty crust.]
89 Like {boo:ts y} ^high pitch, tensed vocal cords&
90 crust.
91 ^laughter&
92 Bob: Oh.
93 Well,
94 Maybe if it’s bootsy,
95 I don’t know.
96 Fred: {Boot[sy! ]} ^falsetto, sing-song&
97 Kate: [̂ coughs&]
98 ^laughter&

These questions display Bob’s nerd identity through her use of puns on the word
crust (lines 85, 87). Punning, as a discourse practice that orients to linguistic
form, is characteristic of nerds’ discourse style (see Table 1). Carrie’s refusal (line
88) to participate in Bob’s punning thus constitutes a negative identity practice –
one which, moreover, indexes a rejection of nerd identity as it has been con-
structed through preceding interactional practices. The refusal is made more
evident by her exploitation (lines 86, 88–90) of Bob’s syntactic template. By
conforming to the syntactic form of Bob’s turn, while failing to conform to the

M A R Y B U C H O L T Z

218 Language in Society28:2 (1999)

cola
Highlight

cola
Highlight

discourse practice of punning, Carrie separates herself from Bob at a point when
the latter is fully engaged in nerdy identity practices.

This analysis is confirmed by Carrie’s choice of upgraded adjective in line
89. Bootsyis a slang term with a negative evaluative sense; it is not used by
other members of the Random Reigns Supreme Club. The introduction of youth
slang into a group that explicitly rejects such linguistic forms is part of a strongly
negative identity practice, and the reactions of Carrie’s interlocutors are corre-
spondingly negative: Bob’s response (lines 92–95) jokingly concedes the point,
while underscoring that Carrie has violated the rules of nerdy argument by
appealing to the authority of cool youth culture. Fred’s mocking repetition of
the term (line 96) demonstrates that the use of slang is itself worthy of com-
ment. With Carrie’s narrative entirely derailed – it never becomes clear how it
is connected to the earlier discussion – she soon afterward moves away from
the group.

The complex interaction presented above reveals Carrie’s peripheral status in
this community of practice. As a non-core member, she moves between friend-
ship groups – in fact, the interaction occurred when Carrie approached the core
group in the middle of lunch period. Carrie’s social flexibility has made her a
cultural and linguistic broker for the Random Reigns Supreme Club, whose mem-
bers become aware of current youth slang in large part through contact with her.
Hence many slang terms that circulate widely in the “cool” groups are labeled by
club members as “Carrie words”.

Yet Carrie also demonstrates her ability and willingness to participate in the
group’s positive identity practices. She does so most obviously by engaging in
sound play in recounting her poem (crusty king, line 44;a boat in the moat, line
60). More significant, though, is the subtle shift in her speech practices at the
beginning of the interaction. Thus Carrie’s questionIs anybody here knowledge-
able about (.) the seeds on top of bagels?(lines 12–13) draws on the formal
register through her choice of the wordknowledgeable. Among nerds, this reg-
ister projects a speaker’s persona as smart and highly educated. But the use of the
formal register is strategic, not a mechanical result of membership in a particular
social category. This point is supported by the fact that Carrie employs the nerd
identity practice only after she asks two related questions in colloquial register
(lines 1, 6). Her unwillingness to overlap her turn with Fred’s (lines 9, 10) further
suggests that the question is a performance of nerdiness, not just a manifestation
of it; she does not produce her utterance until she is assured of an attentive audi-
ence. That is, Carrie is simultaneously displaying and commenting on nerd prac-
tice – showing her awareness of nerdy linguistic forms, and announcing her
willingness to enter a nerdy interactional space by carefully gauging her utter-
ance to match the group’s practices. Thus Carrie’s performance of nerdiness places
her within the community of practice; but her use of slang, as the other members
are quick to let her know, moves her outside it. Such adjustments at interactional
boundaries may reflect adjustments at community boundaries.

“ W H Y B E N O R M A L ? ” : L A N G U A G E A N D I D E N T I T Y P R A C T I C E S

Language in Society28:2 (1999) 219

cola
Highlight

cola
Highlight

cola
Highlight

C O N C L U S I O N

Because all the participants in the above exchange are middle-class European
American girls from the same California city, the traditional sociolinguistic per-
spective would classify them unproblematically as members of the same speech
community. Such an analysis would overlook the details of greatest interest to
language and gender researchers: the performances of identity, and the struggles
over it, which are achieved through language. However, by viewing the inter-
action as the product of a community of practice, we can avoid this problem, as
well as others associated with the speech community model.

The ethnographic method brings into view the social meanings with which
participants invest their practices. These meanings emerge on the ground in local
contexts; thus what it means to display academic knowledge, or to use slang,
depends not on fixed identity categories but on where one is standing. Nor do
participants necessarily agree on the meanings of their actions; nerdiness, like all
identities, is a contested domain in which speakers struggle both over control of
shared values, via positive identity practices (Who’s better at being a nerd?), and
over control of identity itself, via negative identity practices (Who counts as a
nerd?). Such conflicts reveal the heterogeneity of membership in the community
of practice – its constitution through the work of central and peripheral members
alike. In this project, the interactional choices of specific individuals matter. Thus
Carrie’s identity is on display – and at risk – in a way that Loden’s, for example,
is not. These actions must be seen as choices, not as the outputs of interactional
algorithms. While some practices reproduce the existing local social structure (as
does Carrie’s use of the formal register), others undermine it (e.g. her use of
slang). Likewise, some nerdy practices (such as being good students) comply
with the larger social order, while others (such as rejecting femininity) resist it.
Linguistic practices, moreover, have no special status in this process. Instead,
they work in conjunction with other social practices to produce meanings and
identities. Bob’s interactional work to distance herself from hegemonic feminin-
ity, for instance, is part of her overall participation in anti-feminine practices and
her non-participation in feminine practices, as evidenced also by her physical
self-presentation.

For sociolinguists, the community of practice represents an improvement over
the speech community in that it addresses itself to both the social and the linguis-
tic aspects of the discipline. As a well-grounded framework with currency in a
number of fields, practice theory in general, in particular the community of prac-
tice, revitalizes social theory within sociolinguistics. What is more, it does so at
a sufficiently general level to accommodate multiple dimensions of social analy-
sis – including both structure and agency, both ideology and identity, both norms
and interactions. The community of practice also provides an avenue for a more
complete sociolinguistic investigation of identity. Although introduced for gender-
based research, the community of practice has never been restricted to the analy-
sis of a single element of identity. Indeed, it lends itself to the simultaneous

M A R Y B U C H O L T Z

220 Language in Society28:2 (1999)

cola
Highlight

cola
Highlight

cola
Highlight

investigation of multiple aspects of the self, from those at the macro level – like
gender, ethnicity, and class – to micro-identities like Jocks, Burnouts, or nerds.
The framework also allows for the study of interaction between levels of identity.
The concepts of positive and negative identity practices, as proposed in this ar-
ticle, are intended as one way to develop the potential of the community of prac-
tice in this arena.

In addition to its benefits for social analysis, the community of practice offers
an integrated approach to linguistic analysis. By understanding all socially mean-
ingful language use as practices tied to various communities, the model enables
researchers to provide more complete linguistic descriptions – along with social
explanations – of particular social groups. Moreover, the community of practice
provides a way to bring qualitative and quantitative research closer together.
Because both kinds of linguistic data emerge from practice, both can be included
in a single analysis. This richly contextualized approach to both language and
society is one of the great strengths of the community of practice as a sociolin-
guistic framework.

The community of practice, having revolutionized the field of language and
gender almost as soon as it was first proposed, enables researchers of socially
situated language use to view language within the context of social practice.
Perhaps the most valuable feature is that the community of practice admits a
range of social and linguistic phenomena that are not analyzed in other theoretical
models. Local identities, and the linguistic practices that produce them, become
visible to sociolinguistic analysis as the purposeful choices of agentive individ-
uals, operating within (and alongside and outside) the constraints of the social
structure. To describe and explain such complexity must be the next step not only
for language and gender scholars, but for all sociolinguists concerned with the
linguistic construction of the social world.

N O T E S

* My thanks to Janet Holmes, Chris Holcomb, Stephanie Stanbro, and members of the Ethnography/
Theory Group at Texas A&M University for comments on and discussion of the ideas in this article.

1 The work of Barbara Horvath on immigrants in Sydney’s speech community (Horvath 1985,
Horvath & Sankoff 1987) has done a great deal to correct this omission.

2 Santa Ana & Parodi’s model of nested speech communities (1998) is a recent attempt to address
this problem.

3 A fuller discussion of the advantages of practice theory for language and gender research is
provided by Bucholtz 1999.

4 Eckert 1989b calls this simple formulation into question; see also Labov 1990 for a response.
5 Though this is not its actual name, it preserves the flavor of the original. All other names are

pseudonyms chosen by the speakers.
6 Transcription conventions are as follows:

. end of intonation unit; falling intonation
, end of intonation unit; fall-rise intonation
? end of intonation unit; rising intonation
– self-interruption
: length

“ W H Y B E N O R M A L ? ” : L A N G U A G E A N D I D E N T I T Y P R A C T I C E S

Language in Society28:2 (1999) 221

underline emphatic stress or increased amplitude
(.) pause of 0.5 seconds or less
(n.n) pause of greater than 0.5 seconds, measured by a stopwatch
h exhalation (e.g. laughter, sigh); each token marks one pulse
( ) uncertain transcription
^ & transcriber comment; nonvocal noise
{ } stretch of talk over which a transcriber comment applies
[ ] overlap beginning and end
/ latching (no pause between speaker turns)
5 no pause between intonation units

The transcript emphasizes sequential organization in order to highlight speakers’ orientation to one
another. It excludes phonological detail that is necessary for a complete analysis of nerd identity
performance.

R E F E R E N C E S

Bloomfield, Leonard (1933).Language. New York: Holt.
Bourdieu, Pierre (1978).Outline of a theory of practice. Cambridge & New York: Cambridge Uni-

versity Press.
_(1991).Language and symbolic power. Cambridge, MA: Harvard University Press.
Bucholtz, Mary (1998). Geek the girl: Language, femininity, and female nerds. In Natasha Warner

et al. (eds.),Gender and belief systems: Proceedings of the Fourth Berkeley Women and Language
Conference, 119–31. Berkeley: Berkeley Women and Language Group.
_(1999). Bad examples: Transgression and progress in language and gender research. In Mary

Bucholtz et al. (eds.),Reinventing identities. To appear, New York: Oxford University Press.
Cameron, Deborah (1992).Feminism and linguistic theory. 2d edn. New York: St. Martin’s Press.
Chambers, J. K. (1995).Sociolinguistic theory. Oxford: Blackwell.
Certeau, Michel de (1984).The practice of everyday life. Berkeley: University of California Press.
Dorian, Nancy C. (1982). Defining the speech community to include its working margins. In Suzanne

Romaine (ed.),Sociolinguistic variation in speech communities, 25–33. London: Arnold.
Eckert, Penelope (1989a).Jocks and Burnouts: Social categories and identity in the high school. New

York: Teachers College Press.
_(1989b). The whole woman: Sex and gender differences in variation.Language Variation and

Change1:245– 67.
_, & McConnell-Ginet, Sally (1992). Think practically and look locally: Language and gender

as community-based practice.Annual Review of Anthropology21:461–90.
_,_(1995). Constructing meaning, constructing selves: Snapshots of language, gender,

and class from Belten High. In Kira Hall & Mary Bucholtz (eds.),Gender articulated: Language
and the socially constructed self, 459–507. London: Routledge.

Giddens, Anthony (1979).Central problems in social theory: Action, structure, and contradiction in
social analysis. Berkeley: University of California Press.

Guy, Gregory R. (1988). Language and social class. In Frederick J. Newmeyer (ed.),Linguistics: The
Cambridge survey, vol. 4: Language: The socio-cultural context, 37– 63. Cambridge & New York:
Cambridge University Press. 37-63.

Hinton, Leanne, et al. (1987). It’s not just the Valley Girls: A study of California English.Berkeley
Linguistics Society13:117–28.

Holmes, Janet (1997). Women, language and identity.Journal of Sociolinguistics1:195–223.
Horvath, Barbara (1985).Variation in Australian English: The sociolects of Sydney. Cambridge &

New York: Cambridge University Press.
_, & Sankoff, David (1987). Delimiting the Sydney speech community.Language in Society

16:179–204.
Hudson, Richard A. (1980).Sociolinguistics. Cambridge & New York: Cambridge University Press.
Hymes, Dell (1974).Foundations in sociolinguistics: An ethnographic approach. Philadelphia: Uni-

versity of Pennsylvania Press.
Johnstone, Barbara (1995). Sociolinguistic resources, individual identities, and public speech styles

of Texas women.Journal of Linguistic Anthropology5:183–202.

M A R Y B U C H O L T Z

222 Language in Society28:2 (1999)

_(1996).The linguistic individual: Self-expression in language and linguistics. Oxford & New
York: Oxford University Press.
_, & Bean, Judith Mattson (1997). Self-expression and linguistic variation.Language in Soci-

ety 26:221– 46.
Kinney, David A. (1993). From nerds to normals: The recovery of identity among adolescents from

middle school to high school.Sociology of Education66:1.21– 40.
Labov, William (1966).The social stratification of English in New York City. Washington, DC: Center

for Applied Linguistics.
_(1972). The reflection of social processes in linguistic structures. In hisSociolinguistic pat-

terns, 110 –21. Philadelphia: University of Pennsylvania Press.
_(1990). The intersection of sex and social class in the course of linguistic change.Language

Variation and Change2:205–54.
Lave, Jean (1988).Cognition in practice: Mind, mathematics, and culture in everyday life. Cam-

bridge & New York: Cambridge University Press.
_, & Wenger, Etienne (1991).Situated learning: Legitimate peripheral participation. Cam-

bridge & New York: Cambridge University Press.
Luthin, Herbert W. (1987). The story of California (ow): The coming-of-age of English in California.

In Keith M. Denning et al. (eds.),Variation in language: NWAV-XV at Stanford, 312–24. Stanford,
CA: Department of Linguistics, Stanford University.

Milroy, James (1992).Linguistic variation and change: On the historical sociolinguistics of English.
Oxford: Blackwell.

Morgan, Marcyliena (1999). No woman, no cry: Claiming African American women’s place. In Mary
Bucholtz et al. (eds.),Reinventing identities(to appear).

Ortner, Sherry (1996).Making gender: The politics and erotics of culture. Boston: Beacon.
Pratt, Mary Louise (1987). Linguistic utopias. In Nigel Fabb et al. (eds.),The linguistics of writing:

Arguments between language and literature, 48– 66. New York: Methuen.
Rickford, John R. (1986). The need for new approaches to social class analysis in sociolinguistics.

Language and Communication6:3.215–21.
Romaine, Suzanne (1982). What is a speech community? In Suzanne Romaine (ed.),Sociolinguistic

variation in speech communities, 13–24. London: Arnold.
Santa Ana, Otto, & Parodi, Claudia (1998). Modeling the speech community: Configuration and

variable types in the Mexican Spanish setting.Language in Society27:23-51.
Schegloff, Emanuel; Jefferson, Gail; & Sacks, Harvey (1977). The preference for self-correction in

the organization of repair in conversation.Language53:361–82.
Silverstein, Michael (1996). Encountering languages and languages of encounter in North American

ethnohistory.Journal of Linguistic Anthropology6:126– 44.
Tolone, W. L., & C. R. Tieman (1990). Drugs, delinquency and nerds: Are loners deviant?Journal of

Drug Education20:2.153– 62.
Wenger, Etienne (1998).Communities of practice. Cambridge & New York: Cambridge University

Press.
Williams, Glyn (1992).Sociolinguistics: A sociological critique. London: Routledge.

“ W H Y B E N O R M A L ? ” : L A N G U A G E A N D I D E N T I T Y P R A C T I C E S

Language in Society28:2 (1999) 223