Social Networks.
LESLEY MILROY.
An individual's social network is
straightforwardly the aggregate of relationships contracted with
others, and social network analysis examines the differing structures and properties of these
relationships. Such analysis has been applied by variationists fairly extensively over the last two
decades or so to explicate informal social mechanisms supporting language varieties specific to
particular social groups. Researchers have also addressed the question of how some social groups
maintain nonstandard dialects or minority languages, often over centuries, despite pressures (of the
kind described by Li ppi-Green 1 997) to adopt publicly legit imized natio nal languages or varieties .
others, and social network analysis examines the differing structures and properties of these
relationships. Such analysis has been applied by variationists fairly extensively over the last two
decades or so to explicate informal social mechanisms supporting language varieties specific to
particular social groups. Researchers have also addressed the question of how some social groups
maintain nonstandard dialects or minority languages, often over centuries, despite pressures (of the
kind described by Li ppi-Green 1 997) to adopt publicly legit imized natio nal languages or varieties .
Social network is better
treated as a means of capturing the dynamics underlying speakers'
interactional behaviors than as a f ixed social cat ego ry (see Ec kert 2000: 1 - 3 3 for a discus s ion of
dif ferent conce pt s of "speaker variables" in sociolingui st ic s) . Given that the ties contracted by
individuals within and between speech communities may change for many reasons, analysis of change
in the operation of the same social network mechanisms which support localized linguistic codes can
illu minate the pheno meno n of lingu is tic change . Networ k-oriented accou nt s of linguis ti c change have
emerged bot h in variat ionis t s tudies of cont em porary s peech comm unities, and as po st hoc
sociohist ori cal s tudies of changes com plet ed at earlier st ages of t he language (Li p pi-Green 1 9 8 9,
Milroy 1 992 , Milroy and Milroy 1 9 85 , Nevalai nen 2 000) .
interactional behaviors than as a f ixed social cat ego ry (see Ec kert 2000: 1 - 3 3 for a discus s ion of
dif ferent conce pt s of "speaker variables" in sociolingui st ic s) . Given that the ties contracted by
individuals within and between speech communities may change for many reasons, analysis of change
in the operation of the same social network mechanisms which support localized linguistic codes can
illu minate the pheno meno n of lingu is tic change . Networ k-oriented accou nt s of linguis ti c change have
emerged bot h in variat ionis t s tudies of cont em porary s peech comm unities, and as po st hoc
sociohist ori cal s tudies of changes com plet ed at earlier st ages of t he language (Li p pi-Green 1 9 8 9,
Milroy 1 992 , Milroy and Milroy 1 9 85 , Nevalai nen 2 000) .
1 The Concept of Social
Network
Social network analysis of the kind generally adopted
by variationists was developed by social
anthropologis t s mainly du ring t he 1 960s and 1 970s (see Milroy 1 987a, Li 1 996 , Johnson 1 994) .
Contrary t o the as sertions of Mu rray (1 993 : 1 6 2) , it is clear from even a cursory reading of the
literature that no canonical "real" procedure for analyzing social networks can be identified; scholars
from many different disciplines employ the concept for a range of theoretical and practical reasons.
For exam ple , Johnson's (1 994) survey alludes t o a wide range of approaches within anthropology
which hardly overlap with the largely quantitative modes of analysis described by Cochran et al.
(1990). This international and interdisciplinary team of scholars is interested in the role of networks in
providing support for urban families. Accordingly, their methods are to a great extent driven by a
concern with social policy and practice.
anthropologis t s mainly du ring t he 1 960s and 1 970s (see Milroy 1 987a, Li 1 996 , Johnson 1 994) .
Contrary t o the as sertions of Mu rray (1 993 : 1 6 2) , it is clear from even a cursory reading of the
literature that no canonical "real" procedure for analyzing social networks can be identified; scholars
from many different disciplines employ the concept for a range of theoretical and practical reasons.
For exam ple , Johnson's (1 994) survey alludes t o a wide range of approaches within anthropology
which hardly overlap with the largely quantitative modes of analysis described by Cochran et al.
(1990). This international and interdisciplinary team of scholars is interested in the role of networks in
providing support for urban families. Accordingly, their methods are to a great extent driven by a
concern with social policy and practice.
Personal social networks are
always seen as contextualized within a macro-level social framework,
which is "bracketed
off" for purely methodological reasons-i.e. to focus on less abstract
modes of
analysis capable of accounting more immediately for the variable behavior of individuals. Since no one
claims that personal network structure is independent of broader social, economic, or political
analysis capable of accounting more immediately for the variable behavior of individuals. Since no one
claims that personal network structure is independent of broader social, economic, or political
frameworks constraining
individual behavior, a social network analysis of language variation does not
compete with an analysis in terms of a macro-level concept such as social class.
compete with an analysis in terms of a macro-level concept such as social class.
A fundamental postulate of network analysis is that individuals create
personal communities which
provide a meaningful framework for solving the problems of daily life (Mitchell 1986: 74). These personal communities are constituted by interpersonal ties of different types and strengths, and
provide a meaningful framework for solving the problems of daily life (Mitchell 1986: 74). These personal communities are constituted by interpersonal ties of different types and strengths, and
structural relationships between links can vary. Particularly, the
persons to whom ego is linked may
also be tied to each other to varying degrees-ego being the person who, for analytic reasons, forms
the "anchor" of the network. A further postulate with particular relevance to language maintenance or
change is that structural and content differences between networks impinge critically on the way they
directly affect ego. Particularly, if a networ k consi st s chiefly of st rong t ies , and t hose t ies are
multiplex or many-s tranded , and if the networ k is also relatively dense-i .e . many of ego's ties are
linked to each other-then such a network has the capacity to support its members in both practical
and symbolic ways. More negatively, such a network type can impose unwanted and stressful
constraints on its members. Thus, we come to the basic point of using network analysis in variationist
research. Networks constituted chiefly of strong (dense and multiplex) ties support localized linguistic
norms, resisting pressures to adopt competing external norms. By the same token, if these ties
weaken conditions favorable to language change are produced. The idealized maximally dense and
multiplex networ k i s shown in figu re 2 2 . I in cont ras t wit h a loose- knit , u ni plex type of networ k
shown in (figure 2 2 . 2) .
also be tied to each other to varying degrees-ego being the person who, for analytic reasons, forms
the "anchor" of the network. A further postulate with particular relevance to language maintenance or
change is that structural and content differences between networks impinge critically on the way they
directly affect ego. Particularly, if a networ k consi st s chiefly of st rong t ies , and t hose t ies are
multiplex or many-s tranded , and if the networ k is also relatively dense-i .e . many of ego's ties are
linked to each other-then such a network has the capacity to support its members in both practical
and symbolic ways. More negatively, such a network type can impose unwanted and stressful
constraints on its members. Thus, we come to the basic point of using network analysis in variationist
research. Networks constituted chiefly of strong (dense and multiplex) ties support localized linguistic
norms, resisting pressures to adopt competing external norms. By the same token, if these ties
weaken conditions favorable to language change are produced. The idealized maximally dense and
multiplex networ k i s shown in figu re 2 2 . I in cont ras t wit h a loose- knit , u ni plex type of networ k
shown in (figure 2 2 . 2) .
A social network may be seen as a boundless web
of ties which reaches out through a whole society,
linking people to one another, however remotely. However, interest generally focuses on first-order
network ties, constituted by those persons with whom an individual directly interacts. Second-order
ties are those to whom the link is indirect, as shown also in figure 2 2 . I . Within t he firs t order zone , it
is i mport ant for the reasons noted above to distinguish between "strong" and "weak" ties of everyday
life-roughly ties which connect friends or kin as opposed to those which connect acquaintances. To
supplement the notions of multiplexity and density, Milardo distinguishes "exchange" from
"interactive" networ ks ( I 98 8: 26 - 36) . Exchange networ ks consi st of persons such as kin and close
friends with whom ego not only i nt eract s regularly, but also exchanges direct aid, advice, criticism,
and support. Interactive networks on the other hand consist of persons with whom ego interacts
frequently and perhaps over prolonged periods of time, but on whom he or she does not rely for
material or symbolic resources. An example of an interactive tie would be that between a store owner
and cu s tomer . In addi tion to exchange and int eractive ties , Li ( I 994) dis tinguishes a "pas sive" tie ,
which seem s particularly i mport ant t o migrant or mobile individuals . Pas sive ties entail an absence of
regular contact, but are valued by ego as a source of influence and moral support. Examples are
physically distant relatives or friends.
linking people to one another, however remotely. However, interest generally focuses on first-order
network ties, constituted by those persons with whom an individual directly interacts. Second-order
ties are those to whom the link is indirect, as shown also in figure 2 2 . I . Within t he firs t order zone , it
is i mport ant for the reasons noted above to distinguish between "strong" and "weak" ties of everyday
life-roughly ties which connect friends or kin as opposed to those which connect acquaintances. To
supplement the notions of multiplexity and density, Milardo distinguishes "exchange" from
"interactive" networ ks ( I 98 8: 26 - 36) . Exchange networ ks consi st of persons such as kin and close
friends with whom ego not only i nt eract s regularly, but also exchanges direct aid, advice, criticism,
and support. Interactive networks on the other hand consist of persons with whom ego interacts
frequently and perhaps over prolonged periods of time, but on whom he or she does not rely for
material or symbolic resources. An example of an interactive tie would be that between a store owner
and cu s tomer . In addi tion to exchange and int eractive ties , Li ( I 994) dis tinguishes a "pas sive" tie ,
which seem s particularly i mport ant t o migrant or mobile individuals . Pas sive ties entail an absence of
regular contact, but are valued by ego as a source of influence and moral support. Examples are
physically distant relatives or friends.
Eckert notes that the people who
comprise an individual's personal communities change, as indeed do
the everyday problems which such personal communities help to solve (2000: 34). Furthermore,
individuals engage on a daily basis in a variety of endeavors in multiple personal communities. Eckert
em ploys the co nce pt of community of practice, whi ch is closely relat ed to that of social network, to
locate the int eract tonal s ites where social meaning i s indexed by Ii nguis ti c ele ment s, and Ii nguis t ic
change and social meaning are co-constructed. A community of practice may be defined as an
aggregate of people coming together around a particular enterprise (Eckert 2000: 35), and in her
analysis of the social dynamics of language change among Detroit adolescents, Eckert focuses on
intersecting clusters of individuals engaged in such enter pri ses , (2 000: 1 71 -2 1 2) . Such clu s ters
co ns tit ut e gendered subgrou ps ins tant iating t he adolescent s octal categories which part ici pant s
the m selves cons t ruct. Networ k analys is t ypically does not attend t o the identi fication of such clus t er s
or the enter prises u ndertaken by me m bers , but deals primarily wit h the s truct ur al and co ntent
pro perties of t he ties which cons tit ut e egocentric personal networ ks .
the everyday problems which such personal communities help to solve (2000: 34). Furthermore,
individuals engage on a daily basis in a variety of endeavors in multiple personal communities. Eckert
em ploys the co nce pt of community of practice, whi ch is closely relat ed to that of social network, to
locate the int eract tonal s ites where social meaning i s indexed by Ii nguis ti c ele ment s, and Ii nguis t ic
change and social meaning are co-constructed. A community of practice may be defined as an
aggregate of people coming together around a particular enterprise (Eckert 2000: 35), and in her
analysis of the social dynamics of language change among Detroit adolescents, Eckert focuses on
intersecting clusters of individuals engaged in such enter pri ses , (2 000: 1 71 -2 1 2) . Such clu s ters
co ns tit ut e gendered subgrou ps ins tant iating t he adolescent s octal categories which part ici pant s
the m selves cons t ruct. Networ k analys is t ypically does not attend t o the identi fication of such clus t er s
or the enter prises u ndertaken by me m bers , but deals primarily wit h the s truct ur al and co ntent
pro perties of t he ties which cons tit ut e egocentric personal networ ks .
While close-knit networks vary
in their degree of approximation to the idealized representation shown
in figure 2 2 . 1 , networks of this type are the interactional site where localized styles and norms of all
kinds are constructed. Thus, for example, Eckert (2000: 210) comments that for the high school
st udent s s he s tudied , t he cons truct ion of local s t yles was "a fu nct ion of integration into local
networks and access to information. The importance of information is clear at the level of clothing
style." She goes on to note that "[c]ertain aspects of linguistic style are also negotiated consciously. I
can recall explicit discussions in my own high school crowd of _cool' ways to say things, generally in
the form of imit ations of cool peo ple . . . . But in general , lingui st ic influence t akes place wi thout explicit
co mment and all t he more requires direct access to speakers. The adoption of a way of speaking, like
a way of dressing, no doubt requires both access and entitlement to adopt the style of a particular
group" (Eckert 2000: 211). Eckert is here describing very general social mechanisms by which local
conventions and norms-of dress, religion, and general behavior, for example-are negotiated and
created, and linguistic norms are no exception. Close-knit networks of the kind where this activity
takes place are commonly contracted in adolescence. These are the linguistically influential peer
groups which are of such interest to sociolinguists attempting to understand the kinds of language
change as sociated wit h different point s i n the life s pan (see Ker swill 1 996) . However, such networks
al so flou rish i n low-s tatu s co mmu ni ties (both ru rat and urban) in t he absence of social and
geographical mobility and are important in fostering the solidarity ethos associated with the long-
term su rvival of socially di sfavored languages and dialect s .
in figure 2 2 . 1 , networks of this type are the interactional site where localized styles and norms of all
kinds are constructed. Thus, for example, Eckert (2000: 210) comments that for the high school
st udent s s he s tudied , t he cons truct ion of local s t yles was "a fu nct ion of integration into local
networks and access to information. The importance of information is clear at the level of clothing
style." She goes on to note that "[c]ertain aspects of linguistic style are also negotiated consciously. I
can recall explicit discussions in my own high school crowd of _cool' ways to say things, generally in
the form of imit ations of cool peo ple . . . . But in general , lingui st ic influence t akes place wi thout explicit
co mment and all t he more requires direct access to speakers. The adoption of a way of speaking, like
a way of dressing, no doubt requires both access and entitlement to adopt the style of a particular
group" (Eckert 2000: 211). Eckert is here describing very general social mechanisms by which local
conventions and norms-of dress, religion, and general behavior, for example-are negotiated and
created, and linguistic norms are no exception. Close-knit networks of the kind where this activity
takes place are commonly contracted in adolescence. These are the linguistically influential peer
groups which are of such interest to sociolinguists attempting to understand the kinds of language
change as sociated wit h different point s i n the life s pan (see Ker swill 1 996) . However, such networks
al so flou rish i n low-s tatu s co mmu ni ties (both ru rat and urban) in t he absence of social and
geographical mobility and are important in fostering the solidarity ethos associated with the long-
term su rvival of socially di sfavored languages and dialect s .
2 Social Networks and Language
Variation: Methods and Findings.
This sect ion
reviews so me variation! st st udies which have employed t he
networ k conce pt , and begins
by noti ng that the effect of interpersonal relationships on language choices has been explored for a
long ti me in sociolinguis tics ; wit nes s Gauchat's (1 905) account of variation in the vernacular of the tiny Swiss village of Charmey. Much later, Labov's (1972) sociometric analysis of the relationship
by noti ng that the effect of interpersonal relationships on language choices has been explored for a
long ti me in sociolinguis tics ; wit nes s Gauchat's (1 905) account of variation in the vernacular of the tiny Swiss village of Charmey. Much later, Labov's (1972) sociometric analysis of the relationship
between language use and the individual's position in the group
resembles in important respects
Eckert's account of communities of practice as the sites where linguistic norms and social meaning
are co-const ruct ed (see also Cheshire 1 9 82 for a co m parable accou nt of language variatio n in
adolescent peer groups) . Follow! ng an ethnographic , non-quantit at ive tradit ion of research which has
st rongly i nfluenced variation! st methods , Gu mperz's Discourse Strategies (1 9 82) provides an
extensive discus sion of t he effect s of changing networ k s tructu res on language choice in bilingual
co mmu nities . Chambers (1 995) identif ies a s tudy carried out i n Belfast , Northern Ireland , i n the mid-
1 9705 (Mitroy and Mitroy 1 978 , Mitroy 1 987a) as the firs t sys tematic accou nt of the relatio nshi p
between language variation and social network structure in the variationist literature. In this section, I
discuss some of the major methods and findings of this research, before reviewing three later studies
which take it as a point of departure.
Eckert's account of communities of practice as the sites where linguistic norms and social meaning
are co-const ruct ed (see also Cheshire 1 9 82 for a co m parable accou nt of language variatio n in
adolescent peer groups) . Follow! ng an ethnographic , non-quantit at ive tradit ion of research which has
st rongly i nfluenced variation! st methods , Gu mperz's Discourse Strategies (1 9 82) provides an
extensive discus sion of t he effect s of changing networ k s tructu res on language choice in bilingual
co mmu nities . Chambers (1 995) identif ies a s tudy carried out i n Belfast , Northern Ireland , i n the mid-
1 9705 (Mitroy and Mitroy 1 978 , Mitroy 1 987a) as the firs t sys tematic accou nt of the relatio nshi p
between language variation and social network structure in the variationist literature. In this section, I
discuss some of the major methods and findings of this research, before reviewing three later studies
which take it as a point of departure.
An ethnographically-oriented dat a-collect ion procedu re , influenced by
the wor k of John Gum perz
ment toned above , was used i n Belfast whereby t he researcher introduced herself to initial contacts as
a "friend of a friend"-i.e. a second order network contact of the people participating in the study.
These i nit ial cont act s pas sed her on t o ot hers , and observatio n and recording co nti nued u nt il
sufficient s peakers wit h the des ired characteris tic s (e . g. of age and gender) were sampled. Crucially,
the unit of study was the pre-existing social group, rather than a series of isolated individuals as
representatives of particular social categories. By attaching herself to this group and retreating to its
fringes as interactions between members progressed, she was able to obtain large amounts of
spontaneous speech as well as relevant social and demographic information, and the effect of the
observer on the data was lessened. Fieldwork procedures of this general type have been used
extensively in both bilingual and monolingual communities (as described by Mitroy et al. 1 995) , and
problems of access are rarely re ported . Ker swill (1 994) describes their im plementat ion in wes t ern
Norway.
ment toned above , was used i n Belfast whereby t he researcher introduced herself to initial contacts as
a "friend of a friend"-i.e. a second order network contact of the people participating in the study.
These i nit ial cont act s pas sed her on t o ot hers , and observatio n and recording co nti nued u nt il
sufficient s peakers wit h the des ired characteris tic s (e . g. of age and gender) were sampled. Crucially,
the unit of study was the pre-existing social group, rather than a series of isolated individuals as
representatives of particular social categories. By attaching herself to this group and retreating to its
fringes as interactions between members progressed, she was able to obtain large amounts of
spontaneous speech as well as relevant social and demographic information, and the effect of the
observer on the data was lessened. Fieldwork procedures of this general type have been used
extensively in both bilingual and monolingual communities (as described by Mitroy et al. 1 995) , and
problems of access are rarely re ported . Ker swill (1 994) describes their im plementat ion in wes t ern
Norway.
Data recorded in the Belfast study were examined to compare the language
patterns of 46 speakers
from three low status urban working-class communities-Ballymacarrett, Hammer, and Clonard. Eight
phonological variables, all of which were clearly indexical of the Belfast urban speech community,
were analyzed in relation to the network structure of individual speakers. In all three communities
networks were relatively dense, multiplex, and often kin-based, corresponding to those described by
many investigators as characteristic of traditional, long-established communities minimally impacted
by social or geographical mobili ty (see , e . g . , You ng and Wilmott 1 96 2, Cohen 1 982) .
from three low status urban working-class communities-Ballymacarrett, Hammer, and Clonard. Eight
phonological variables, all of which were clearly indexical of the Belfast urban speech community,
were analyzed in relation to the network structure of individual speakers. In all three communities
networks were relatively dense, multiplex, and often kin-based, corresponding to those described by
many investigators as characteristic of traditional, long-established communities minimally impacted
by social or geographical mobili ty (see , e . g . , You ng and Wilmott 1 96 2, Cohen 1 982) .
Although a social class index of the kind used in the early days of
socio-linguistics could not
discriminate between these speakers, the extent of individuals' use of vernacular variants was found
to be strongly influenced by level of integration into neighborhood networks. For example, some
people worked outside the neighborhood and had no local kin and few local ties of friendship, while
others were locally linked in all these capacities. Such differences in personal network structure clearly
spring from many complex social and psychological factors, and so interact with a number of other
variables; examples are generation cohort , t he recent his t ory of the neighborhood, and gender. Since
the gender / networ k relationshi p has proved to be suggestive of a general explanation of gendered
patterns of language variation and change, I shall comment on it here.
discriminate between these speakers, the extent of individuals' use of vernacular variants was found
to be strongly influenced by level of integration into neighborhood networks. For example, some
people worked outside the neighborhood and had no local kin and few local ties of friendship, while
others were locally linked in all these capacities. Such differences in personal network structure clearly
spring from many complex social and psychological factors, and so interact with a number of other
variables; examples are generation cohort , t he recent his t ory of the neighborhood, and gender. Since
the gender / networ k relationshi p has proved to be suggestive of a general explanation of gendered
patterns of language variation and change, I shall comment on it here.
Men in the Belfast neighborhoods generally contracted
denser and more multiplex localized network
ties than women, and network structure correlated with language use patterns differently for men and
women. A similar disjunction between the effect of male and female networks is reported in Dubois
and Horvath's (1998) variationist account of Cajun English. Eckert confirms and elaborates the
st rongly gendered character bot h of networ k clus ters and of t he networ k/ language relationshi p
(2000: 1 2 0- 4) . Noti ng a t endency for wo men t o contract ties acros s a wider social s pectrum,
Chambers (1 995: 1 2 4-8) attributes the frequently observed tendency of men to approximate more
closely than women to vernacular norms to this difference in network structure. In broader
investigations of the social trajectory of language change in two different urban locations, Mitroy and
Mitroy (1 99 3) and Docher ty et al. (1 997) ex plore the interaction between gender and network; male
nor ms are as sociated wit h localized variant s, and female norm s with su pra-local (bu t not neces sarily
st andardized) variant s. Mi troy (1 999) attributes the relative conservative , localized charact er of male
speech behavior to the particularly constraining effect of male peer networks. The interacting effect of
network and gender on patterns of language variation and change remains an extremely important research topic which will be alluded to again in this chapter.
ties than women, and network structure correlated with language use patterns differently for men and
women. A similar disjunction between the effect of male and female networks is reported in Dubois
and Horvath's (1998) variationist account of Cajun English. Eckert confirms and elaborates the
st rongly gendered character bot h of networ k clus ters and of t he networ k/ language relationshi p
(2000: 1 2 0- 4) . Noti ng a t endency for wo men t o contract ties acros s a wider social s pectrum,
Chambers (1 995: 1 2 4-8) attributes the frequently observed tendency of men to approximate more
closely than women to vernacular norms to this difference in network structure. In broader
investigations of the social trajectory of language change in two different urban locations, Mitroy and
Mitroy (1 99 3) and Docher ty et al. (1 997) ex plore the interaction between gender and network; male
nor ms are as sociated wit h localized variant s, and female norm s with su pra-local (bu t not neces sarily
st andardized) variant s. Mi troy (1 999) attributes the relative conservative , localized charact er of male
speech behavior to the particularly constraining effect of male peer networks. The interacting effect of
network and gender on patterns of language variation and change remains an extremely important research topic which will be alluded to again in this chapter.
2.1 Measuring social network
structure.
Given the ethnographic
orientation of social network analysis, a major challenge for
variationist
researchers is to devise a procedure for characterizing differences in network structure which reflects
the everyday social practices of speakers. The Belfast study developed a Network Strength Scale to
as sess s peakers' networ k characteris tics on five indicat ors of mutt i plexity and density. Mitroy (1 98 7a:
1 41 f) describes in det ail both the rationale for selecting the indicators and procedures for
operationalizing the Network Strength Scale. Briefly, a score of one or zero was assigned to each
indicator, and a total network strength score calculated for each individual which was the sum of
individual indicator scores. The indicators focused on various relationships within the nelhborhood
of kin, work, and friendship which had emerged in the course of the fieldwork as significant to
participants. They were
researchers is to devise a procedure for characterizing differences in network structure which reflects
the everyday social practices of speakers. The Belfast study developed a Network Strength Scale to
as sess s peakers' networ k characteris tics on five indicat ors of mutt i plexity and density. Mitroy (1 98 7a:
1 41 f) describes in det ail both the rationale for selecting the indicators and procedures for
operationalizing the Network Strength Scale. Briefly, a score of one or zero was assigned to each
indicator, and a total network strength score calculated for each individual which was the sum of
individual indicator scores. The indicators focused on various relationships within the nelhborhood
of kin, work, and friendship which had emerged in the course of the fieldwork as significant to
participants. They were
·membership of a high density,
territorially based group (e.g. a bingo or card-playing group,
a gang, or a football team or football supporters' club);
a gang, or a football team or football supporters' club);
·having kinship ties with more
than two households in the neighborhood;
·same workplace as at least two
others from the neighborhood;
·same workplace as at least two
others of the same gender from the neighborhood;
·voluntary association with workmates in leisure hours.
A series of statistical
analyses revealed a clear relationship between personal network structure and
phonological variation, usually complicated by the interaction of other
social variables such as the age
and gender of the speaker. The strongest vernacular speakers were generally those whose
and gender of the speaker. The strongest vernacular speakers were generally those whose
neig hbor hood networ k ties
were the st ronges t . Figu re 2 2 .3 illu st rates this t endency in Baily
macar rett, where
patterns of use for the variable (th) are plotted against network structure.
The variable pattern
represented here is the presence vs. absence of the
voiced interdental fricative [6] in intervocalic
contexts in such words as mother and brother; deletion scores for each speaker, whose age group
and gender are also specified, are plotted as percentages against individual network strength scores.
The interacting effects of gender and network noted above are also evident; not only are women's
network scores lower than men's, but the woman with the highest (th) index scores a full 10 percent
lower than the man who scores the lowest.
contexts in such words as mother and brother; deletion scores for each speaker, whose age group
and gender are also specified, are plotted as percentages against individual network strength scores.
The interacting effects of gender and network noted above are also evident; not only are women's
network scores lower than men's, but the woman with the highest (th) index scores a full 10 percent
lower than the man who scores the lowest.
It is on the basis of language/network
relationships such as these across several linguistic variables
and many more speakers that the close-knit network is interpreted as an important mechanism of
dialect maintenance.
and many more speakers that the close-knit network is interpreted as an important mechanism of
dialect maintenance.
2.2 Network concept in
small-scale communities: some examples.
The attractions for variationis t s of a networ k ap proach can be s tated
quit e briefly. Fir st , it provides a
set of procedures for studying small groups where speakers are not discriminable in terms of any
set of procedures for studying small groups where speakers are not discriminable in terms of any
kind of social class index-as
for example the eastern US island communities investigated by Wolfram
et al. (1999). Other examples are minority ethnic groups, migrants, rural populations, or populations
in nonindustrialized societies.
et al. (1999). Other examples are minority ethnic groups, migrants, rural populations, or populations
in nonindustrialized societies.
A second advantage of a social network approach is
that it is intrinsically a participant rather than an
analyst concept, and so has the potential to elucidate the social dynamics driving language variation
and change. Finally, network analysis offers a procedure for dealing with variation between individual
speakers, rather than between groups constructed with reference to predetermined social categories.
Eckert (2000: 1-33) discusses in some detail the very different concepts of a social variable implied
here . These methodolog ical i ssues are exem plified in many s tudies carried ou t in the 1 9 80s and
1 9905, such as t hat by Rus sell (1 982 ) in Mo mbasa, Kenya; Schm idt (1 9 85) of Aus tral ian Aboriginal
adolescent s ; Borto ni-Ricardo (1 9 8 5) of changes in t he language of r ural m ig rant s to a Brazil ian city ; V.
Edwards (1 9 86) of t he language of Britt sh black adolescent s ; School ing (1 990) of language dif ferences
among Melanes ians in New Caledonia; Lip pi-Green (1 9 8 9) o n dynam ic s of change in t he ru ral alpine
vi Ilage of Gros sdorf, Aus tria; W. Edwards (1 992) of variat ion in an African-Amer ican community in
inner-cit y Det roit ; and Maher (1 9 96) of t he pers is tence of language differences in the isolated island
co mmu ni ty of St. Bart hlemy, French Wes t Indies . Li ppi -Green (1 9 8 9), Edwards (1 992 ) and Borto ni-
Ricardo (1 9 8 5 ) are briefly reviewed below, to illustrate a range of different applications of the network
idea.
analyst concept, and so has the potential to elucidate the social dynamics driving language variation
and change. Finally, network analysis offers a procedure for dealing with variation between individual
speakers, rather than between groups constructed with reference to predetermined social categories.
Eckert (2000: 1-33) discusses in some detail the very different concepts of a social variable implied
here . These methodolog ical i ssues are exem plified in many s tudies carried ou t in the 1 9 80s and
1 9905, such as t hat by Rus sell (1 982 ) in Mo mbasa, Kenya; Schm idt (1 9 85) of Aus tral ian Aboriginal
adolescent s ; Borto ni-Ricardo (1 9 8 5) of changes in t he language of r ural m ig rant s to a Brazil ian city ; V.
Edwards (1 9 86) of t he language of Britt sh black adolescent s ; School ing (1 990) of language dif ferences
among Melanes ians in New Caledonia; Lip pi-Green (1 9 8 9) o n dynam ic s of change in t he ru ral alpine
vi Ilage of Gros sdorf, Aus tria; W. Edwards (1 992) of variat ion in an African-Amer ican community in
inner-cit y Det roit ; and Maher (1 9 96) of t he pers is tence of language differences in the isolated island
co mmu ni ty of St. Bart hlemy, French Wes t Indies . Li ppi -Green (1 9 8 9), Edwards (1 992 ) and Borto ni-
Ricardo (1 9 8 5 ) are briefly reviewed below, to illustrate a range of different applications of the network
idea.
The relative socioeconomic homogeneity of
the inner-city Detroit neighborhood studied by Walter
Edwards (1992) made social network analysis an attractive procedure for dealing with intra-
community linguistic variation. While the principal factor associated with choice of variant was age,
the most important factor which distinguished age-peers of a comparable social and educational
background was participation in neighborhood culture. Edwards interpreted such participation as
indicative of relative integration into local networks, and measured this integration by means of a
Vernacu lar Cu Itu re Index . Thi s was const ruct ed from res ponses to 1 0 s tat ement s which could range
from Strongly Disagree (1 point) to Strongly Agree (4 points). Five statements were designed as
indicators of the individual's physical integration into the neighborhood, and, like the Network
Strength Scale u sed i n Belfast , focused on localized int eract ions with ki n, wor kmates , and f fiends (e . g.
"Most of my relatives live in this neighborhood or with me"; "Most of my friends live in this
neighborhood"). Convinced of the importance of attitude in accounting for variation, Edwards
designed the other five statements to indicate evaluations of the neighborhood and of black/white
friendship ties (e.g. "I would like to remain living in this neighborhood"; "I do not have white friends
with whom I interact frequently").
Edwards (1992) made social network analysis an attractive procedure for dealing with intra-
community linguistic variation. While the principal factor associated with choice of variant was age,
the most important factor which distinguished age-peers of a comparable social and educational
background was participation in neighborhood culture. Edwards interpreted such participation as
indicative of relative integration into local networks, and measured this integration by means of a
Vernacu lar Cu Itu re Index . Thi s was const ruct ed from res ponses to 1 0 s tat ement s which could range
from Strongly Disagree (1 point) to Strongly Agree (4 points). Five statements were designed as
indicators of the individual's physical integration into the neighborhood, and, like the Network
Strength Scale u sed i n Belfast , focused on localized int eract ions with ki n, wor kmates , and f fiends (e . g.
"Most of my relatives live in this neighborhood or with me"; "Most of my friends live in this
neighborhood"). Convinced of the importance of attitude in accounting for variation, Edwards
designed the other five statements to indicate evaluations of the neighborhood and of black/white
friendship ties (e.g. "I would like to remain living in this neighborhood"; "I do not have white friends
with whom I interact frequently").
Yet another set of indicator s
was relevant to Li ppi-Green's (1 989) s t udy of language change in
progress in Grossdorf, an isolated Austrian Alpine village with 800 inhabitants. Noting the
unhelpfulnes s of macro-level conce pt s such as clas s in u ncovering the relationshi p between language
variation and social structure, Lippi-Green examined in considerable detail the personal network
structures of individuals, constructing a scale which used 16 differentially weighted indicators. Some
of these were associated with the familiar domains of work, kin, and friendship, while others dealt
more s pecifically wit h local condit ions-such as the nu mber of grand parent s familiar t o the speaker
who was a core member of the village, or the involvement of the speaker's employment with the tourism industry. Particularly important were indicators which linked speakers to major village family
progress in Grossdorf, an isolated Austrian Alpine village with 800 inhabitants. Noting the
unhelpfulnes s of macro-level conce pt s such as clas s in u ncovering the relationshi p between language
variation and social structure, Lippi-Green examined in considerable detail the personal network
structures of individuals, constructing a scale which used 16 differentially weighted indicators. Some
of these were associated with the familiar domains of work, kin, and friendship, while others dealt
more s pecifically wit h local condit ions-such as the nu mber of grand parent s familiar t o the speaker
who was a core member of the village, or the involvement of the speaker's employment with the tourism industry. Particularly important were indicators which linked speakers to major village family
networks. Overall, the best correlate of linguistic
behavior (conservative vs. innovatory) was
integration into three important networks, including those which involved workplace and exposure to
non-local language varieties. However, the subtlety of Lippi-Green's network measurement scale
allowed her to examine correlations both with all of it and some part s of it , revealing for example
gender- s pecific social t raject ories of language change and variation.
integration into three important networks, including those which involved workplace and exposure to
non-local language varieties. However, the subtlety of Lippi-Green's network measurement scale
allowed her to examine correlations both with all of it and some part s of it , revealing for example
gender- s pecific social t raject ories of language change and variation.
In addressing the changing language behavior of mobile
individuals , Bo rtoni-Ricardo's (1 9 85) accou nt
of the sociolinguis t ic adjus t ment of rur al mig rant s t o Br azlandia, a satellite cit y of Brasilia exem plifies
a very different ap plication of the networ k conce pt fro m those di scu s sed above . Again , the social
clas s conce pt is not particularly u seful i n this cont ext, since it does not discriminate between the
individuals studied by Bortoni-Ricardo, all of whom were relatively poor. Taking the group's own
linguistic norms as a starting point, Bortoni-Ricardo examined the extent to which speakers had
moved away from thei r st ig matized Cal pi ra dialect , rat her than att em pting t o ident ify a linguis tic
st andard "target"
of the sociolinguis t ic adjus t ment of rur al mig rant s t o Br azlandia, a satellite cit y of Brasilia exem plifies
a very different ap plication of the networ k conce pt fro m those di scu s sed above . Again , the social
clas s conce pt is not particularly u seful i n this cont ext, since it does not discriminate between the
individuals studied by Bortoni-Ricardo, all of whom were relatively poor. Taking the group's own
linguistic norms as a starting point, Bortoni-Ricardo examined the extent to which speakers had
moved away from thei r st ig matized Cal pi ra dialect , rat her than att em pting t o ident ify a linguis tic
st andard "target"
Bortoni-Ricardo's main
hypothesis is that the change in social structure associated with rural to
urban
migration involves a move from an "insulated" network consisting largely of kinsfolk and neighbors to
an "integrated" urban network where links are less multiplex but contracted in a wider range of social
migration involves a move from an "insulated" network consisting largely of kinsfolk and neighbors to
an "integrated" urban network where links are less multiplex but contracted in a wider range of social
co nt
ext s . The linguis tic count er part of t his change is increasi ng dialect
dif fuse-Mes s-a movement
away f rom t he relat ively focu sed norm s of the Cal pi ra dialect ( see fu rther Le Page and Tabouret - Keller
1 9 8 5) . Two se parate networ k i ndices are cons tructed to measu re the changing patterns of the
away f rom t he relat ively focu sed norm s of the Cal pi ra dialect ( see fu rther Le Page and Tabouret - Keller
1 9 8 5) . Two se parate networ k i ndices are cons tructed to measu re the changing patterns of the
mig rant s' s ocial relat
ionshi ps ; t he integration index and the urbanization
index. The integration index
assesses relevant characteristics of the three persons with whom each migrant most frequently
assesses relevant characteristics of the three persons with whom each migrant most frequently
interacts-for
example, whether they are kin or non-kin, whether ties were contracted prior to
migration. The final score measures progress in the transition from an insulated to an integrated type
of network-effectively the gradual loosening of close-knit network ties. These changes are correlated
migration. The final score measures progress in the transition from an insulated to an integrated type
of network-effectively the gradual loosening of close-knit network ties. These changes are correlated
with a linguis tic movement away from t he norms of the Cal pi ra dialect .
The urbanization index focu
ses not on the migrant , but the characteris t ic s of members of his or her
personal network, such as
educational level and mobility; indicators are selected to assess the extent
to which the migrant's contacts are integrated into urban life. In developing these two quite different
types of index Bortoni-Ricardo extends the application of the network concept beyond an analysis of
small, close-knit groups of the kind described so far to consider the extent to which individuals have
detached themselves from such groups and the linguistic consequences of that detachment.
to which the migrant's contacts are integrated into urban life. In developing these two quite different
types of index Bortoni-Ricardo extends the application of the network concept beyond an analysis of
small, close-knit groups of the kind described so far to consider the extent to which individuals have
detached themselves from such groups and the linguistic consequences of that detachment.
In an extended di scu ssion of
sociolinguis t s' use of the networ k conce pt, Mu rray (1 993) is
particularly
cr itical of the quantit ative analysis employed by Labov (1 972) and Milroy (1 987a) on grounds which
are more controversial than he claims, and are moreover not always entirely clear. Murray's strongly
stated claims of what constitutes an appropriate approach to social network analysis and an
appropriate statistical method are disputable. Moreover, a subsequent clarification in American
Speech (Bu tter s 1 995 : 2 0) point s out that s pecific crit icis ms of Milroy's s tat is tical procedures and
result s res t on a mis reading of her text.
cr itical of the quantit ative analysis employed by Labov (1 972) and Milroy (1 987a) on grounds which
are more controversial than he claims, and are moreover not always entirely clear. Murray's strongly
stated claims of what constitutes an appropriate approach to social network analysis and an
appropriate statistical method are disputable. Moreover, a subsequent clarification in American
Speech (Bu tter s 1 995 : 2 0) point s out that s pecific crit icis ms of Milroy's s tat is tical procedures and
result s res t on a mis reading of her text.
3 Language Maintenance and
Shift in Bilingual Communities.
Although the discus s ion so far has concentrat
ed on the language / networ k relationshi p in monolingual
co mmu nities , researchers inves ti gati ng the social mechanis ms of language mai nt enance and shift in
bilingual co mmu nities have em ployed a variant of the same general principle: networks constituted
chiefly of strong ties function as a mechanism to support minority languages, resisting institutional
pressures to language shift, but when these networks weaken, language shift is likely to take place.
This section reviews some of this work, starting with a consideration of the network structure
characteristic of immigrant communities.
co mmu nities , researchers inves ti gati ng the social mechanis ms of language mai nt enance and shift in
bilingual co mmu nities have em ployed a variant of the same general principle: networks constituted
chiefly of strong ties function as a mechanism to support minority languages, resisting institutional
pressures to language shift, but when these networks weaken, language shift is likely to take place.
This section reviews some of this work, starting with a consideration of the network structure
characteristic of immigrant communities.
It has sometimes been
suggested that close-knit networks such as those studied in Belfast and
Detroit are marginal to contemporary urban life; for example, there is a large sociological literature on
"the stranger", the mobile, marginal individual who is often seen as typical of a modern city dweller
(Harman 1 988) . While this perce ption cert ainly reflect s important as pect s of contemporary urban life
(we shall later discuss the linguistic consequences of social and geographical mobility) it does not tellthe whole story. Certainly traditional working-class communities like the Italian American "urban
Detroit are marginal to contemporary urban life; for example, there is a large sociological literature on
"the stranger", the mobile, marginal individual who is often seen as typical of a modern city dweller
(Harman 1 988) . While this perce ption cert ainly reflect s important as pect s of contemporary urban life
(we shall later discuss the linguistic consequences of social and geographical mobility) it does not tellthe whole story. Certainly traditional working-class communities like the Italian American "urban
villagers" described by Cans (1 96 2) or t he close- knit Yorkshire
mini ng communities described by
Dennis et at. (1 95 7) have all but disappeared . However , Giddens (1 989) point s out that
neighbor hoods involving close kinship and personal ties seem still to be created rather than
discouraged by city life, since those who form part of urban ethnic communities gravitate to form ties
with, and often to live with, others from similar linguistic or ethnic backgrounds. Hence, the older
style of close-knit working class community is apparently being replaced in industrialized countries
by similar types of community created by newer immigrants. Dabne and Moore (1995) describe the
supportive function of such migrant networks during the period when immigrants are developing
resources to integrate more fully into urban life.
Dennis et at. (1 95 7) have all but disappeared . However , Giddens (1 989) point s out that
neighbor hoods involving close kinship and personal ties seem still to be created rather than
discouraged by city life, since those who form part of urban ethnic communities gravitate to form ties
with, and often to live with, others from similar linguistic or ethnic backgrounds. Hence, the older
style of close-knit working class community is apparently being replaced in industrialized countries
by similar types of community created by newer immigrants. Dabne and Moore (1995) describe the
supportive function of such migrant networks during the period when immigrants are developing
resources to integrate more fully into urban life.
The type of close-knit network structure which seems to help maintain
community languages is
therefore by no means a residue of an earlier type of social organization; not only immigrants, but
also long-term stigmatized and marginalized minorities, like the New York Puerto Ricans studied by
Zentella (1997), construct personal communities which function as powerful support systems in a
hostile environment . Gal (1 978) and Li (1 994) , whose work i s discus sed i n more det ail below, have
correlated observed patt ems of language u se with s pecific networ k patterns in much the same way as
researchers working in monolingual communities. Indeed, Gal explicitly compares her model of
language shift to a variationist model of language change, in being both gradual and rooted in
synchronic patterns of variation in language use. Zentella also adopts a broad variationist perspective,
but like Gum perz (1 982) , u ses the conce pt of network informally and non-quantitatively.
therefore by no means a residue of an earlier type of social organization; not only immigrants, but
also long-term stigmatized and marginalized minorities, like the New York Puerto Ricans studied by
Zentella (1997), construct personal communities which function as powerful support systems in a
hostile environment . Gal (1 978) and Li (1 994) , whose work i s discus sed i n more det ail below, have
correlated observed patt ems of language u se with s pecific networ k patterns in much the same way as
researchers working in monolingual communities. Indeed, Gal explicitly compares her model of
language shift to a variationist model of language change, in being both gradual and rooted in
synchronic patterns of variation in language use. Zentella also adopts a broad variationist perspective,
but like Gum perz (1 982) , u ses the conce pt of network informally and non-quantitatively.
Gumperz's (1 982) accou nt of the Slovenian / German bilingual co mmu ni ty
in a remote part of Austria's
Gail Valley associates the move towards monolingualism with economic changes. Members of this
poor and socially stigmatized farming community had traditionally been embedded in classic close-
knit networks of mutual support which linked them in many capacities-as co-workers, neighbors, and
friends who socialized together within the boundaries of their community. However, such behaviors
changed as the economy shifted from a dependence on subsistence farming to a primarily service
economy. Improvement s in the road sys tern gave rise to a host of other changes which affected
network structure and, ultimately, language behavior followed. Farmers sold produce to incomers and
to factories rather than dealing with other local farmers; farm buildings were converted into tourist
accommodation for the many visitors entering the area; work and leisure activities were no longer
confined to the immediate locality. As many day to day interactions came to be with urban outsiders,
villagers lost their reliance on the local support network. Although of course local conditions give rise
to variations, the pattern Gumperz describes here appears to be a very general one in much of
western Europe (and probably elsewhere in t he developed and develo ping world) . 0' Riagain_s (1 997)
description of a series of st udies car ried ou t in Ireland between 1 973 and 1 99 3 sugges t s a sit uation
broadly si milar t o that in t he Gail Valley where change to a service economy triggers associated
change in personal social network structures. Consequent changes in the categories of individual
involved in face-to-face encounters shifted the balance from bilingual Irish-speaking insiders to
monolingual English-speaking outsiders, inevitably resulting in the further decline of Irish.
Gail Valley associates the move towards monolingualism with economic changes. Members of this
poor and socially stigmatized farming community had traditionally been embedded in classic close-
knit networks of mutual support which linked them in many capacities-as co-workers, neighbors, and
friends who socialized together within the boundaries of their community. However, such behaviors
changed as the economy shifted from a dependence on subsistence farming to a primarily service
economy. Improvement s in the road sys tern gave rise to a host of other changes which affected
network structure and, ultimately, language behavior followed. Farmers sold produce to incomers and
to factories rather than dealing with other local farmers; farm buildings were converted into tourist
accommodation for the many visitors entering the area; work and leisure activities were no longer
confined to the immediate locality. As many day to day interactions came to be with urban outsiders,
villagers lost their reliance on the local support network. Although of course local conditions give rise
to variations, the pattern Gumperz describes here appears to be a very general one in much of
western Europe (and probably elsewhere in t he developed and develo ping world) . 0' Riagain_s (1 997)
description of a series of st udies car ried ou t in Ireland between 1 973 and 1 99 3 sugges t s a sit uation
broadly si milar t o that in t he Gail Valley where change to a service economy triggers associated
change in personal social network structures. Consequent changes in the categories of individual
involved in face-to-face encounters shifted the balance from bilingual Irish-speaking insiders to
monolingual English-speaking outsiders, inevitably resulting in the further decline of Irish.
Gal's (1 978) analysis of language shift i n the bilingual German / Hu
ngarian com mu nit y in Oberwart,
Aus tr ia, identif ies Sim ilar tr iggers . Individuals are measu red in t er ms of the relative " peasant nes s" (a
local social cat egory) of t heir networ ks . This variable o perat es dif ferent ly for men and wo men ( recall
thi s pattern in other networ k s tudies) bu t is fou nd to correlat e more closely t han individual peasant
st at us wit h patterns of language choice . Like t he Gail Valley and the Irish Gaeltacht, Oberwart had
been bilingual for several centuries, and again, changes in network structure are associated with
higher -level economic changes . We now t urn to Li's (1 994) and Zentella's (1 997) wor k in i mmigrant
co mmu nities where t ypically pressu re to as similate to t he monolingual nor m of t he hos t cou nt ry is
int ense and , i n contras t t o the co mmu nities discu s sed above , a pattern of language shift over three
generat ions is co mmon. Grosjean (1 982) and Jorgensen (1 998) describe these pres su res in t he Unit ed
States and Euro pe res pectively.
Aus tr ia, identif ies Sim ilar tr iggers . Individuals are measu red in t er ms of the relative " peasant nes s" (a
local social cat egory) of t heir networ ks . This variable o perat es dif ferent ly for men and wo men ( recall
thi s pattern in other networ k s tudies) bu t is fou nd to correlat e more closely t han individual peasant
st at us wit h patterns of language choice . Like t he Gail Valley and the Irish Gaeltacht, Oberwart had
been bilingual for several centuries, and again, changes in network structure are associated with
higher -level economic changes . We now t urn to Li's (1 994) and Zentella's (1 997) wor k in i mmigrant
co mmu nities where t ypically pressu re to as similate to t he monolingual nor m of t he hos t cou nt ry is
int ense and , i n contras t t o the co mmu nities discu s sed above , a pattern of language shift over three
generat ions is co mmon. Grosjean (1 982) and Jorgensen (1 998) describe these pres su res in t he Unit ed
States and Euro pe res pectively.
Despite a general sense that Spanish in the United States is resistant
to shift (Bourhis and Marshall
1999) Zentella provides evidence of this three-generational shift pattern in a Puerto Rican community
in New York City (et b/oque). In an account which is compiled from long-term participant oservation,
she notes some characteristic sociolinguistic patterns. First, while choice of code is heavily network-
dependent, several distinguishable varieties of both Spanish and English give rise to multiple-code repertoires (Zentella 1 997: 4 8) . For example, while you ngs ters and you ng mother s have access t o a
1999) Zentella provides evidence of this three-generational shift pattern in a Puerto Rican community
in New York City (et b/oque). In an account which is compiled from long-term participant oservation,
she notes some characteristic sociolinguistic patterns. First, while choice of code is heavily network-
dependent, several distinguishable varieties of both Spanish and English give rise to multiple-code repertoires (Zentella 1 997: 4 8) . For example, while you ngs ters and you ng mother s have access t o a
range of Spani sh and Engl ish codes , t hey favor a Puerto Rican variety
of English. Older men and
women however prefer Puerto Rican Spanish, while "young dudes" favor African-American Vernacular
English but also have access to varieties of Spanish and English. Many of the children speak very little
Spanish, mixing Spanish into their English to produce the code popularly described as Nuyorican;
interestingly, only one child in Zentella's sample is monolingual in English. Patterns of code choice
from this multidimensional repertoire are thus systematically associated with a range of distinctive
gender- and age-related networks.
women however prefer Puerto Rican Spanish, while "young dudes" favor African-American Vernacular
English but also have access to varieties of Spanish and English. Many of the children speak very little
Spanish, mixing Spanish into their English to produce the code popularly described as Nuyorican;
interestingly, only one child in Zentella's sample is monolingual in English. Patterns of code choice
from this multidimensional repertoire are thus systematically associated with a range of distinctive
gender- and age-related networks.
Zentella comments on the significance of what she describes as
"the Puerto Rico language learning
connection" in offering an explanation of the strength and persistence of Spanish in New York City
and elsewhere in the United States. She cites the combined effect of continuing network ties of
immigrants to individuals in adjacent Spanish-speaking countries of Latin America, and cyclic
patt ems of immi grat ion (see agai n Bourhis and Marshal I 1 999) . Thi s Puerto Rican connect ion (and it s
counter part in ot her S Danish- speaki ng co mmu nities in t he United St ates) may explain why young
peo ple u se a mixed S Danish-Engl ish code; in accordance with the expected pattern of language shift
in immigrant communities, they have shifted substantially to English mono-lingualism but still need
to communicate with Spanish monolingual speakers.
connection" in offering an explanation of the strength and persistence of Spanish in New York City
and elsewhere in the United States. She cites the combined effect of continuing network ties of
immigrants to individuals in adjacent Spanish-speaking countries of Latin America, and cyclic
patt ems of immi grat ion (see agai n Bourhis and Marshal I 1 999) . Thi s Puerto Rican connect ion (and it s
counter part in ot her S Danish- speaki ng co mmu nities in t he United St ates) may explain why young
peo ple u se a mixed S Danish-Engl ish code; in accordance with the expected pattern of language shift
in immigrant communities, they have shifted substantially to English mono-lingualism but still need
to communicate with Spanish monolingual speakers.
Li (1 994) and Milroy and Li (1 995) re port an i nvestigation of
social traject ories of language shift which
as sociates different network t ypes with variable patterns of language use. However, they describe a
much less complex community repertoire than Zentella, providing a quantitative analysis of both
network types and language patterns. Three migrant groups are distinguished, overlapping with (but
not exactly corres pondi ng to) a grand parent , parent, and child generation . Each grou p co ntract s
charact eristically di fferent types of networ k ties, t he fir st as sociating mainly with ki n, the second
chiefly with other British Chinese, and the third more extensively with non-Chinese peers. Variable
network patterns were in turn correlated with seven different patterns of language choice, where
English and Chinese were used either monolingually or in different combinations. Following Milardo
(1988), interactive and exchange networks were distinguished, corresponding roughly to "weak" and
"strong" types of tie.
as sociates different network t ypes with variable patterns of language use. However, they describe a
much less complex community repertoire than Zentella, providing a quantitative analysis of both
network types and language patterns. Three migrant groups are distinguished, overlapping with (but
not exactly corres pondi ng to) a grand parent , parent, and child generation . Each grou p co ntract s
charact eristically di fferent types of networ k ties, t he fir st as sociating mainly with ki n, the second
chiefly with other British Chinese, and the third more extensively with non-Chinese peers. Variable
network patterns were in turn correlated with seven different patterns of language choice, where
English and Chinese were used either monolingually or in different combinations. Following Milardo
(1988), interactive and exchange networks were distinguished, corresponding roughly to "weak" and
"strong" types of tie.
Since the Chinese in Tyneside did not live within a specificable
neighborhood, assessments of network
strength could not be based on the territorially restricted strong ties as in several of the studies
reviewed in the previous section. Instead comparative analysis of individual exchange networks was
based on a list of up to 20 persons who constituted significant and regular contacts for each
individual, adapting the procedure described by Mitchell (1986). These sets of 20 could then be
compared on relevant dimensions-for example, different ethnic compositions. Not surprisingly, the
strongest ethnic networks were associated both with the oldest generation and with the most
extensive use of Chinese, and the weakest with the 8ritish-born generation and with the most
extensive use of English. There were, however, many subtleties associated with different network
patterns within each group; particularly interesting is the role of the TrueJesus Church, one of the
community's institutions.
strength could not be based on the territorially restricted strong ties as in several of the studies
reviewed in the previous section. Instead comparative analysis of individual exchange networks was
based on a list of up to 20 persons who constituted significant and regular contacts for each
individual, adapting the procedure described by Mitchell (1986). These sets of 20 could then be
compared on relevant dimensions-for example, different ethnic compositions. Not surprisingly, the
strongest ethnic networks were associated both with the oldest generation and with the most
extensive use of Chinese, and the weakest with the 8ritish-born generation and with the most
extensive use of English. There were, however, many subtleties associated with different network
patterns within each group; particularly interesting is the role of the TrueJesus Church, one of the
community's institutions.
Li (1 995) sugges t s t he raison d't re of this
church to be a support mechanism for cultural and
socialization activities rather than a religious institution, noting that member families were distinctive
in having contracted pre-migration network ties on the island of Ap Chau, close to Hong Kong. He
further documents a very much stronger pattern of Chinese language maintenance amongst the
young Brit ish- born members of the True Jesu s Church t han among s t the you ng co mmunity as a
whole , a patt ern att ribut ed t o the st rong t ies maint ai ned by True Jesu s young st ers with church
members monoli ngual in Cant onese . Interestingly, Li also notes a pattern of fluent Cant onese / English
code m ixing as character! st ic of the True Jesus teenagers , which he explains in much the same way as
Zentel la (1 997) explains wides Dread S panish/ Engli sh mix ing by New Yor k Cit y Puert o Rican
young st ers . In both cases the young people have come up with a similar solution to a similar problem;
proficiency in the community language is limited, but they maintain network ties which require them
to communicate with non-English speakers. Thus, a network analysis can help explain not only the
social trajectory of language shift, but specific patterns of code switching (see also Labrie's (1988)
network-based account of code switching by It al tans in Montreal) 4 Weak Ties and Theories of Language Change.
socialization activities rather than a religious institution, noting that member families were distinctive
in having contracted pre-migration network ties on the island of Ap Chau, close to Hong Kong. He
further documents a very much stronger pattern of Chinese language maintenance amongst the
young Brit ish- born members of the True Jesu s Church t han among s t the you ng co mmunity as a
whole , a patt ern att ribut ed t o the st rong t ies maint ai ned by True Jesu s young st ers with church
members monoli ngual in Cant onese . Interestingly, Li also notes a pattern of fluent Cant onese / English
code m ixing as character! st ic of the True Jesus teenagers , which he explains in much the same way as
Zentel la (1 997) explains wides Dread S panish/ Engli sh mix ing by New Yor k Cit y Puert o Rican
young st ers . In both cases the young people have come up with a similar solution to a similar problem;
proficiency in the community language is limited, but they maintain network ties which require them
to communicate with non-English speakers. Thus, a network analysis can help explain not only the
social trajectory of language shift, but specific patterns of code switching (see also Labrie's (1988)
network-based account of code switching by It al tans in Montreal) 4 Weak Ties and Theories of Language Change.
Social network analysis has most commonly been employed in communities
where ties between
speakers are generally strong. While studies such as those reviewed above show that it is relatively
straightforward to operationalize the network concept in this way, how to handle socially and
geographically mobile speakers whose personal network ties are not predominantly dense or
multiplex i s much les s obviou s . I n fact, networ k- based accou nt s of such s peakers are rare , and the
only st udy exami ned so far which has attempted anything like this is Bortoni-Ricardo's (1985) account
of the progressive urbanization patterns of Brazilian rural migrants. Geographical and social mobility
is, however, the rule rather than the exception in contemporary cities, and an increasing amount of
work carried out by variationists within dialect contact frameworks focuses on such speakers (see for
exam ple Tr udgill 1 9 86, Trudg ill and Brit ai n forthcoming , Chambers 1 995: 5 2 -6 5 , Kerswill and
Will tarns 2000) . Chambers (1 992) po int s out t hat sociolingu is tic s as well as dialectology is quite
generally oriented to non-mobile speakers in isolated communities; the focus of network studies over
two decades has reflected this orientation.
speakers are generally strong. While studies such as those reviewed above show that it is relatively
straightforward to operationalize the network concept in this way, how to handle socially and
geographically mobile speakers whose personal network ties are not predominantly dense or
multiplex i s much les s obviou s . I n fact, networ k- based accou nt s of such s peakers are rare , and the
only st udy exami ned so far which has attempted anything like this is Bortoni-Ricardo's (1985) account
of the progressive urbanization patterns of Brazilian rural migrants. Geographical and social mobility
is, however, the rule rather than the exception in contemporary cities, and an increasing amount of
work carried out by variationists within dialect contact frameworks focuses on such speakers (see for
exam ple Tr udgill 1 9 86, Trudg ill and Brit ai n forthcoming , Chambers 1 995: 5 2 -6 5 , Kerswill and
Will tarns 2000) . Chambers (1 992) po int s out t hat sociolingu is tic s as well as dialectology is quite
generally oriented to non-mobile speakers in isolated communities; the focus of network studies over
two decades has reflected this orientation.
At a purely operational level, loose-knit networks are hard to work with.
Analysis of close-knit
networks involves comparing speakers who differ from each other in certain respects (for example,
multiplexity of ties contracted at the workplace) but are still similar enough in other relevant ways for
a comparison to be meaningful. But it is difficult to see how the loose-knit network structures of
individuals who differ from each other in many different respects (educational level, occupation,
region of origin, mobility inter alia) might meaningfully be compared with each other. This problem
was noted in the Be|fas t suburbs of Andersonst own and Braniel (Milroy 1 98 7b: I 0 8) and was
encou ntered also in an att empt to ap ply social networ k analyst s i n the prosperous Bert in subu rb of
Zehlendorf (Ditt mar and Schlobins ki 1 98 8) . However, from t he pers pective of a per son who has
changed em ployment and place of residence several times, the networks of speakers studied in
Belfast, Detroit, and Grossdorf are all close-knit , and m ight be compared in a general way with those
of more mobile speakers (see Kerswill and Williams 1 999) .
networks involves comparing speakers who differ from each other in certain respects (for example,
multiplexity of ties contracted at the workplace) but are still similar enough in other relevant ways for
a comparison to be meaningful. But it is difficult to see how the loose-knit network structures of
individuals who differ from each other in many different respects (educational level, occupation,
region of origin, mobility inter alia) might meaningfully be compared with each other. This problem
was noted in the Be|fas t suburbs of Andersonst own and Braniel (Milroy 1 98 7b: I 0 8) and was
encou ntered also in an att empt to ap ply social networ k analyst s i n the prosperous Bert in subu rb of
Zehlendorf (Ditt mar and Schlobins ki 1 98 8) . However, from t he pers pective of a per son who has
changed em ployment and place of residence several times, the networks of speakers studied in
Belfast, Detroit, and Grossdorf are all close-knit , and m ight be compared in a general way with those
of more mobile speakers (see Kerswill and Williams 1 999) .
To identify these
operational difficulties is not to suggest
that loose-knit networks are uninteresting
to the variationist: quite the contrary. For if a close-knit network structure supports localized
lingu is tic norms and res is t s change ori ginating f rom out side the network, the corollary, that
communities composed of weak ties will be susceptible to such change, is also likely to hold.
Follow! ng Granovetter's (1 973) argu ment that "weak" and ap parently i ns ig nif icant inter personal ties
(of "acquaintance" as op posed to "friend" , for example) are im portant channels through which
innovation and influence flow from one close-knit group to another Milroy and Milroy (1 985 ) have
pro posed t hat Ii nguis t ic i nnovators are I ikely to be i ndividual s who are in a position to contract many
weak ties. Since such weak ties link close-knit groups to each other and to the larger regional or
national speech community, they are likely to figure prominently in a socially accountable theory of
linguistic diffusion and change.
to the variationist: quite the contrary. For if a close-knit network structure supports localized
lingu is tic norms and res is t s change ori ginating f rom out side the network, the corollary, that
communities composed of weak ties will be susceptible to such change, is also likely to hold.
Follow! ng Granovetter's (1 973) argu ment that "weak" and ap parently i ns ig nif icant inter personal ties
(of "acquaintance" as op posed to "friend" , for example) are im portant channels through which
innovation and influence flow from one close-knit group to another Milroy and Milroy (1 985 ) have
pro posed t hat Ii nguis t ic i nnovators are I ikely to be i ndividual s who are in a position to contract many
weak ties. Since such weak ties link close-knit groups to each other and to the larger regional or
national speech community, they are likely to figure prominently in a socially accountable theory of
linguistic diffusion and change.
Milroy and Mitroy (1985) argue that a "weak tie" model of change
can account rather generally for the
tendency of some languages to be more resistant to change than others (Icelandic vs. English, or
Sardinian vs. Sicilian, for example). They suggest that a type of social organization based on
overlapping close-knit networks will inhibit change, while one characterized by mobility (for whatever
reason), with a concomitant weakening of close ties, will facilit at e it. Grace (1 992 ) explains in a s imilar
way so me puzzling development s among t he Au st rones ian languages , whi ch show widely di ffering
patt ems of su sce pti bility t o change inexplicable in t erms of tradit tonal assu m ptions (see also Grace
1 990) . As well as ex plat ning dif ferent large -scale Ii ngui st ic out comes by co m paring different t ypes of
social organizat ion , the weak tie model can account for specific problematic examples of change, of
which two are considered below.
tendency of some languages to be more resistant to change than others (Icelandic vs. English, or
Sardinian vs. Sicilian, for example). They suggest that a type of social organization based on
overlapping close-knit networks will inhibit change, while one characterized by mobility (for whatever
reason), with a concomitant weakening of close ties, will facilit at e it. Grace (1 992 ) explains in a s imilar
way so me puzzling development s among t he Au st rones ian languages , whi ch show widely di ffering
patt ems of su sce pti bility t o change inexplicable in t erms of tradit tonal assu m ptions (see also Grace
1 990) . As well as ex plat ning dif ferent large -scale Ii ngui st ic out comes by co m paring different t ypes of
social organizat ion , the weak tie model can account for specific problematic examples of change, of
which two are considered below.
Innovations have been widely observed to skip from
city to city, bypassing intervening territory. This
appears to be the pattern of the Northern Cities Shift, a vigorous change in vowel systems affecting
cities of the northern USA from western New England to an unspecified point westward (Labov 1991,
Wolfram and Schilling-Estes 1 998: 1 3 8) . On the ot her side of t he At|ant ic, Trudgill (1 988) notes t he
relatively recent adoption by young speakers in Norwich of a merger between / f / / q/ and / v/
/ 6/ (as in fin thin," lava lather) . Milroy (1 996) subsequently documents thi s merger by young
working-class speaker s in the northern English cities of Sheffield and Derby, and Stuart-Smith (1999) reports its recent appearance yet further north in Glasgow. As it saliently indexes working-class
appears to be the pattern of the Northern Cities Shift, a vigorous change in vowel systems affecting
cities of the northern USA from western New England to an unspecified point westward (Labov 1991,
Wolfram and Schilling-Estes 1 998: 1 3 8) . On the ot her side of t he At|ant ic, Trudgill (1 988) notes t he
relatively recent adoption by young speakers in Norwich of a merger between / f / / q/ and / v/
/ 6/ (as in fin thin," lava lather) . Milroy (1 996) subsequently documents thi s merger by young
working-class speaker s in the northern English cities of Sheffield and Derby, and Stuart-Smith (1999) reports its recent appearance yet further north in Glasgow. As it saliently indexes working-class
London speech, this change, in the British context,
appears to be contact-induced rather than to
originate from within communities. While attributing
its rapid spread to NonNich to greater mobility
and contact between speakers, Trudgill points out that
the teenagers who use the merged variants are
less mobile than their seniors and tend to contract
close ties locally. For this reason, it is hard to
explain the precise mechanisms of diffusion in terms
of close contact between London and Norwich
speakers. The same objection can be made even more
tellingly since this very vigorous change has
extended across large areas of urban Britain in the
years since Trudgill's observations of its spread to
Norwich. A weak tie model would however predict its
diffusion from one community to another
through multiple weak ties; in fact, Trudgill suggests
tourists and football supporters as individuals
who might be implicated in the diffusion process
(1986: 54).
A second problematic case of change to which a
weak-tie model offers a principled solution involves
the alternating phonolexical variable (u), as in pull,
pus/7, foot, discussed in detail by, for example,
Milroy (1987a) and Milroy (1992). A rounded and an
unrounded vowel variant appear in a lexically
limited set, the unrounded variant strongly indexing
working-class identity. Figure 22.4 provides clear evidence of cross-community
consensus between young (but not middle-
aged) speakers on the use of these alternative
realizations to index gender also in the three working-
class communities of Ballymacarrett, Hammer, and
Clonard. However, the fact that middle-aged
Ballymacarrett speakers use (u) to index gender in
much the same way as younger speakers in all
three communities suggests that the change is
diffusing from Ballymacarrett. The problem is this:
how has this change crossed the notorious sectarian
lines of demarcation in the Belfast of the 19705,
since the civil disturbances which began during the
childhood years of these young speakers
prevented them from contracting close ties across the
sectarian divide? Older speakers, who report
having contracted many cross-community ties prior to
the civil disturbances, display no cross-
community consensus on the use of (u). The age-related
pattern of this change is puzzling if we
adopt the commonsense assumption that innovations are
transmitted via strong ties (for such a
model, see Labov 1980: 261; Labov and Harris 1986),
but not if we assume that the conduits of
innovation are the multiple weak ties of everyday
urban interaction in the neutral areas outside close-
knit community territories (for further discussion of
this issue, see Milroy and Milroy 1985).
Although it may at first seem counter-intuitive, a
“weak tie” model of change is plausible for several
reasons, as suggested by Granovetter (1973). First,
persons central to a close-knit, norm-enforcing
group are likely to find innovation of any kind
socially risky, but the adoption of an innovation already
on the fringes of the group less so. Second (in the
networks of mobile individuals at least), weak ties are more numerous than
strong ties, providing links to many more individuals; consider, for example,
the ties set up by participants at academic or
business conferences, which link cohesive groups
associated with each institution and provide conduits
for new ideas and information. Conversely,
information relayed through strong ties tends not to
be innovatory, since strong tie contacts are likely
to be shared (that is to belong to overlapping
networks). Thus, mobile individuals who have
contracted many weak ties, but occupy a position
marginal to any given cohesive group, are in a
favorable position to diffuse innovation.
Interestingly, this argument is consistent with the traditional
assumption of historians of language that the
emergent, mobile merchant class were largely
responsible for the appearance of northern (and other)
dialectal innovations in Early Modern London
English (see, for example, Baugh and Cable 1978: 194).
Using standard variationist methods,
historical sociolinguists have begun to examine
systematically the social trajectories of such earlier
changes with attention to the effects not only of
social network, but of gender and social status, with
promising results (Nevalainen 1999, Nevalainen and
Raumolin-Brunberg 1996, Tieken-Boon van
Ostade et al. in press). Sociohistorial linguistics
continues to develop as a significant subfield of
variationist theory.
5 Social Network, Social Class and Mobility.
The weak tie model of change discussed above can
illuminate the dynamics of dialect leveling-that is
the eradication of socially or locally marked variants
(both within and between linguistic systems) in
conditions of social or geographical mobility and
resultant dialect contact. Leveling might reasonably
be viewed as a linguistic reflex of the large-scale
disruption endemic in the modern world of close-
knit, localized networks which have historically
maintained highly systematic and complex sets of
socially structured linguistic norms. Such disruption
arises from (for example) internal and
transnational migration, war, industrialization, and
urbanization. While these dynamics have operated
earlier and more intensively in colonial contexts, as
discussed by Chambers within a broad social
network framework (1995: 57-66), they continue to
affect geographically or socially mobile
populations. In any event, leveling gives rise to
homogenization and a tendency for the localized
norms of the kind supported by a close-knit network
structure to become obliterated (Chambers
1995, Trudgill et al. 2000, Kerswill and Williams
2000, Britain 1997, Watt and Milroy 1999). This
process raises interesting psycholinguistic as well as
sociolinguistic issues concerning the functions
of close-knit networks, which are briefly explored
below.
On the basis of evidence from language attitudes
research, sociolinguists generally assume an
ideological motivation to underlie the long-term
maintenance of often stigmatized norms in the face
of pressures from numerically or socially more
powerful speech communities; speakers want to sound
(for example) Welsh, Irish, Northern English, New Zealandish,
Canadian, African-American, American
Southern and unlike whatever social group they
perceive themselves as contrasting with. The dialect
loyalty of such speakers and their resistance to
change originating from outside the group is usually
said to be motivated by their desire to index group
identity. Yet, motivations alone are insufficient to
maintain nonstandard varieties reliably (see Wolfram
et al. 1999 for a discussion of this issue in an
American context). Relevant here is Payne's (1980)
demonstration of the social conditions needed for
children to learn the highly localized phonolexical
complexities of the Philadelphia system;
particularly, their parents needed to be locally born
for such learning to take place. What this amounts
to is that if a close-knit community network structure
loosens and members become mobile, the
social prerequisites for supporting highly localized
norms disappear, and dialect leveling takes place.
Thus, not only does a community's sense of
distinctiveness become redundant as network ties loosen
(a social and ideological issue), but from a
psycholinguistic perspective speakers lack the extensive
and regular input needed to maintain localized norms.
Such norms are sometimes complex; for example, Belfast
speakers whose networks are relatively
loose-knit reduce the number of linguistically
conditioned allophones of /a/ by eliminating the
extreme back and front variants characteristic of the
vernacular system, often converging on a very
narrow area of vowel around the centre of the
vernacular range (Milroy 1982, Milroy 1999). Thus,
close-knit networks may be viewed not only as social
and sociolinguistic support mechanisms which
facilitate the construction and maintenance of local
distinctiveness; from the point of view of the
language learner, they also provide the intensive
input required to master complex, localized
linguistic structures which lack the support of
institutional models. Leveling, which from this cognitive perspective can be
viewed as a simplification strategy, takes place when such input is no longer
present. Trudgill (1989, i992) and Schilling-Estes
(2000) provide relevant discussions of the social
conditions in which both simplification and structural
complexity flourish.
We turn now to consider more specifically the links
between mobility, social network structure, and
social class. Following Giddens (1989: 205-73), class
is viewed here as one of four systems of
stratification which promote inequality in society.
While the other three (slavery, caste, and estates)
depend on institutionally sanctioned inequalities,
class divisions are not officially recognized, and
since an individual's class position is to some extent
achieved, class stratification is accompanied by
varying degrees of mobility. Issues of power
inequalities between groups and individuals are raised in
this discussion, which so far have only been touched
upon.
Different types of network structure seem to be
broadly associated with different social classes:
loose-knit networks with the socially and
geographically mobile mainly middle classes, and close-knit
ties with very low status and very high status
speakers. In terms of the predictions of the weak tie
model of change discussed above, this association is
consistent with Labov's principle that innovating
groups are located centrally in the social hierarchy,
characterized as lower-middle or upper-working
class (i 980: 254). The question then arises of how an
integrated model of change and variation might
be constructed which takes account of the relationship
between social class and social network
structures. Such an integration is desirable, since
the association of different network types with
different social class groups is not arbitrary, but
springs from the operation of large scale social,
political and economic factors (contra Guy 1988, who
views network and class as unrelated, but
pertaining respectively to a micro- and macro-level of
analysis).
Traditionally, sociolinguistics has assumed a
consensus model of class, where the community is said
to be fundamentally cohesive and self-regulating. Yet,
the vitality and persistence of nonstandard
vernacular communities highlighted by network studies
is more readily interpreted as evidence of
conflict and division than of consensus. Accordingly,
Milroy and Milroy (1992) argue that a dynamic
model of class as a process which splits the community
into subgroups (characterized by different
orientations to work, leisure and family) is helpful
in constructing an integrated theory of variation
and change. Hojrup's (i 983) analysis of these
subgroups as characterized by different lifemodes with
different network structures “falling out” from those
lifemodes is proposed as an approach which links
the variables of class and network in an illuminating
way.
With the link between social class and network
structure as their point of departure, Kerswill and
Williams (1999) have recently investigated the
relationship between social class, mobility, and
susceptibility to change by comparing the language
behavior of low and high mobility speakers of
different social statuses in the English towns of
Reading and Milton Keynes. They conclude that
network structure has the predicted effect-that is,
close-knit networks maintain localized norms,
while loose-knit networks facilitate change. However,
they argue that the variables of class and
network need to be considered independently, given the
different language behaviors of mobile high
status and mobile low status groups.
While the relationship between class, network, and
mobility is evident, its precise character is as yet
unclear as are the linguistic outcomes associated with
interactions between these social variables.
However, since they are constructed at different
levels of abstraction, it is likely that a two-level
sociolin-guistic theory would be helpful. Such a
theory should link the small-scale networks where
individuals are embedded and act purposively in their
daily lives with larger-scale social structures
which determine relationships of power at the
institutional level. The different sociolinguistic patterns
associated with both strong and weak ties would need
to be considered, with attention to recent work
on the sociolinguistics of mobility. For, while strong
ties give rise to a local cohesion of the kind
described by network studies of close-knit
neighborhoods such as those in Belfast or Detroit, they
lead also to overall fragmentation in the wider
community. Conversely, it is weak ties which give rise
to the linguistic uniformity across large territorries
such as that described by Chambers in Canada,
Labov in the United States, and Trudgill et al. (2000)
in New Zealand. The social dynamics underlying
both diversity and uniformity lie at the core of an
accountable theory of language variation and
change.
23. The Speech Community.
PETER L. PATRICK.
Subject Sociolinguistics » Language Variation and
Change.
The speech community (SpCom), a core concept in
empirical linguistics, is the intersection of many
principal problems in sociolinguistic theory and
method. I trace its history of development and
divergence, survey general problems with contemporary
notions, and discuss links to key issues in
investigating language variation and change. I neither
offer a new and correct definition nor reject the
concept (both misguided efforts), nor exhaustively
survey its applications in the field (an impossibly
large task).
1 General Problems with Speech Community as a Concept
Every branch of linguistics that is concerned with
representative samples of a population; that takes
individual speakers or experimental subjects as
typical members of a group; that studies langue as
attributable to a socially coherent body (whether or
not it professes interest in the social nature of
that body); or that takes as primitive such notions as
“native speaker,” “competence/ performance,”
“acceptability,” etc., which manifestly refer to
collective behavior, rests partially on a concept
equivalent to the SpCom. Linguistic systems are
exercised by speakers, in social space: there they are
acquired, change, are manipulated for expressive or
communicative purposes, undergo attrition, etc.
Whether linguists prefer to focus on speakers,
varieties or grammars, the problem of relating a
linguistic system to its speakers is not trivial.
In studying language change and variation
(geographical or social), reference to the SpCom is
inescapable, yet there is remarkably little agreement
or theoretical discussion of the concept in
sociolinguistics, though it has often been defined.
Some examples from research reports suggest the
degree of its (over-)extension (Williams 1992: 71).
The term “SpCom” has been used for geographically
bounded urban communities, both large
(Philadelphia; Labov 1989) and small (Anniston,
Alabama; Feagin 1996); for urban neighborhoods
(“Veeton” in Kingston, Jamaica; Patrick 1999) and
subgroups-Belfast vernacular speakers (Milroy and
Margrain 1980, but see Macaulay 1997: 15) and “the
French-speaking minority of Ontario,
Canada” (Mougeon and Beniak 1996: 69). It has been
denied for other cities (London; Wardhaugh
1998: 123) but used for Anglo-Saxon England (Labov
1982: 35), for urban immigrants, as distinct
from both their source and target groups (Kerswill
1994), and for the “national unity of a
people” (Dittmar 1976: 106). Cutting across geographic
and class lines, it has been used of very
general assemblages such as children (Romaine 1982: 7)
and women (Coates 1993: 140), as well as
specific and temporary ones such as members of ajury
(Durant 1999).
For rural populations, it has been used to pick out
named settlements of Warlpiri speakers (Bavin
1989), but also for a discontinuous, larger region-the
Caeltacht-in Ireland (Watson 1989) where
speakers do not define their communities in linguistic
terms. Joly (1 981) calls the Afro-Hispanic
No hay comentarios.:
Publicar un comentario