Você está na página 1de 14

9

LESLEY MILROY

Language Ideologies
and Linguistic Change

1. Introduction and preliminaries

Ronald Macaulay’s (1977) account of the Glasgow speech community would today
be described as “variationist,” that is, contributory to the core area of sociolinguistics,
which focuses primarily on developing socially sensitive accounts of language varia-
tion and processes of language change. Macaulay also dealt extensively, but sepa-
rately, with language attitudes. In this chapter I present work in progress, which
similarly addresses both language attitudes and processes of language variation and
change. My goal is to propose a framework for incorporating into mainstream varia-
tionist work an account of language attitudes, treated as manifestations of locally
constructed language ideologies.
In one sense this is a rather uncontroversial goal since from the publication of
Weinreich et al.’s (1968) classic article the evaluative dimension has been viewed
as a major component of a comprehensive account of language change. And indeed,
Labov’s (1963) classic analysis of socially motivated change in Martha’s Vineyard
treats attitude as central. For the most part, however, variationist accounts of lan-
guage attitudes and ideologies and of language variation and change have tended to
proceed along independent, parallel tracks. Influential work on attitudes was carried
out not by sociolinguists but by social psychologists (see also Giles and Coupland
1991; Milroy and Preston 1999) and was seldom integrated into basic accounts of
variation and change.1 This disjunction between analyses of attitudes and of socio-
linguistic patterns is evident in Macaulay’s (1977) monograph, where attitudes to
language in Glasgow and analysis of the sociolinguistic variables to which these

161
162 ATTITUDES AND IDEOLOGIES

attitudes refer are treated quite separately. Similarly, the recent work of Preston and
his colleagues on attitudes is carried out independently of analyses of variation and
change (e.g., Preston and Niedzielski 2000). In this chapter, I outline a framework
that treats attitudes expressed by individuals embedded in social groups as only one
kind of instantiation of ideologies, which may be defined initially as thoroughly
naturalized sets of beliefs about language intersubjectively held by members of speech
communities. A larger goal is to consider how ideologies interact with internal lin-
guistic constraints to structure patterns of variation and trajectories of change. My
orientation is quite similar to that of Labov’s (1963) Martha’s Vineyard study and is
indebted to its insights.
In the period since the appearance of that influential work, surely one of the
gems of sociolinguistics, variationists have demonstrated extensively the capacity
of phonological elements to index group collectivities of many kinds. For example,
Macaulay showed in great detail the massive consistency of language variation as a
fine-grained index of social class; in a set of composite scores for four Glasgow vowel
variables, only 1 of 48 speakers was out of the rank constructed in accordance with
social-class indexes. Moreover, the indexicality revealed in these rankings supported
an analysis of social class in Glasgow as tripartite, consisting of a unitary working-
class group, a white-collar group, and a professional and managerial group (Macaulay
1978:138). Indexicality of language, primarily with respect to the social catego-
ries of class, gender, and ethnicity, was demonstrated by other early work (e.g., Labov
1972; Trudgill 1974). A few years later, in L. Milroy ([1980] 1987a) and Milroy and
Milroy (1985), I argued further for the relevance of individual social network struc-
tures to accounts of the social dynamics of language maintenance and trajectories of
language change. Le Page and Tabouret-Keller (1985) objected to the treatment in
much early sociolinguistic work of social categories as “given,” independent of the
actions and attitudes of speakers. They therefore proposed a multidimensional ac-
count of individual “acts of identity” as speakers indexed their multiple and shift-
ing allegiances to different groups at different times. Eckert’s (2000) more recent
work also implicitly grants agency a central role in determining patterns of varia-
tion and trajectories of change. She argues for an ethnographically driven investi-
gation of social categories indexed by language, treating categories as constructed
by social actors. Hence, the researcher’s task is to discover rather than to assume
the relevance of particular categories. Her approach is similar to that of Le Page
and Tabouret-Keller, not only in its focus on agency, but also in treating indexicality
as a product of social practice that simultaneously projects more than one analyzable
“identity.”

2. Standard English and language attitudes

Almost all variationist work uses the standard language as a pivotal reference point,
thus rendering attitudes to the standard a crucial component of the variationist frame-
work. Yet the concept of the standard is surprisingly underspecified and undertheorized.
Standard English is the construct which I shall consider in this section, and I shall argue
that it is both misleading and unhelpful to treat it as a cross-culturally comparable and
LANGUAGE IDEOLOGIES AND LINGUISTIC CHANGE 163

sociolinguistically unproblematic entity. In early work by Labov, Trudgill, and others,


the standard is treated ad hoc as the norm to which speakers shift in their most careful
speech. This norm is generally identified with a prescriptive standard. For example,
Labov (1972:64–65) discusses the emergence of postconsonantal [r] as a prestige norm
in school systems and radio networks in the 1940s as a reaction against “an Anglophile
tradition which taught that (r-1) was a provincial feature, an incorrect version of the
consonant and that the correct pronunciation of orthographic r in car was (r-0).” Al-
though this spoken standard to which speakers shift in careful styles is thus oriented to
the norm prescribed by the New York City school system, for Trudgill (1974) the codi-
fied British elite accent, Received Pronunciation, provides the standard reference point.2
Even such a limited comparison between what Labov and Trudgill mean by “Standard
English” reveals the quite different sociologies underlying British and American im-
ages of the standard.
Labov’s (1972) claim that postvocalic [r] in New York City emerged as a new
prestige norm on the heels of a reaction against an Anglophile tradition (cf. Mencken
1948:28) is, in fact, not self-evidently correct. Bonfiglio (2002) offers an alternative
account, starting from the observation that Americans imagine the standard to be
located in the so-called midwestern, agricultural “Heartland,” not in the old and cul-
turally influential cities of the East Coast. In view of the cultural antecedents of stan-
dards in Europe (and elsewhere), this is an oddity parallel to the British standard being
imagined in relation to the dialects of rural Yorkshire or Dorset, or the French stan-
dard in relation to an agricultural region like Aquitaine. Bonfiglio presents compel-
ling evidence to support an argument that in the early twentieth century, American
xenophobia became focused on the immigrant cities of the East Coast, with a con-
comitant loss of linguistic capital for their nonrhotic dialects. Industrial centers were
seen as sources of contamination of race, religion, and language, and purist ideolo-
gies developed that associated American virtues of all kinds with the rural areas of
the Midwest. These ideologies are clearly manifested in (for example) the writings
of Brander Matthews, who taught at Columbia from 1892 to 1924. His 1921 article
“Is the English Language Degenerating?” concluded that it was in a bad way—even
worse in America than in Britain. For Matthews, “Linguistic decay [was] inseparable
from racial decadence” the reason being “its immediate contact with a host of other
and inferior tongues” (Bonfiglio 2002:147). In Matthews’s view, standardization was
a means of maintaining linguistic and racial purity, and Bonfiglio provides many
examples of such discourse that focus on [r] pronunciation as indexical not only of
racial purity but more generally of virtue and national vigor.
Mencken (1948:28) characterizes general American (i.e., the variety spoken west
of the eastern seaboard) as “mainstream” and as an ideal candidate for the standard.
This rhotic variety was adopted by the radio networks, starting with CBS in the 1930s,
and constructed as the homogeneous language of the majority. Bonfiglio (2002) ar-
gues that the linguistic ideologies that locate an American standard in the Midwest
spring from a rejection not only of urban immigrants but also of African Americans
and white Southerners. This analysis appears to be consistent with an American ide-
ology that foregrounds race and ethnicity (see also L. Milroy 2000). It also high-
lights the relationship between such an ideology and the trajectory of a linguistic
change in mid-twentieth-century New York City. An analysis of the social motivations
164 ATTITUDES AND IDEOLOGIES

underlying the changing social evaluation of (r), which treats race and ethnicity as
salient, is also consistent with important variationist preoccupations; consider par-
ticularly the emergence during the twentieth century of a set of supralocal norms
that increasingly distinguish African-American English from ambient white dialects
(Wolfram and Thomas 2002:11). It is difficult, on the other hand, to find much evi-
dence of “Britishness” as a salient social category either in contemporary American
culture or in American sociolinguistics, despite the traditionally favorable attitude
to British accents noted by Lippi-Green (1997). Yet, this is what Labov’s (1972)
analysis of the revalorization of (r) in New York City seems to assume.
The association of (r) revalorization under these circumstances with a new
American standard supports the general proposition that standard languages are
best treated as constructs that are emerging from the particulars of a nation’s history
and social structure. This being so, we would not expect “Standard English” to be a
comparable sociolinguistic entity in different English-speaking countries. Although,
indeed, RP is a class-defined reference accent with a social positioning very differ-
ent from Labov’s (1972) standard in New York City, Norwich speakers appeared to
orient to RP norms in a way that made it possible for Trudgill (1974) to adapt Labov’s
(1966) framework without much difficulty. However, Scotland and Ireland have long
histories of independence from and opposition to the English metropolis, and socio-
linguists who attempt to employ Labov’s framework in those countries report sig-
nificant difficulties in specifying the norms to which individuals might be shifting in
careful speech (L. Milroy [1980] 1987a:105). An example of such a problem is the
behavior of Belfast working-class speakers, who certainly modified vernacular norms
in careful spontaneous speech but not when reading word lists. Furthermore, in both
Edinburgh (Romaine 1978) and Belfast (L. Milroy [1980] 1987a), a range of differ-
ent middle-class and educated accents could be heard, so unlike the situation reported
in New York City and Norwich, more than one high-status linguistic model was
available. And as noted by Harris (1991) and Gunn (1990), ethnicity in Belfast is
indexed by a speaker’s choice of norm in careful or rhetorical speech more overtly
than in the vernacular varieties discussed by (L. Milroy [1980] 1987a). This situa-
tion was not envisioned by earlier work, which assumed a relatively unitary and fo-
cused set of standard norms. These norms were associated in an unproblematic way
with social class and with stylistic variability, represented on intersecting continua
from most to least standard.
It is clear also that the target phonological system(s) of careful speakers in
Glasgow, Edinburgh, and Belfast were rarely oriented to RP. Indeed, the phonolo-
gies of these Celtic fringe dialects could not be mapped directly onto RP, as is shown
by the example of the low vowel /a/ in both Glasgow and Belfast (Macaulay 1977,
1978; L. Milroy [1980] 1987a). RP and many Anglo-English dialects are character-
ized by a clear phonological distinction between short high front and long back vari-
ants of this vowel in pairs like Sam: psalm and have: halve. Variation in length and
quality in Scotland and Northern Ireland, however, is allophonic rather than phono-
logically distinctive. A very few careful middle-class speakers in Glasgow—and
indeed in Belfast—display an orientation to an RP type of norm with respect to /a/.
However, most Belfast speakers (Macaulay’s methods do not reveal this pattern in
Glasgow) construct careful styles by reducing the range of allophonic variation, thus
LANGUAGE IDEOLOGIES AND LINGUISTIC CHANGE 165

eliminating the extreme back and front variants characteristic of the vernacular sys-
tem ([ε,,a,ä,ɑ,ɔ]).Such speakers sharply contract the allophonic range and over-
whelmingly realize /a/ class items with allophones in the region of [a,ä], avoiding
back or front-raised variants, which strongly index working-class status in the city.
James Milroy (1982) describes this pattern as “normalization,” but following
more recent sociolinguistic work that borrows insights from dialect contact models,
I would now describe it as leveling. Leveling is typically a linguistic process that has
the effect of reducing variability both within and across language systems and which
in principle operates independently of an institutional norm (Watt and Milroy 1999;
L. Milroy 2002). Standardization as manifested in careful or higher-status speech
typically displays an orientation to an institutionally supported norm.3 In any event,
the analytic difficulties produced by the near irrelevance of RP as a careful speech
model in the Celtic fringe cities is yet another reason for variationists to deconstruct
the notion of the standard. All the upper working- and lower-middle-class Belfast
speakers interviewed in the 1970s in the research projects described by Milroy
(L. Milroy 1987b: see also Milroy and Gordon 2003) were oriented to a leveled norm,
which is more plausibly conceptualized as a modification of the vernacular than as
an externally supported norm of the kinds identified in Norwich and New York City.
I conclude this section with an account of a particular problem for variationists
that emerges if the linguistic standard is treated as “given” and therefore not requir-
ing further analysis. It is generally presented as a neutral reference point for descrip-
tions of variability, and although sociolinguists are aware of its intrinsically hegemonic
character, by treating it as neutral they both have their cake and eat it. Silverstein
(1996:284) points out correctly (if a little opaquely) that “once the debate is focused
on linguistic issues in terms of The Standard versus whatever purportedly polar
opposites [sic], then the fact that the situation is conceptualized in terms of The Stan-
dard indicates what we might term its hegemonic domination over the field of con-
troversy, no matter what position is taken with respect to it.” Since there is no neutral
reference point and no neutral way of reacting to and analyzing language variation,
scholars imbue their sociolinguistic analyses with unintended ideological significance
when they focus on the characteristics of some variety by comparing it with a sup-
posedly neutral standard. James Milroy (2001) explores in some depth such ideo-
logical effects on the theories and descriptive frameworks of linguists, but there is
little systematic discussion in the variationist literature of the particular contradic-
tions that arise when the linguistic standard is subjected to the oxymoronic treatment
described by Silverstein. These contradictions occasionally become visible, as do other
unacknowledged reference points arising from sociolinguistic hegemonies. For ex-
ample, in a manner uncontroversial to linguists, Labov and Harris’s account of di-
vergence between black and white vernaculars treats both varieties as linguistically
equal. However, when the possibility of an ultimate convergence is discussed, the
curtain is briefly lifted and the assumption of a concealed reference point is evident:
“If the contact is a friendly one, and we achieve true integration in the schools, the
two groups may actually exchange socially significant symbols, and black children
will begin to use the local vernacular of the white community” (Labov and Harris
1986:21; my emphasis). In one of the few discussions in the literature of the hege-
monies underlying the neutral standard assumption, Walters (1996) details the
166 ATTITUDES AND IDEOLOGIES

problems and inconsistencies visible to African-American students who are intro-


duced to elementary linguistics textbooks. These texts proclaim the linguistic equal-
ity of all varieties, while simultaneously presenting a supposedly neutral standard
variety as the embodiment of the language ‘English’.

3. Language ideology as social process

I have argued so far that an infrastructural prerequisite for an analysis of language


attitudes and ideologies is a systematic account of the concept of the standard, and
related covert linguistic hegemonies. It is evident that such an account has implica-
tions also for analytic procedures (as in the case of leveled vowel systems in Belfast)
and for a motivated analysis of the particular ideologies that structure the trajecto-
ries of linguistic changes [in the case of (r) in New York City]. Taking as a starting
point the account of the revalorization of (r) set out in the previous section, I turn
now more generally to the capacity of language ideologies to structure language
change, and I consider a style of ideological analysis that seems potentially able to
contribute to an integrative account of language attitudes and language use. The
approach of scholars such as Woolard, Silverstein, Irvine, and Gal, set out in Kroskrity
(2000), is particularly promising in this respect. This approach does not treat ideolo-
gies primarily as manifestations of hegemonic practices, as has commonly been done
by sociolinguists (cf. Lippi-Green 1997; Ronkin and Karn 1999), but rather models
the semiotic processes that give them their social significance. They are thus pre-
sented as a central feature of the sociolinguistic landscape rather than as cultural
models embraced by the prejudiced and linguistically naive.
In accordance with this approach, language ideologies are defined by Silverstein
(1979:193) as “sets of beliefs about language articulated by users as a rationaliza-
tion or justification of perceived language structure or use” and by Irvine (1989:255)
as “the cultural system of ideas about social and linguistic relationships, together with
their loading of moral and political interests.” These definitions are mutually consis-
tent but slightly differently focused. Silverstein suggests a motivation for construct-
ing ideologies in the first place; speakers construct “stories” to explain observed
linguistic phenomena. Irvine emphasises the political and other interests that struc-
ture intersections between cultural conceptions of language and the social world. For
both Irvine and Silverstein, ideologies provide a window on everyday understand-
ings of the social significance of language, playing an important role in delimiting
and defining salient social groups and, indeed, whole nations. They involve not only
beliefs about language variation and language users but also the creation of lineages
and histories for national standard languages that arise from these beliefs. J. Milroy
(2002) and Woolard (2002) illustrate this dimension of ideological activity in analy-
ses of the lineages created, respectively, for English and Spanish. Both scholars show
clearly how these ideologized histories emerge from specific political-economic and
social circumstances. This historical dimension is crucial for an understanding of how
ideologies work since, typically, they are historically deep-rooted and thoroughly
naturalized—hence their resistance to analysis or argument. Thus, far from mani-
festing a misguided approach to language variation, which serious investigators should
LANGUAGE IDEOLOGIES AND LINGUISTIC CHANGE 167

ignore, language ideologies link the cultural world of the language user to macro-
level social and political forces. This is, of course, the very link that sociolinguists
attempt to create when they model sociolinguistic structure with reference to social
categories.
Noting that languages and language forms index speakers’ social identities fairly
reliably in most communities, Silverstein (1992, 1995) further elaborates his notion
of ideology as a system for making sense of this inherent indexicality. Importantly,
indexicality can usefully be ranked into different orders of generality. For him, first-
order indexicality entails the association by social actors of a linguistic form or va-
riety (accurately or otherwise—accuracy is beside the point) with some meaningful
social group such as female, Asian, Spanish, working class, aristocratic, and so forth.
Although social class, gender, and ethnicity are the time-honored sociolinguistic social
categories, they by no means exhaust the indexical potential of language: consider,
for example, the use of phonological resources in rural Wales to index association
with different villages or different chapels (L. Milroy 1982; Thomas 1988) or in
Detroit high schools to index association with particular adolescent social categories
(Eckert 2000). The point here is that an ideological analysis treats social categories
as locally created by social actors and discoverable by analysis, rather than as given.
Consequently, an ideologically oriented account of language variation and change
treats members of speech communities as agents, rather than as automatons caught
up ineluctably in an abstract sociolinguistic system.
Whereas Silverstein’s (1992, 1995) first-order indexicality refers rather straight-
forwardly to the association of linguistic form with social category, second-order
indexicality is a metapragmatic concept, describing the noticing, discussion, and
rationalization of first-order indexicality. It is these second-order indexical processes
that emerge as ideologies. But crucially, language varieties are likely to be differ-
ently noticed, rationalized, and evaluated from community to community and from
nation to nation. In different communities, different varieties are foregrounded, and
the kind of people who speak these varieties are differently ideologized. Particular
ideologies need to be explained in terms of local histories and local social, political,
and economic conditions.
The key point here is that an analysis of ideologies in terms of Silverstein’s (1992,
1995) concept of second-order indexicality provides for relative prominence in cul-
tural models of particular social groups, and the recession of others as first-order
indexicalities are rationalized in different ways. This sociolinguistic relativity accounts
for the evident salience of race and ethnicity in American language ideology and of
social class in British ideology, as noted by (L. Milroy 2000). And since ideologies
purport to explain and rationalize the source and significance of linguistic differences,
they restructure and distort relationships between the index (i.e., the linguistic form)
and the social group indexed, locating linguistic forms “as part of, and as evidence for,
what they believe to be systematic behavioral, aesthetic, affective and moral contrasts
among the social groups indexed” (Irvine and Gal 2000:37). Hence the pervasiveness
of strongly held but palpably counterfactual beliefs about (for example) the superior-
ity of the standard, the impoverished character of working class or ethnically distinc-
tive dialects, the superiority of English or French over other languages, of Colombian
and Argentinian varieties of Spanish over other New World varieties, and so forth.
168 ATTITUDES AND IDEOLOGIES

It is important for sociolinguists that second-order reactions to basic indexicality


are visible not only in such attitudes and beliefs about variability but also in other
kinds of rather familiar behaviors. For example, the kind of sociolinguistic variables
described by Labov (1972) as stereotypes, as distinct from indicators and markers,
show the effects of ideological construction particularly clearly. Reactions to such
forms by participants are evidenced in stylistic shifting and hypercorrection (such as
Cuba[r], idea[r] in nonrhotic dialect areas of the United States). Because they in-
stantiate a reaction to a hegemonic standard, these behaviors are sometimes inter-
preted as manifestations of linguistic insecurity. Similar reactions to forms associated
with nonstandard varieties in which speakers misanalyze a system as they orient to
local dialects are not usually cited as evidence of linguistic insecurity—another ex-
ample of the unacknowledged effect on variationist frameworks of standard as covert
reference point. Trudgill (1986: 66–78) describes such misanalyses as “hyperdia-
lectisms.” Thus, for example, in Tyneside, northern England: He divvn’t drink “he
doesn’t drink” is a traditional dialect form. Although *he div drink “he does drink”
is not, such forms do in fact occur, spoken chiefly by young male speakers of a dia-
lect that is progressively losing extremely localized forms (see Watt and Milroy 1999).
All such misanalyses display a sensitivity to the indexical relationship between lan-
guage and social category. Therefore, regardless of the status of the social group trig-
gering the misanalyses, they can be viewed as instantiations of second-order indexical
processes and hence as evidence of local ideological systems. Crucially, an ideological
analysis in these terms addresses both language attitudes and sociolinguistic patterns
evident in the distribution of variants.
Citing a familiar example, Irvine and Gal (2000:47) exemplify the kind of ac-
count of socially motivated linguistic change (incorporating both language attitudes
and patterns of language use) that such an analysis might offer. They somewhat re-
focus Labov’s (1963) interpretation of high levels of centralization in the first ele-
ments of the diphthongs /ay/ and /aw/ in Martha’s Vineyard, emphasising the content
of the ideology that gives rise to them. Thus, distinctions among ethnic groups on
the island (Yankees, Portuguese, and Indians) were important in the 1930s but re-
ceded into the background as an opposition between islanders and mainlanders be-
came salient 30 years later in response to changing economic pressures. This social
change structures the trajectory of change in /ay/ and /aw/ since centralized variants
of these phonological elements are associated particularly with island versus main-
land affiliation, whereas distinctions among ethnic groups are no longer ideologized:
“The sudden increase in centralization began among the Chilmark fishermen, the most
close-knit group on the island, the most independent, the group which is most stub-
bornly opposed to the incursions of the summer visitors” (Labov 1972:37). The re-
sulting divergence between variants that index mainlander versus islander affiliation
is ideologically mediated in the sense that it depends upon “local images of salient
social categories that shifted over time” (Irvine and Gal 2000:47).
Forty years after Labov’s (1963) landmark study, Blake and Josey (2003) re-
port the results of a real-time study of /ay/ variants on the island. Adopting the ana-
lytic framework set out by Irvine and Gal (2000) and elaborated by L. Milroy (2003),
they note that social conditions on the island have changed in several ways. First, the
island economy has diversified considerably, so that average incomes are now higher
LANGUAGE IDEOLOGIES AND LINGUISTIC CHANGE 169

than 40 years ago and unemployment rates lower. Second, the socially constructed
distinction emphasised by Labov between incomers as “others” and islanders as
“locals” has virtually disappeared. Third, as a result of a movement toward a less
localized economy and the development of service industries, the fishermen consti-
tute a less distinctive community. The localized fishing industry has been absorbed
into larger conglomerates, and the close-knit networks previously characteristic of
the fishing community have been disrupted. Blake and Josey note that whereas local
identity is still a salient social category on the island, it is no longer associated with
the Chilmark fishermen or even with year-round residence on the island. Most Vine-
yarders, with the important exception of the Native Americans of Gay Head, can be
assigned a middle-class status.
This social restructuring, which has obliterated the ideological distinction be-
tween the social categories “islander” and “mainlander,” also appears to have struc-
tured the trajectory of recent changes in the social distribution of /ay/ variants. Blake
and Josey (2003) observe that the trend toward high degrees of centralization among
the fishermen reported by Labov has halted. The fishermen’s mean index score for
/ay/ is not significantly different from that of other social groups, and no speaker or
group of speakers emerges as distinctive. The authors argue that the socially moti-
vated pattern of change affecting /ay/ as reported by Labov has disappeared. They
characterize the subsequent trajectory of change as the phonologically related but
linguistically conditioned process known as Canadian Raising, which affects many
more communities than Martha’s Vineyard in different parts of the English-speaking
world (Chambers 1973; J. Milroy 1996; Britain 1997). Unlike the trajectory noted
by Labov 40 years earlier, Canadian Raising cannot easily be described as socially
motivated by local ideological factors.

4. Social and language internal constraints


on change: Some points of intersection

At this point, let me review and elaborate on the major issues discussed so far. We
have considered rather broadly the role of ideological and attitudinal factors in an
account of language variation and change, and an ideological analysis of specific
changes was exemplified by the well-known cases of (r) in New York City and /ay/
and /aw/ in Martha’s Vineyard. Ideologically motivated change is explicated with
reference to local images of language variation, which construct some social groups
and their language forms as salient while others are consigned to the background;
and of course, different groups may be foregrounded at different times. Such an
analysis addresses central concerns of variationist research, treating the social cate-
gories not only as constructed by social actors but also as variably salient. A motiva-
tion is explicitly provided for speakers to structure variation and change in accordance
with their understanding of changing social configurations. The identification of
relevant social factors is therefore seen as part of the analysis rather than a frame-
work for structuring a subsequent analysis. Dyer’s (2002) account of the relation-
ship between changing social saliences and phonological change in a town in the
English Midlands provides a clear example of how such an analysis works in practice.
170 ATTITUDES AND IDEOLOGIES

Indeed, a good deal of recent sociolinguistic work is quite consistent with this analy-
sis (see particularly Dubois and Horvath 1999 and, of course, Eckert 2000).
More broadly, ideological motivations underlie the long-term maintenance of
stigmatized norms in the face of pressures from focused and codified standard lan-
guages so familiar to sociolinguists; speakers want to sound like Southerners, Afri-
can Americans, Tynesiders, Belfasters, New Zealanders, and other groups with whom
they align. They also want to sound unlike whatever group they currently perceive
themselves as contrasting with. The loyalty of such speakers to their own dialects
and their resistance to language forms associated with others constructed as opposi-
tional is usually described by sociolinguists as exemplifying the capacity of spoken
language to index identity. The identity factor is, in fact, precisely what an ideologi-
cal framework addresses, assuming also (and crucially) that salient social groups with
which speakers identify are subject to change.
Factors constraining4 change, which are generally described as “language inter-
nal,” appear to operate independently of local ideologies, and in this section I shall
attempt to examine the ways in which such factors work together with ideological
motivations to give rise to particular linguistic outcomes. We can exemplify the point
at issue here by referring to Eckert’s (2000) account of the interaction between com-
plex configurations of adolescent social category and gender, on the one hand (so-
cial factors), and the trajectories of change constrained by the rotations of the Northern
Cities Shift on the other (internal factors). Eckert emphasises the local construction
of the social categories that provide the dynamic for change, and her account of the
ideologies underlying them is entirely consistent with the approach developed in this
chapter. However, the interaction between these changes and the principles of change
in vowel systems that constrain the Northern Cities Shift is not clear. The remainder
of this section focuses on points of intersection such as this one.
Internally constrained changes are generally understood to be explicable in terms
either of human speech production or perception mechanisms or of intrasystemic con-
straints on possible changes. They are therefore widely distributed across the world’s
languages (see Campbell 1999:chap.11). For example, C-J. Bailey’s (1996:369) no-
tion of connatural change (i.e., internally constrained as opposed to socially motivated)
adopts the first kind of explanation, assuming neurobiological rather than phonologi-
cal constraints. Ohala (1993) emphasises the role of perceptual mismatches, propos-
ing that misperceptions of the speech signal by the listener result in a change in phonetic
target; if this target is phonologized, sound change results (see also Beddor and Krakow
1998). Using a more overtly teleological kind of reasoning, Lindblom (1986, 1990)
argues that vowel systems evolve to maximize perceptual distance between units in
the system and that natural phonetic processes underlie many regularities.
Much of Labov’s most influential recent work has focused on internally con-
strained change. Although he alludes to universal articulatory or perceptual constraints
of the kind mentioned above (1994:220–221), he attributes internally constrained
change primarily to the symmetry-preserving pressure of the system. Vowel changes
are seen as constrained by a limited set of language-internal principles, replacing an
earlier, more atomistic analysis of individual phonological variables. The chain shift
notion is central to Labov’s account of language internal change, the fundamental
principle being that movement of one vowel in the system triggers a series of further
LANGUAGE IDEOLOGIES AND LINGUISTIC CHANGE 171

changes, as adjacent vowels move sequentially to fill emerging gaps. A series of


principles derived from a survey of a large number of documented chain shifts is
said to constrain possible changes; for example, long vowels are said to rise, short
vowels to fall, and back vowels to move to the front (1994:115–154). The nature of
language-internal factors in language change, and the relationship between language-
internal and social factors are perennial subjects of debate in historical linguistics.
Accordingly, parts of Labov’s work have drawn criticism from historical linguists,
particularly Lass (1997:34ff) and Stockwell and Minkova (1997, 1999), chiefly be-
cause the chain shift formulation carries with it unacknowledged implications of
causality. Also problematic is the applicability of the chain shift concept to work in
contemporary speech communities (see also Gordon 2000; Watt 2000).
Reservations about chain shifts as an appropriate model do not invalidate Labov’s
(1994) principles of vowel change since empirical research can confirm whether long
vowels do indeed rise and short vowels fall or back vowels move to the front, and it
is certainly clear that such principles frequently capture recurrent types of change.
For example, Trudgill et al. (2000) appeal to internal principles to account for a set
of changes associated particularly with Southern Hemisphere varieties of English,
southern British English and the Southern dialect of the United States. These changes
are described by Wells (1982) as “diphthong shift” to refer to ongoing developments
whereby the first elements of the rising diphthongs associated with the Great Vowel
Shift (/i: u:, ei, ou, ai, au/) continue to move in an apparently coordinated fashion.
Labov (1994:217) discusses the same set of changes in a somewhat different way, as
a subsystem of the Southern Shift. Although these changes cannot be accounted for
within an overarching theory, some can be accommodated by universal principles of
the kind posited by Labov, and all appear to be attributable to properties of the lan-
guage rather than to any detectable social factor.
Many other sets of vowel changes identified by sociolinguists as operating in
widely separated communities are explained with reference to language-internal
constraints. Canadian Raising is one such change, and we noted already that Blake
and Josey (in press) describe recent changes in Martha’s Vineyard in relation to this
process. The point here is that the ideological system motivating an earlier associa-
tion of /ay/ raising with the Chilmark fishermen changed when that group ceased to
be salient. Thus, the socially motivated pattern of change reported by Labov (1963)
disappeared. What is observable now is a phonologically similar pattern, which is,
however, linguistically rather than socially constrained and arises independently in
different communities (see again Chambers 1973; L. Milroy 1996; Britain 1997).
The general proposal here is that local social factors (treated as ideologically driven
processes rather than as the social categories more familiar to sociolinguists) oper-
ate as constraints on changes driven by internal factors. If the local social bound-
aries set in place by ideological processes weaken or disappear, language-internal
changes can take their course, uninhibited by local ideologies.
An analysis of this kind offers an answer to a pair of related questions about the
Northern Cities Shift: why does this major series of vowel shifts have a particular
geographical distribution and why does it stop where it does? The shift is reported to
be most vigorous in the cities settled after the opening up of the Great Lakes by the
Erie Canal in an era of greatly increased mobility ushered in by the steamship, the
172 ATTITUDES AND IDEOLOGIES

railroad, and the automobile (see further Wolfram and Schilling-Estes 1998). How-
ever, it appears to be blocked when it encounters the social boundaries set up by
historically well-established and linguistically focused communities; hence it does
not affect the older East Coast cities or eastern New England. The suggestion here is
that this set of language-internal changes is blocked by speech communities where
historically long-standing and thoroughly ideologized patterns of indexicality are
firmly in place. It might further be predicted that if social changes in a particular speech
community weakened the indexical relationship between language and local social
categories, the NCS changes would spread into that community.
This account of the intersection between local ideologies and internal factors as
catalysts of change builds upon a good deal of solid variationist research that focuses
either on social motivations or on internal constraints. Discussing the issue of the
relationship between internally motivated and externally constrained change in an
adjacent field, Clyne (2003:3) similarly emphasises their complementarity. Although
Jones and Esch (2002:1) are surely correct in noting a tendency to focus disjunc-
tively on either social or internal constraints on change, accounts that attempt to in-
tegrate both kinds of factors are not new, and some are in certain respects similar to
the one presented here. Consider the example of the set of coordinated changes in
Southern Hemisphere English described as diphthong shift, which was discussed
earlier in this section. Noting its continuation along a trajectory already present in
nineteenth-century local accents of southeastern England, John Wells suggests that
the changes were free to take a natural course in the colonies, particularly Australia,
because they were uninhibited by the sociolinguistic constraints of RP: “In a village
or small town . . . in 1800, a man would be in regular contact with RP speakers (squire,
rector, doctor) and the social pressures to admire and imitate qualities associated with
this culturally dominant class were strong” (1982:593).
Although Wells (1982) characterizes a relationship between language internal-
and ideologically motivated change similar to the one proposed here, he appears to
restrict to high-status models the effect of ideology in structuring change. I would
suggest a larger generalization of the kind already proposed to account for the barri-
ers encountered by the American Northern Cities Shift, namely, that Southern Hemi-
sphere diphthong shift can take place because it is uninhibited by the social boundaries
associated with first-order indexicalities of all kinds. This conception does not privi-
lege varieties that index the high status of the English rector and squire but includes
also those that index other salient social groups in the migrants’ original homes. Simi-
lar comments might be made about Kroch’s (1978) theory of change, which charac-
terized ideology as suppressing “natural” phonetic processes. Like Wells, Kroch
restricts his discussion of the inhibiting effects of ideology to standard or high-status
varieties. It is also worth commenting here on Labov’s (1994:78) long-standing
distinction between “change from above” and “change from below,” which refers
simultaneously to levels of social awareness and positions in the socioeconomic hi-
erarchy. “Change from above” appears to be ideologically motivated, being triggered
by a response to a particular social group [as in the case of (r) in New York City].
However, its characterization with reference to a high-status group of speakers is
problematic since the social semiotics underlying ideologically motivated change are
not thus restricted.
LANGUAGE IDEOLOGIES AND LINGUISTIC CHANGE 173

5. Conclusion

My main goal in this chapter has been to develop an analysis of the role of speaker
attitude in language change that can be incorporated into a variationist account. To this
end, I have adapted an ideological analysis of the kind developed by scholars such as
Gal, Irvine and Silverstein. Since attitudes toward the standard in speech communities
are usually very salient, I began by looking at different local images of Standard English.
I argued particularly that the concept of the standard was underspecified in socio-
linguistics, and I discussed some consequences of this underspecification. The way in
which an ideological analysis might handle the task of characterizing trajectories of
socially motivated change was illustrated by several examples, some very prominent
in the sociolinguistic literature. In addition, L. Milroy (2003), and Anderson et al.
(2002) discuss changes in the back vowel system in the African-American commu-
nity in Detroit with reference to this framework.
The previous section focused on the intersections between “internally” and
“externally” constrained change. The fundamental proposal is that local social fac-
tors, discussed here in terms of ideologically driven processes rather than as social
categories, operate as constraints on language-internal change. The corollary to this
proposal is that if social boundaries become permeable and ideological systems were
disrupted as a result of social change, language-internal changes will take their course,
uninhibited by social barriers; hence (for example) the spread of the Northern Cities
Shift in the United States across large territories. Building upon existing variationist
work on language change, I have attempted to integrate perspectives that usually
appear disjunctively in our large and rich body of research into sociolinguistic pat-
terns in speech communities.

Notes
Versions of parts of this chapter were presented in October 1999 at NWAVE-28, Toronto,
and in November 1999 at the conference organized at Pitzer College, Claremont, Cali-
fornia, in honor of Ronald Macaulay. I am grateful to colleagues at both of these meet-
ings for helpful contributions. I thank James Milroy for helpful comments on the entire
chapter.
1. Bell’s work (1984, 1999) and very recently that of Garrett et al (2003) is a notable
exception to this generalization.
2. However, whereas Trudgill characterises RP as the relevant norm at the standard end
of the sociolinguistic continuum, he elsewhere (most recently Trudgill 1999) treats British
standard English as a dialect that can be spoken with any accent.
3. This distinction simplifies matters somewhat. It does not, of course, preclude the
emergence in speech communities of leveled varieties as institutionally supported spoken
standards, as seems to have happened in the United States (L. Milroy 2000).
4. A pedantic terminological point is in order here. Given that speakers are agents with
attitudes and motivations, it is reasonable to speak of socially motivated change. But for
two reasons it is preferable to speak of internal constraints on change rather than internal
motivations. First, it is not easy to attribute motives to abstract systems; second, the term
motive and its derivatives implies a teleological way of thinking about language-internal
change that is not consistent with the proposal developed in this chapter. In a recent article
that addresses a range of issues overlapping the subject matter of this chapter (Milroy 2003),
174 ATTITUDES AND IDEOLOGIES

I have commented on some difficulties arising from the dichotomy between language-inter-
nal and social factors in language change.

References
Anderson, Bridget, Lesley Milroy, and Jennifer Nguyen. 2002. Fronting of /u/ and /U/ in
Detroit African American English: Evidence from Real and Apparent Time. Paper pre-
sented at NWAV 31, Stanford University, Stanford, Cal.
Bailey, Charles-James. 1996. Essays on Time-based Linguistic Analysis. Oxford: Clarendon.
Beddor, Patrice, and Rena Krakow. 1998. Perceptual Confusions and Phonological Change:
How Confused Is the Listener? In Proceedings of the Twenty-fourth Annual Meeting of
the Berkeley Linguistics Society: General Session and Parasession on Phonetics and
Phonological Universals, ed. Benjamin Bergen, Madelaine Plauché, and Ashlee Bailey,
320–335. Berkeley, Cal.: Berkeley Linguistics Society.
Bell, Allen. 1984. Language Style as Audience Design. Language in Society 13:145–204.
———. 1999. Styling the Other to Define the Self: A Study in New Zealand Identity Mak-
ing. Journal of Sociolinguistics 3(4):523–541.
Blake, Renee, and Meredith Josey. (2003). The /ay/ diphthong in a Martha’s Vineyard Com-
munity: What Can We Say 40 Years after Labov? Language in Society 32(4):451–485.
Bonfiglio, Thomas P. 2002. Race and the Rise of Standard American. Berlin and New York:
Mouton de Gruyter.
Britain, David. 1997. Dialect Contact and Phonological Reallocation: “Canadian Raising” in
the English Fens. Language in Society 26(1):15–46.
Campbell, Lyle. 1999. Historical Linguistics: An Introduction. Cambridge, Mass.: MIT Press.
Chambers, Jack K. 1973. “Canadian Raising.” Canadian Journal of Linguistics 18:113–135.
Clyne, Michael. 2003. Dynamics of Language Contact. Cambridge: Cambridge University
Press.
Dubois, S., and B. M. Horvath. 1999. When the Music Changes You Change Too: Gender
and Language Change in Cajun English. Language Variation and Change 11(3):287–
313.
Dyer, Judith. 2002. “We All Speak the Same Round Here”: Dialect Leveling in a Scottish-
English Community. Journal of Sociolinguistics 6(1):99–116.
Eckert, Penelope. 2000. Linguistic Variation as Social Practice. Oxford: Blackwell.
Garrett, Peter, Nikolas Coupland, and Angie Williams. 2003. Investigating Language Atti-
tudes: Social Meanings of Dialect, Ethnicity and Performance. Cardiff: University of
Wales Press.
Giles, Howard, and Nikolas Coupland. 1991. Language: Contexts and Consequences. Milton
Keynes: Open University Press.
Gordon, Matthew J. 2000. Small-town Values, Big-city Vowels: A Study of the Northern Cit-
ies Shift in Michigan. Publications of the American Dialect Society 84. Durham, N.C.:
Duke University Press.
Gunn, Brendan. 1990. The Politic Word. Ph.D. diss., University of Ulster, Jordanstown,
Northern Ireland.
Harris, J. 1991. Ireland. In J. Cheshire (ed.), English around the World: Sociolinguistic Per-
spective. Cambridge: Cambridge University Press.
Irvine, Judith T. 1989. When Talk Isn’t Cheap: Language and Political Economy. American
Ethnologist 16:248–267.
Irvine, Judith T., and Susan Gal. 2000. Language Ideology and Linguistic Differentiation. In
Regimes of language, ed. Paul Kroskrity, 35–84. Santa Fe, N.M.: School of American
Research Press.

Você também pode gostar