WILLIAM BRIGHT.Two fundamental facts of language are (a) that it is always changing, in all areas of structure (phonology, grammar, discourse style, semantics, and vocabulary) and (b) that it changes in different ways at diverse places and times. Some societies have made efforts to check the mutability of language; where literacy is present, special efforts have been made to stabilize written languages in particular. Such attempts typically involve prescriptive grammars, as well as codified orthographies made accessible through authoritative dictionaries; these identify conservative usages, linked with traditional literature and established social values, and they discourage departure from the established norms. In some countries, learned institutions like the Academie Francaise or the Academia Real de la Lengua Espanola have been given official responsibility for maintaining the linguistic status quo. In such a literate milieu, the written language is typically held up by educational institutions as a model for the spoken language; and innovative linguistic usages are discouraged in speech as well as writing - such as slang vocabulary (e.g., Eng. booze instead of liquor), analogical simplification in grammar (he don't instead of he doesn't), and departure from the orthographic norm in pronunciation (such as the merger of the vowel in bad and bared with that of beard [biad] in some New York City speech). To the extent that such efforts to retard change are effective, they may be said to exemplify socially conditioned nonchange of language.
What is more typical, however, is that efforts to control language change have only limited success. Even in written English, which has been subjected to efforts at prescriptive control since the eighteenth century, changes of vocabulary, grammar, and spelling have taken place steadily; this is clear to anyone who now reads novels written in the eighteenth and nineteenth centuries. In spoken language, prescriptivism is even less effective, and numerous changes can be observed within an individual lifetime. Speakers of English nowadays say and write ice cream with no awareness that, in 1900, this form was considered a vulgar error for iced cream; they also say, and sometimes write, ice tea for iced tea, as well as can peaches and smoke fish for canned peaches and smoked fish. Phonological changes
by which American English whine [hwain] merges with wine [wain], or caught [lo:t] with cot [kat], or pin [pin] with pen [pen] - or, indeed, Mary['meiji] and many['rnaaii] with merty['meii] - are common over broad geographical areas. In Great Britain, by contrast, the merger of whine/wine is standard, but the others are rare. Furthermore, such mergers are mostly below the level of awareness, and typically go uncensured by schoolteachers (who are likely to use the same pronunciations themselves).
The fact that language is universally changeable, and that it changes in different ways at different times and places, is of course the basic fact of historical linguistics. This is the reason behind the facts that Modern English is spoken in different ways in London, New York, Cape Town, and Sydney; that Modern English shows many differences from, as well as similarities to, the Old English of King Alfred the Great, with which it is mutually unintelligible; that French and Spanish are mutually unintelligible with each other and with Latin, although the modern languages show systematic correspondences with Latin; and that a more remote level of systematic correspondences involving Old English and Latin - as well as Ancient Greek, Sanskrit and other languages - enables us to relate those languages historically as "sisters" to each other, and as descendants of a prehistoric language that we call Proto-Indo-
European.
Types of influence In view of the above, it is understandable that linguists have wanted to understand the reasons why linguistic change occurs. Certain types of changes - involving a more or less simultaneous effect on large groups of people, up to entire societies - may be called macrolinguistic, these involve entire language structures, and often involve deliberate, conscious decisions, institutionally promulgated as part of language planning programs. One such process is that of standardization, in which a single dialect is put forward as the official norm for an entire multidialectal area. Again, when languages come into contact on a large scale, such as Spanish and English in the US, bilingualism may become common (sometimes with encouragement from governments and schools); this is likely to produce such typical language contact phenomena as code-switching between Spanish and English, the introduction of loanwords from one language into the other, and the assimilation of grammatical patterns toward those of the language to which social value is attached (in this case, English). A further result in some cases may be the limitation of Spanish language use to more restricted social contexts (e.g., the home), even to the point of the obsolescence of Spanish in some communities, and ultimate complete language shift in the direction of English. The ultimate stage of obsolescence is, of course, language death. Still other changes which may be called macrolinguistic, although they do not involve institutional actions, are pidginization and creolization, in which contact between two or more languages - e.g., in the situation of a colonial plantation economy - results in a new language with vocabulary mainly derived from the socially dominant language, but with a drastically simplified grammar.
On a more microlinguistic level, linguistic changes may be initiated by a single individual, or by a small group, and subsequently imitated by others who attribute social value to them; in some cases, such innovations may spread through an entire society. In the case of new vocabulary items, the motivation may be conscious, in the form of a new concept or invention such as radar, for which the English term was coined in the twentieth century; in such cases, the person who initiated the item, and the circumstances of its spread, are often well known. However, such new vocabulary items are far from typical of linguistic changes in general. When unconscious changes occur in grammar, as when the older plural kine is replaced by cows, or in pronunciation, as in which [hwitf] > [witf], we cannot pinpoint the initiating individuals or the paths of imitation. It is precisely the difficulty of discovering when and how such changes have occurred in the past, or of "catching them in the act" in the present, that has captured the imagination of many linguists and led them to study the mechanisms of language change.
Two types of misapprehension have often put obstacles in the way of this study. First, it used to be thought that it would never be possible to capture unconscious language change "in the act," simply because its operation required too long a period of time. In this view, trying to observe language change is like trying to observe the motion of a clock's hands: One cannot see the change, but if we look again later, we perceive that change has occurred. Second, people have sometimes thought of language history in terms of abrupt changes from one literary period to another, such as the change of Middle English soote to Modern English sweet. However, as has been pointed out in particular by William Labov, whose research on language change and its causes has been especially important, we can observe language change operating in "apparent time" (1972: 275), simply by listening to the speech of three generations living in the same household. For example, in many American families, members of the oldest generation never merge the vowels of caughtand cot, but their children do so sometimes - most often when they are speaking informally - and their grandchildren do so always (see also Bailey et al., 1991). Furthermore, phonetic change does not take place abruptly between one historical period and another, but rather occurs in small gradations, with socially conditioned variation among alternative forms. Thus the locus of language change is not in large abstractions like "the English language," but rather in the variable daily use of individual speakers over time. If we study the speech of a single individual, the variation that we find may seem to contain a great deal of randomness; but as Labov has shown, if we undertake statistical and comparative study - involving multiple speakers, social contexts, generations, and geographical locations - it is possible to discover coherent patterns of variation and change that characterize contemporary spoken English. Of course, confirmation of linguistic change in real-time studies remains desirable.
a universal tendency toward the use of minimum effort. Prescriptivists have even characterized the processes of change as resulting from laziness. In these terms, English whale merged with wall in many dialects because people found it "too effortful" to pronounce the h; caught merged with cot because some speakers found that the vocal calibration required to distinguish the vowels was just too much trouble. But since such changes have always been in operation, we would expect that they would have ultimately reduced all speech to the easiest possible sound, perhaps [aIII] - which has not happened. In fact, two other overriding motives are characteristic of language change: Speakers want to be able to understand each other, and they want to use language to express their social identity. All speakers frequently produce inadvertently simplified pronunciations of particular words; but in most cases, this low-level variability remains unimportant. The innovations are not repeated on later occasions by the original speakers, nor are they imitated by other speakers.
Again, in grammar, we can discern a tendency to simplify structure by analogy:Since English has / don't, you don't, we don't, they don't, we can save the trouble of mastering an irregularity if we replace he doesn't by he don't. But in fact he don't is stigmatized by most English speakers as rustic or uneducated, and there seems little chance of its becoming widely accepted. Furthermore, it is easy to find grammar changes which increase structural complexity. For instance, most English verbs have "weak" past tenses, formed by suffixation as in walk/-ed; but a minority are "strong," with vowel change, as in sing/sang. The tendency to simplify grammar by analogy should change verbs from the strong to the weak pattern, and it is a commonplace that children learning English often produce "incorrect" verb forms such as swimmed. But more complex types of analogical change, from weak to strong, also occur, as in the currently used sneak/snuck and squeeze/squoze. Complex irregularities remain a feature of English grammar.
Functional factors in language change It has been proposed, especially by Andre Martinet (1955), that phonological change in particular is constrained and guided by the need to preserve communicative function in language. This view is associated in particular with arguments for patterns of internal change in language, involving chain shifts, in which one change is followed by others which serve to preserve contrasts that distinguish meaning; thus in Swedish, the fronting of u to high-mid [e] was followed by the raising of o to [u], and this in turn by the raising of a (originally [3]) to [o] (Hock, 1991: 156-7). Such patterns are discussed in detail by Labov (1994), in volume 1 of his Principles of Language Change. However, it is clear that, whatever the importance of language-internal factors may be, a major role in language change is played by sociolinguistic factors; these have formed a central area of Labov's research for many years, and are the topic of his forthcoming volume 2.
The role of imitation A long-standing hypothesis in discussions of language change has been the idea that, once a change is initiated by a single individual (for whatever reason), its subsequent spread throughout a language community occurs when, and to the extent that, it is imitated by other speakers. This process indeed seems to operate in the area of new vocabulary, in cases where one person coins a new technical term (or creates a colorful new slang usage) and is then imitated by others. The way the process operates in the less conscious areas of phonology and grammar, however, is not so clear. There are anecdotes about how a change of [s] to [0] in Castilian Spanish arose because a king of Spain had a lisp, or of how a French [r] changed to uvular [K] because a French king had a speech defect - in each case the populace is supposed to have imitated the prestige of their monarch's speech - but these stories are hard to authenticate. In any case, such proposed explanations say, in effect, that the innovating pronunciation was borrowed; but linguistic borrowing is not always characterized by complete regularity across all vocabulary items, which is observable in the Castilian and French cases. In addition, the notion that imitation follows prestige models raises the question of how prestige itself is defined. It has sometimes been too easily assumed that prestige is any quality that people imitate! In fact, it is often observed that higher social classes adopt features of lower-class speech, as in many cases where middle-class White speakers have borrowed slang vocabulary from lower-class Black speakers; for some individuals, the borrowed features have connotations of masculinity or of natural authenticity - a kind of "inverse prestige." (Although many lower-class members assign low value to their own speech, they often feel that it would be "sissified" for them to imitate the upper classes.) Thus it is necessary to recognize both "change from above," where the language of higher social strata is dominant, and "change from below," where the model is the language of lower social strata. In short, although imitation must indeed be an important factor in language change, the concept is still not fully understood.
Grammatical change and phonological change.
Two major factors in grammar change are often cited. First, phonological change often does away with distinctions among morphological elements, or may delete them entirely; then new syntactic constructions may be elaborated to take over the functions formerly served by morphology. Thus the noun case suffixes of Old English were phonologically reduced and lost (except for genitive s); in compensation, the language developed rules of word order to differentiate subject from object, which earlier had distinct suffixes, and increased the use of prepositions to specify other relationships involving nouns. A second factor in grammatical change is the operation of analogy, discussed above; although it may not be possible to predict exactly when or how analogical processes will operate, their role is clearly important. In addition, some recent research on syntactic change indicates that external social forces may be important (Kroch, 1989).
Sociolinguistic motivations for change The research which opened the way for much subsequent work on social factors in phonological change was the study carried out by Labov (1963) on Martha's Vineyard, an island off the coast of Massachusetts. There he found that a centralization of [a] toward [a], in the diphthongs [ai] and [au], was a social marker of loyalty to the Island community, as opposed to the outside world. In subsequent much more detailed work in New York City (Labov, 1966c), Labov showed that what had appeared as random phonetic variation, when studied in the speech of individual New Yorkers, was statistically patterned when correlated with, on the one hand, social class, and, on the other hand, the degree of formality in speech. The latter factor could be categorized on a scale which included five styles, in order of increasing formality: (A) casual style, (B) careful style, (C) reading aloud from text, (D) reading aloud from word-lists, and (D') pronunciation of minimal word-pairs. Furthermore, he found that higher rank on the scale of social class was correlated with higher degrees on the scale of formality, both favoring conservative pronunciations; alternatively, if an innovation occurred more frequently in working-class speech, it would occur more frequently in the informal speech of all speakers. This kind of quantitative social dialectology, which has become almost synonymous with sociolinguistics for some people, is often discussed under the label variation theory.
One social class in particular was found to play a special role, namely the upward-aspiring lower middle class. This social group demonstrated hypercorrect behavior in the sense that, when speaking formally, they went beyond the highest-status group in adopting new prestige features (Labov, 1966b). Other behavior patterns typical of the lower middle class were a wide range of variation among styles
in speech, a high degree of phonetic fluctuation within a given style, and a conscious striving for "correctness." Perhaps the most striking feature of what Labov has called the linguistic insecurity (1972: 117) of this group was demonstrated when they were asked to give subjective evaluations of their own speech: They were strongly negative about it. Yet, as was later shown clearly in Labov's Philadelphia research (see below), the lower middle class occupies a key role in the processes of language change in the urban eastern US.
The interaction of class stratification and style with the dimension of apparent time, as seen in change over generations, is shown with reference to syllable-final r in figure 5.1. In Style A (casual speech), for the two oldest age groups (50-75 and 40-49), there is little indication that the occurrence of ris significant. However, among speakers under 40 years old, r suddenly becomes a prestige marker for class 9 only (upper middle class). This sudden change in the status of ris apparently associated with the population changes accompanying World War II. Reading downward in the table, from more casual to more formal styles, we find a regular increase in the use of r. The larger left-to-right pattern shows the role of age group in two different ways: In class 9, younger speakers show more use of rthan older speakers, but in the other classes, older speakers tend to use r more than younger ones. Reading from left to right, we see the familiar pattern of class behavior by which the lower middle class (levels 6-8) leads the working class and the lower class in the use of r.
C, reading; D, word lists; D', minimal pairs. Within each style, the vertical scale shows index scores (maximum 100) for the occurrence of syllable-final r. The horizontal scale shows four age categories, and within each of those, socioeconomic class levels 0-8. Class level 9, the upper
middle class, is indicated by the dotted line. The hatched areas represent the degree to which a given index score exceeds the level of the upper middle class. (Adapted from Labov, 1972, fig. 4.3, p. 116 and fig. 5.4, p. 137.)
middle class, is indicated by the dotted line. The hatched areas represent the degree to which a given index score exceeds the level of the upper middle class. (Adapted from Labov, 1972, fig. 4.3, p. 116 and fig. 5.4, p. 137.)
The hatched areas in the figure point up the phenomenon of hypercorrection. Although all class and age groups tend to use more ras formality increases, it is lower middle-class speakers (class levels 68) of the middle-aged group (40-49 years) who show the greatest tendency to increase their use of r in formal styles - until, in the most formal styles D and D', they far surpass the level of the upper middle class (class 9). But there is a generational difference in the middle class: Younger speakers seem not to be fully aware of the prestige attached to the new r pronunciation, and have not acquired it to the
same extent as their elders (Labov, 1972a: 59). The overall pattern shown by this figure reflects the relatively recent introduction, and subsequent spread through generations, of syllable-final ras a sociolinguistic prestige marker.
From such research by Labov and his colleagues, a new sociolinguistic approach to language change emerged in the 1960s, and was put forward in a classic essay by Weinreich, Labov, and Herzog (1968). The main features of this approach (conveniently summarized by Hock, 1991: 648-9) include the following: All speech involves low-level variation in pronunciation, with no consistent function. But sometimes, for linguistically arbitrary reasons, a feature of pronunciation may become associated with membership in a social group. At this point, the feature becomes important for speakers'knowledge of their language; they arrive at a generalization - a "rule" from the linguist's viewpoint - as to how the variable use conveys socially relevant information. This generalization may then be extended to new environments, new word classes, or related segments. At the same time, the social parameters for the generalization of the variable rule may be expanded to include additional individuals. If there are no opposing social pressures, the rule may expand throughout the entire lexicon and the entire speech community; if it achieves maximum expansion, it changes from variable to categorical status, so that a regular sound change can be recognized. Labov (1972a: 275) introduced the term uniformitarian principle for the hypothesis that this pattern of sound change implementation, generally observable in changes currently in progress, must have applied to all sound changes in past history.
In subsequent years, Labov and his students have concentrated on linguistic variation in the Philadelphia area, moving out from there to other parts of the US and the English-speaking world. Important collections of papers on the social and geographical dialectology of English, and on current changes in English phonology, have been published by Labov (1980b) and Eckert (1991).
The actuation problem The heart of Labov's system is the question of actuation: "Why do changes in a structural feature take place in a particular language at a given time, but not in other languages with the same feature, or in the same language at other times?" (Weinreich et al., 1968: 102). Labov has approached this question "by searching for the social location of the innovators: asking which speakers are in fact responsible for the continued innovation of sound changes, and how their influence spreads to affect the entire speech community" (Labov, 1980a: 252). Earlier linguists had suggested that innovation should originate more often among the lower social classes, because of their lesser exposure to the influence of the standard language - or alternatively, that innovation should originate among the upper classes, who would then provide a prestige model for lower classes to imitate. However, work in Philadelphia by Labov and his associates revealed that sound change could originate in any class otherthan the highest, and that the principal source of innovation was in fact located centrally in the social hierarchy (1980a: 253-4): namely, in that same lower middle class which had figured prominently in his New York research. But the riddle of actuation remained: What was the force that led to the continued renewal of sound change? It seemed evident that a key factor was the entrance of new ethnic and racial groups into the community; in Philadelphia this has been accompanied by strong emphasis on local ethnic identification, and by changes in both Black and White dialects which have the effect of making them increasingly divergent from each other (p. 263).
The influence of literacySince written language generally changes more slowly than spoken language, it has been supposed that the presence of literacy in a community might act as a "drag" to retard change in the spoken language. However, the picture is complicated by the fact that in societies with limited literacy, as in India, literacy tends to exist mainly in the upper social classes; so, although we find that Brahmin speech is relatively conservative, it is not clear what the relative roles of social class and literacy are in producing this effect. Bright and Ramanujan (1964) attempted to study this question by examining the speech of Brahmin and non-Brahmin speakers of Kannada and Tulu, two neighboring and closely related Dravidian languages of South India. Kannada has a long-established writing system, in which most Brahmins (and many non-Brahmins) are literate. By contrast, Tulu has no commonly used script; when speakers of Tulu as a first language (L1) learn to read and write, they do so in a second language (L2) - Kannada, Sanskrit, or English. Studies revealed that, in Kannada, the Brahmin dialect was clearly more conservative than non-Brahmin speech; but in Tulu, the Brahmin dialect was as likely to innovate as that of non-Brahmins. This suggests that literacy in Ll does indeed exercise a conservative effect, independently of social class. For more recent work on the relationship between literacy and sociolinguistic change see Toon, 1991.
between men's and women's speech (see chapter 8) and the question arises: What is the role of sex/gender differentiation in language change? The topic has recently been discussed by Labov (1990). He notes that, in a situation of stable sociolinguistic stratification, women are more conservative linguistically; they tend to favor variants with overt social prestige, whereas men do the reverse. However, in a situation of ongoing change - in which overt social prestige comes from outside the group - women tend to use a higher frequency of new forms than men; among women, it is the "hypercorrect" behavior of the lower middle class (as described above) which is especially important. Labov believes that women play a crucial role in change from below, precisely because of the sexual asymmetry of the care-giving situation: Most of the early language input received by young children is from mothers and other female caregivers.
Lexical diffusion It has been pointed out by William S.-Y. Wang and his associates (Wang, 1977) that phonological change sometimes seems to operate not in such a relatively rapid and sweeping manner as that envisioned by Labov, but rather through the slow borrowing or "diffusion" of individual lexical items between sister dialects. To the extent that sound change does proceed by lexical diffusion it is not gradual in the sense that it proceeds by phonetic gradations, but it spreads gradually through the lexicon and through social groups. When a large number of such borrowings occur, all involving the same phonological correspondence, we speak of a "sound change" as having taken place; however, numerous exceptions are typically found. Thus Early Modern English oo [u:] became modern [uw] in a large number of words, such as loom and boot, but in other words it becomes [uw] varying dialectally with [U] in other words (e.g., roof, root), or [U] alone (wood, book), or even [A] (flood, blood cf. Hock, 1991: 657). The linguistic and/or social factors that distinguish these two types of phonological change remain a matter of controversy.
Sociohistorical linguistics An approach developed by Suzanne Romaine (1982) focuses on the sociolinguistic analysis of historical texts, as illuminated by our understanding of language variation occurring in present-day societies. In accordance with the uniformitarian principle, it is held that language variation in the past must have been similar to that observable today. From this it follows that we should be able to reconstruct some of the sociolinguistic mechanisms which underlie the variation observable in philological study. In other writing, Romaine (1984: 30-1) has noted that, instead of studying variation in terms of statistics-based "predictive" rules - which in fact cannot tell us what variant an individual will choose on a given occasion, or why - we might consider a hermeneutic approach, aimed at understanding rather than predicting.
Social class and social network Whereas Labov has operated with social class as a key factor in sociolinguistic change, an alternative view emphasizes the concept of social network (cf. Blom and Gumperz, 1972). This approach has recently received increased attention and quantification O. Milroy, 1992; L. and J. Milroy, 1992). In this framework, an individual's social network is definable as the sum of relationships which he or she has contracted with others, and these may be spoken of in terms of relatively strong or weak ties. The Milroys'research suggests that strong ties within communities result in dialect maintenance and resistance to change; but individuals who have large numbers of weak ties outside the community tend to be innovators, and to serve as instigators of language change. This approach does not replace the concept of social class, but coexists with it in a two-level theory: Small-scale networks, in which individuals conduct their daily lives, coexist with larger-scale social classes, which determine relationships of power at the institutional level.
Conclusion The importance of sociolinguistic variation in the processes of unconscious language change, especially phonological change, has clearly been established beyond doubt; historical linguists of the future cannot escape being, to some degree, sociolinguists. Nevertheless, many questions remain to be answered. When a variant pronunciation first acquires social value, given that its linguistic character is arbitrary, can we identify the exact social factors which cause that precise variant to take on such significance, rather than some other? When a new phonological element approaches the crucial stage of actuation, can we pinpoint the social factors which cause that particular element to move from variable to categorical status, rather than some other element? Whether or not these problems can be definitively solved, sociolinguists have shed much light on the processes involved. Here as in other areas, progress must be made not by collecting linguistic facts and looking for social correlates, or by collecting social facts and looking for linguistic correlates - or even by collecting sets of correlations between linguistic and social facts - but by attempting to formulate a comprehensive viewpoint in which scholars can deal with linguistic and social aspects of communication simultaneously, as indeed we all do in our everyday lives.
All thanks and no blame go to Florian Coulmas, Hans Hock, William Labov, Lise Menn, and Suzanne Romaine for helpful comments.
7. Dialect in Society. WALT WOLFRAM.
1 Introduction.
For as long as observations about language have been recorded, the symbolic function of dialect in society has been recognized. Over three thousand years ago, the sh versus s pronunciation of shibboleth in the Hebrew word for "ear of corn" was used to detect impostors from true allies among the fleeing Ephraimites who attempted to disguise themselves as Gileadites. As indicated in the Biblical account, the social consequences of the dialect difference were quite severe:
The present-day social consequences of dialect differences may not be quite as gruesome as those described in the account given in the Old Testament, but the diagnostic differentiation of social groups on the basis of dialect remains symbolically just as significant.
The term dialect is used here to refer to any regional, social, or ethnic variety of a language. The language differences associated with dialect may occur on any level of language, thus including pronunciation, grammatical, semantic, and language use differences. At first glance, the distinction between "dialect" and "language" seems fairly straightforward - dialects are subdivisions of language. However, on closer inspection, the boundary between dialects and languages may become blurry as simple criteria such as structural affinity or mutual intelligibility break down. Thus, many of the so-called dialects of Chinese such as Pekingese (Mandarin), Cantonese, and Wu (Shanghai), are mutually unintelligible in their spoken form. By the same token, Swedes and Norwegians are generally able to understand each other although their distinct cultures and literatures warrant their designation as different languages.
In a similar way, the notions of regional, social, and ethnic dialect are not nearly as obvious as we might assume at first glance. Speakers are at the same time affiliated with a number of different groups and their varying memberships may contribute to the variety of language they employ. Speakers located within the same geographical territory may be affiliated with quite different ethnic and/or social groups, and thus end up speaking quite disparate varieties even as they share a subset of regional language peculiarities. While it is certainly convenient to use the term dialect as we do here, to refer to the general notion of a language variety, more precise definition of the term relies on its correlation with the particular parameters of social structure that determine its existence in a given
Given the apparent inevitability of dialect differences and their widespread social recognition, it is somewhat surprising that the social patterning of these differences has not been accorded more systematic study in the examination of language and/or society. The methodical study of dialects per se did not begin until the latter part of the nineteenth century (Wenker, 1877; GilHeron, 1902-10), and the serious study of dialects in their social context did not begin until the 1960s.
As it has developed over the past several decades, the systematic investigation of dialects in society has challenged some of the established perspectives of both linguistics and dialectology. Linguistics, as it progressed in the second half of the twentieth century, focused on the formal structure of language as an abstract cognitive system, with little attention given to the kinds of variants that were central to the examination of dialect variation. In accordance with more formal descriptive and explanatory goals, the primary data base became native speaker intuitions vis-à-vis actual language usage because of the insight these intuitions could provide to the cognitive processes underlying language. From this perspective, the social context of language was considered outside the purview of an abstract, cognitively based model of language description.
At the same time, dialectology in the twentieth century became more aligned with geography and history as it focused on the distribution of particular variants in geographical space and time. Accordingly, isolated sets of dialectally diagnostic lexical and phonological items collected through the direct elicitation of single instances of forms became primary data (e.g., Kurath and McDavid, 1961; Orton, Sanderson, and Widdowson, 1978; Carver, 1987). The social significance of dialect variants was examined to some extent through the correlation of dialect variants with the background of interview subjects, but it was secondary. For example, in the various surveys of the Linguistic At/as of the United States and Canada (Kurath, 1939: 44), subjects were classified according to three different social types in terms of education and social contacts. However, the systematic examination of dialect in society lay beyond the goals of traditional dialectology.
The pioneering investigations of dialect in society by William Labov (1963, 1966, 1972a, b) clearly broke precedent with some of the reified traditions of both linguistics and dialectology in their assumptions, methodology, description, and explanation. For example, the investigation of language in its social context was seen by Labov (1966c) as central to the solution of fundamental problems in linguistic theory and description rather than as a specialized, interdisciplinary subfield combining distinct traditions of inquiry and description. Furthermore, the use of conversational speech data collected through the sociolinguistic interview was based on the assumption that naturally occurring speech reflected the most systematic data for the examination of language variation (Labov, 1972c) and on the assumption that the characterization of systematic variation should be integrated into the description of a language. And it was maintained that the description of language change and variation had to appeal to the role of language in its social context in order to achieve its ultimate goal of explanation.
The line of investigation that developed in social dialectology over the past several decades has altered in a significant way our fundamental understanding of the nature of dialect variation in society with respect to both the linguistic and social sides of the sociolinguistic equation. In the following sections, we consider the nature of dialect variation, the patterning of this variation within society, and the kinds of social consequences that obtain from the socially situated distribution of dialect.
2 The Nature of Dialect Variation
According to popular beliefs, dialect patterns are relatively straightforward and simple: All members of one group invariably use one particular dialect form while members of a different group categorically use another one. While this subjective impression is of sociolinguistic import, the objective reality of dialect distribution within society is far more complex and variable than this popular perception. On one level there is an intricate interaction between the systematic patterning of language and social structure; on another level, however, linguistic variation is inherent in the linguistic system, existing apart from the social meaning that may ultimately be assigned to it. In fact, one of the major shortcomings of traditional dialectology as it developed in North America and the United Kingdom over the past half century was its failure to come to grips with the underlying
linguistic-systemic principles that guided the organization and direction of much linguistic variation subsumed under the rubric of dialect. The commonly adopted premise in dialectology that "each word has its own history" (GilHeron, 1902-10) unfortunately often precluded the extended consideration of the internal linguistic-systemic principles that guide the orderly distribution of language variation. The following sections illustrate a couple of ways in which dialect variation is guided by the internal mechanisms of language systems. From that point, we proceed to show how such variation may distribute itself in society, and the kinds of social meaning that this variation is assigned.
2.1 The internal motivation of dialects
Following the "each word has its own history" edict of traditional dialectology, dialect differences have often been described as if they consisted of unrelated sets of items. Thus various phonetic productions of two different English vowels in assorted dialects of English, such as the /u/ of words such as boot and tube and the of words such as boughtand caught, would be viewed as structurally independent entities because they involve different phonological units within the system. Similarly, the use of the socially diagnostic English reflexive hisselfversus himse/fin Kirk liked hisselfand the subject-verb concord pattern We was down there yesterdayversus We were down there yesterdayare viewed as socially diagnostic items quite independent of each other since there appears to be no inherent structural relationship between these forms. While the patterned co-occurrence of forms such as these may be noted as a part of an overall dialect profile, their coexistence within a given dialect is viewed as arbitrary from a descriptive-theoretical perspective.
Such a viewpoint seems far too limited in its assessment of the nature of language variation and change that serves as the foundation of dialect differentiation. Furthermore, there is empirical evidence that argues for a set of underlying principles that guide dialect variation, or at least the tendencies of variation, which exist independent of dialect contact and diffusion (Chambers, 1993). For example, vernacular dialects of English throughout the world (Wolfram and Fasold, 1974; Cheshire, 1982; Bailey and Gorlach, 1983; Trudgill, 1990) with no apparent common diffusional source, share the feature of negative concord in sentences such as They don't do nothing to nobody about nothing. Such uniformity among vernacular dialects of English suggests that there are underlying, language-internal pressures that guide some types of dialect variation. In the case of negative concord, the predisposition of languages to generalize processes is a natural, internal mechanism that may account for the representation of this process among different, independent, vernacular varieties. A negative marking rule that specifies that the negative should be attached to a verb phrase element and all post-verbal indefinites is a more general rule than one that may select only one position for the placement of negation. The fact that other language adaptation situations manifest negative concord supports the contention that an underlying change and variation principle is at work in this case. For example, in both first language (Brown, 1973) and second language acquisition (Schumann, 1978), speakers go through a negative concord stage regardless of their normative target dialects.
In a similar vein, we may appeal to the process of analogical levelling, which exerts internally induced systemic pressure to align exceptional forms in conformity with dominant patterns, as the basis for explaining the widespread existence among unrelated vernacular varieties of English of the reflexive hisse/fwithin the paradigmatic set of my-/your-/her-/our-self(ves)or the extensive regularization of the past tense finite form of be in I/you/she/we/you/they was found. The underlying principles that govern dialect variation are essentially the same as those that govern language change in general (e.g., Kiparsky, 1989; Joseph and Janda, 1987; Kroch, 1989), but the ratification of variation and hence its identification as language change is derived from the social context in which the linguistic variation occurs. Although the social interpretation of language forms involved in variation may appear somewhat capricious, it is not altogether whimsical. For example, the fact that standard dialects typically include more marked language forms - items or structures less natural in their linguistic composition - than their vernacular counterparts may be related to the fact that prescriptive norms often require speakers to recognize language forms on a conscious level. The practical dialect consequence of this socially ascribed conscious attention to marked forms is the rejection of some natural linguistic changes that have resulted in unmarked forms (Kroch, 1978). For example, the persistence of the marked, irregular plural oxen instead of regularizing it to oxes can only be attributed to such socially ascribed, conscious attention.
Labov's delineation of vowel rotation alternatives in English (Labov, Yaeger, and Steiner, 1972; Labov, i991, 1994) is a prototypical illustration of how dialects shift their vowel systems in orderly and predictable ways which are then assigned social meaning. Given the nature of vowel production, it is convenient to view different vowels as occupying “phonetic spaces” in a continuum of vowel positions.
The notion of phonetic space is important because the shift of one vowel in phonetic space often has an effect on adjacent vowels. As one vowel moves (e.g., becomes higher or more backed in its phonetic position) phonetically closer to or further away from an adjacent vowel, the next vowel may shift its phonetic value to maintain adequate phonetic distance in relation to the vowel that has moved initially. A whole sequence of vowel rotation may thus be set in motion. The pattern of phonetic rotation in vowels, known as chain shifting or the push-pul/ chain, is actively involved in differentiating the current character of long vowels. In the southern vowel shift, the vowel of bedtakes on a glide, becoming more like beya’[bld]. Meanwhile, the front long vowels (the vowels of beet and late) are moving downward and somewhat backward, and the back vowels are moving forward. This scheme is represented in figure 7.2. Figures 7.1 and 7.2 show quite different rotational shifts in terms of directionality, but a unifying set of principles dictates the systematic movement of
vowels in both shift patterns. In chain shifts, several primary descriptive principles are:
Principle I: Peripheral vowels (typically long and tense) rise.
Principle ll: Non-Peripheral (typically short and lax) nuclei fall.
Principle Ill: Back vowels move to the front.
Various subsets of vowel rotations may be noted, as in the northern cities vowel shift and the
southern vowel shift, and vowels may change their status with respect to peripherality so that systems
may be altered rather drastically, but the underlying principles seem to be generalizable. While the
operation of the underlying principles is more detailed than that presented here, and the underlying
explanation of these descriptive shifts in terms of phonetic production and/or communicative
strategy needs explication, principles that apply generally to vowel subsystems illustrate in an
important way how dialect differences are sensitive to language-internal principles of organization
and change. The search for underlying principles guiding dialect change and variation does not
eliminate the need to view some aspects of dialect difference as isolated units (particularly with
respect to lexical variation), but the appeal to internal-systemic principles of change and variation has
taken the linguistic understanding of dialect description a giant step forward.
2.2. Systematic variability.
Another dimension that needs to be admitted into the perspective on dialects in society is the
systematic nature of variability. One of the important discoveries to emerge from the detailed study of
dialects over the past several decades is the fact that dialects are sometimes differentiated not by the
discrete or categorical use or nonuse of forms, but by the relative frequency with which different
variants of a form occurred. For a number of phonological and grammatical dialect features, it can be
shown that dialects are more typically differentiated by the extent to which these features are found
rather than the mere absence or presence of particular variants. For example, studies of the
alternation of -in [in] and -ing [i] in words like swimmin'or swimming show that, while practically all
dialects of English show this alternation, different dialects are distinguished by the relative frequency
with which we find -in'and -ing in particular language varieties. Thus we found in a study of speaker
representing different social classes in Detroit, Michigan, that the mean use of in'ranged from almost
20 percent use for speakers demographically defined as upper middle class to approximately 80
percent usage by speakers designated as lower working class (Shuy, Wolfram, and Riley, 1967). It is
important to note that ALL of the individual speakers exhibit variability between -ing and -in’. In the
study of variation, frequency levels are computed by first noting all those cases where a form like in’
MIGHT HAVE occurred (namely, an unstressed syllable), followed by a tabulation of the number of
cases in which -in'ACTUALLY occurred.
The fact that there is fluctuation between forms such as -ing and in'does not mean that the
fluctuation is random or haphazard. Although we cannot predict which variant might be used in a
given utterance, there are factors that can increase or decrease the likelihood that certain variants will
occur. These factors are known technically as constraints on variability. The constraints are of two
major types. First, there are various social or externa/factors which systematically increase or
decrease the likelihood that a particular variant will occur. For example, the reference to social class
above is an appeal to an external factor, since we can say that a speaker from the lower working class
is more likely to use -in'for -ingthan are speakers from other classes.
Not all of the systematic effects on variability, however, can be accounted for simply by appealing to
various social factors. There are also aspects of the linguistic system itself, known as internal factors,
that may affect the variability of particular forms apart from social constraints. Particular kinds of
linguistic contexts, such as the kinds of surrounding forms or the type of construction in which the
form occurs, may also influence the relative frequency with which these forms occur. The systematic
effect of linguistic and social factors on the relative frequency of particular forms can best be
understood by way of an actual case from phonology. Consider the example of word-final consonant
cluster reduction as it affects sound sequences such as st, na’, id, kt, and so forth in various English
dialects. The rule of word-final consonant cluster reduction may reduce items such as west, wind cold, and act to wes' win, col', and ac' respectively. The incidence of reduction is quite variable, but
certain linguistic factors systematically favor or inhibit the operation of the reduction process. The
linguistic factors or constraints include whether the following word begins with a consonant as
opposed to a vowel (more precisely, a nonconsonant), and the way in which the cluster is formed.
With respect to the phonological environment that follows the cluster, the likelihood of reduction is
increased when the cluster is followed by a word beginning with a consonant. This means that cluster
reduction is more frequent in contexts such as west coast or cold cuts than in contexts like west end
or cold apple. An individual speaker might, for example, reveal consonant cluster reduction in 75
percent of all cases where reduction could take place when the cluster is followed by a word
beginning with a consonant (as in west coast) but in only 25 percent of all cases where the cluster is
followed by a nonconsonant (as in west end). The important observation is that reduction may take
place in both kinds of linguistic contexts, but it is consistently FAVORED in those contexts where the
word following the cluster begins with a consonant.
As mentioned earlier, cluster reduction is also influenced by the way in which the cluster is formed.
Clusters that are a part of an inherent word base, such as windor guest, are more likely to undergo
reduction than clusters that are formed through the addition of an -ed suffix, such as guessed(which
ends in [st] phonetically - that is, [gest]) and wined(which ends in [nd] phonetically - that is, [waind]).
Again, reduction takes place in both types of clusters, but it applies more frequently when the cluster
is an inherent part of a word rather than the result of the -ed suffix addition. When we compare the
relative effects of different linguistic constraints on final consonant cluster reduction, we find that
some linguistic factors have a greater influence on variability than others. Thus, in some dialects of
English, the influence of the following segment (the consonant versus nonconsonant) is more
important than the constraint based on cluster formation type (e.g., non -edversus -edclusters).
In many cases, linguistic constraints on variability can be ordered differently across varieties. Table
7.1 presents a comparison of word-final cluster reduction for different dialects of English, based
upon a sample of speakers in each population. As seen in Table 7.1, all the varieties of English
represented here show clusters to be systematically influenced by the following phonological context
and the cluster formation type, although the proportional differences and the relationship between
the linguistic constraints is not always the same. In some cases, such as Standard English and
Appalachian vernacular English, the influence of the following consonant is more important than the
cluster type, whereas in other cases, such as southern white working-class and southern African-
American working-class speech, the influence of the cluster type is a more important constraint than
the following phonological context.
The analysis of linguistic constraints on variability can get much more sophisticated than the simple
kinds of “raw” frequency tabulations and comparisons introduced here (Guy, I993), but this overview
reveals the subtle and complex ways in which dialect differences are internally constrained, in terms
of both their qualitative and quantitative linguistic dimensions.
3. The Social Distribution of Dialect.
In many respects, describing the social distribution of language variation is dependent upon the kinds
of group affiliations, interactional relations, and sociocultural ideologies operating within a society.
The range of social factors and conditions that may be correlated with linguistic variation is quite
wide-ranging (e.g., Preston, I986a), but there is little agreement about a definitive set of social
factors that vary with linguistic variation or an underlying, unitary, social or sociopsychological
explanation of these parameters. Covariation between linguistic variation and social variation is
multifarious and multidimensional, as the individual members of a society are at once affiliated with a
range of overlapping groups with varying sociocultural ideologies, assume a variety of functional roles
within and across groups, and participate in an assortment of interactional situations. In early studies
of dialect in society (e.g., Labov, I966c; Wolfram, I969) it was common for linguists to appropriate a
set of predetermined background demographic variables such as region, socioeconomic class,
ethnicity, age, and sex, and to show the covariance of linguistic forms with these variables, either in
isolation or, more commonly, in intersecting arrays. Later descriptions focused on the nature of
communication networks (L. Milroy, I980), the dynamics of situational context (Biber and Finegan,
I993), and the projection of social identity (LePage and Tabouret-Keller, I986) in an effort to
describe more authentically the social reality of dialect in society. (See chapters 8, 9, and I9 by Wodak
and Benke, Eckert, and Tabouret-Keller for a discussion of some of these factors.) For our purposes
here, it is sufficient but critical to recognize that many of the social variables typically appealed to in
studies of covariance are abstractions extracted from an intricate, interactive, and multidimensional
social reality. For example, McConnell-Ginet (I 988) and Eckert (I 989) point out that dialect
differences correlated with gender differences assume a social construction based upon the biological
category of sex. But the social construction of gender may be exceedingly complex, as it involves
roles and ideologies creating differential ways for men and women to experience life, culture, and
society. As Eckert notes, “there is no apparent reason to believe that there is a simple, constant
relation between gender and variation” (Eckert, I989: 247). Similar provisos could be offered for
virtually any of the traditional variables examined in the covariation of social and linguistic factors.
The perspective on dialect in society implied in the preceding discussion is an ethnographically
informed one, since only such a vantage point can reveal the local kinds of affiliations, interactions,
and ideologies that lead to the symbolic functions of dialect within a given community. Our recent
studies of Ocracoke, a post-insular island community located 20 miles from the mainland state of
North Carolina, reveal how an ethnographic perspective is needed to inform the role of dialect in a
community setting (Schilling-Estes and Wolfram, I994; Wolfram and Schilling-Estes, I995). In this
post-insular setting, some of the patterns of covariation between linguistic variables and traditional
background demographic factors and/or social networks do not match the patterns of covariation
that might be expected, based upon traditional sociolinguistic studies. For example, we do not find a
unilateral regression pattern between age or socioeconomic class and the loss of traditional island
dialect variants. Thus the use of a classic island variable, in which the aydiphthong of high and tide is
backed and raised so that it is pronounced close to the oyvowel of boyor Boyd, is often found more
frequently among a group of middle-aged, college-educated men who have a fairly wide range of
contact outside the local networks than among their older and younger cohorts. Furthermore, the
incidence of this island feature does not correlate neatly with the density and multiplicity of the social
networks of these middle-aged men. For example, several of the men who use the highest incidence
of this traditional island pronunciation are currently married to outsiders and have a fairly wide set of
social networks extending well beyond local community members. At the same time, our
ethnographic studies showed that these men are part of a highly symbolic, local “poker game
network” which consists of a small, indigenous group of men who meet a couple of times a week to
play poker. This group reflects strong, traditional island values, including the projection of island
identity through the use of symbolic dialect choices that include the now stereotypical oy production
in words such as hoi wide for “high tide.” In fact, islanders are often referred to as hoi raiders, in
reference to this production. While the descriptive details of the selective retention and enhancement
of the by production are much more intricate sociolinguistically than outlined here (see Wolfram and
Schilling-Estes, I995), the essential point is that our understanding of the local social meaning of
varying dialect forms is dependent to a large extent on understanding the local social relations,
ideology, and values of a community caught in the midst of a fairly dramatic transition from a self-
contained, marine-based economy to a vibrant tourist industry. This uneasy but symbiotic relation
between dingbatters, as outsiders are called by community members, and ancestral islanders, whose
genealogy on the island can be traced for at least several generations, is obviously manifested in the
curvilinear relationship shown in the distribution of dialect forms in apparent time. Group affiliation,
communication networks, social identity, and social context all come into play in determining the role
of this dialect variable in the Ocracoke community. But we can only ascribe social meaning to the
patterns of covariance between dialect variables and social variables as we understand the
sociohistorical background, the interactions, the ideologies, and the identities that define the local
social context of dialect.
3. Patterns of distribution.
Quite obviously, not all dialect structures are distributed in the same way within society. Given varying
histories of dialect contact, dialect diffusion, and internal dialect change, and the varieties of social
meaning ascribed to dialect forms, linguistic variables may align with given social groupings in a
variety of ways. The pattern of dialect distribution which most closely matches the popular perception
of dialect differences is referred to as group-exclusive usage, where one group of speakers uses a
form but another group never does. In its ideal interpretation, group-exclusive usage means that ALL
members of a particular community of speakers would use the dialect form whereas NO members of
other groups would ever use it. This ideal pattern is rarely, if ever, manifested in dialects. The kinds of
social grouping that take place in society are just too complex for this pattern to work out so neatly.
In some cases, distinctions between groups exist on a continuum rather than in discrete sets.
Furthermore, as we mentioned above, the definition of a social group usually involves a constellation
of characteristics rather than a single dimension, thus making the simple correlation of a linguistic
form with social structure intricate and multidimensional.
Notwithstanding the qualifications that have to be made when talking about group-exclusive dialect
features, there certainly are items that are not shared across groups of speakers. The essential aspect
of these dialect forms, however, seems to be the fact that speakers from other groups do NOT use
these forms rather than the fact that all the members of a particular group use them. Group-exclusive
usage is therefore easier to define negatively than positively. Viewed in this way, there are many
dialect features on all levels of language organization that show group-exclusive social distribution.
According to Smith (I 985), group-exclusive patterns of dialect distribution may be saturatedor
unsaturated. Saturated patterns refer to those that typify the vast majority of speakers within a
particular social group or speech community and unsaturated patterns refer to those that are less
pervasive, but still group-exclusive. For example, among younger working-class African-Americans,
the “habitual be” form with verb + ing as in They usua/iy be going to the movies might be considered
a saturated form since the majority of speakers in this group use this form at one time or another.
Note that the definition of the group in this case must include at least ethnicity, status, and age. By
the same token, speakers of other varieties of English do not typically use this construction. In
contrast, the use of the specialized future perfect construction be done as in The chicken be done
jumped out the pen for the same population of working-class African-American speakers might be
considered an unsaturated, group-exclusive form, since few, but only speakers of this variety, have
been found to use this construction.
Descriptive qualifications such as “saturated” and “unsaturated” group-exclusive usage are useful
approximate labels, but they have not yet been defined with any rigor. That is, the classification of a
form as saturated or unsaturated is not determined on the basis of a specific proportion of speakers
sampled within a given population (e.g., more than 75 percent of the speakers in a representative
sample use the form in saturated usage and less than 20 percent of the speakers use the form in
unsaturated usage). These designations are imprecise and limited, although admittedly convenient as
informal characterizations of dialect patterns.
Group-exclusive dialect forms may be taken for granted in one dialect while, at the same time, they
are quite obtrusive to speakers from other dialect areas. In American English, speakers from other regions may thus be quick to comment on how strange forms like youns “you pl.,” The house needs
painted, and gumbana’“rubber band” seem of them when visiting Pittsburgh, Pennsylvania, much to
the surprise of the lifetime resident of Pittsburgh who has assumed that these were in common use.
With increased interaction across dialect groups, however, speakers may become aware of some of
their own group-exclusive usages. As the consciousness about these forms is raised, some of them
may take on symbolic significance in identifying people from a given locale or social group. And from
these features come the stereotypes of particular regional and ethnic dialects found in popular
caricatures. However, it is important to remember that the stereotypical, symbolic caricatures by
outsiders, and sometimes even by insiders, are often not linguistically faithful to the actual use of the
form by speakers from the particular speech community.
In contrast to group-exclusive forms, group-preferentia/forms are distributed across different
groups or communities of speakers, but members of one group simply are more likely to use the form
than members of another group. For example, a fine-grained spectrum of color terms (e.g., mauve,
p/um, etc.) is often associated with females in the United States, but there are certainly males who
make similar distinctions, and, of course, there are females who do not make such refined color
designations. The association of a narrowly defined color spectrum with female speakers is thus
statistically based, as more women make these distinctions than men. We refer to this narrowly
defined color spectrum as a group-preferential pattern rather than a group-exclusive one. Group-
preferential patterns may derive from the nature of the dialect variable or the nature of the social
reality that underlies the social variable. For example, as we noted earlier, there are dimensions of
group affiliation, interactional relationship, and ideological perspective that make the social construct
of gender far more complex than a designation of group membership based solely on biological sex.
We would not expect the symbolic effect of a group-preferential pattern to be as socially distinct as a
group-exclusive marking, although popular stereotypes of group-preferential dialect patterns
sometimes treat them symbolically as if they were group-exclusive. The popular characterization of
vernacular dialects of English in their use of dese, dem, and dose for these, them, and those is such
an instance, where the stereotype of group-exclusive behavior actually betrays a fairly complex
pattern which is really group-preferential and also highly variable.
Variable use of socially diagnostic forms may also be distributed in different ways with respect to
social variables. For example, given a continuous axis of age, a particular linguistic variable may show
continuous unilateral regression between the incidence of a linguistic variant and age, that is, we find
a regular slope between decreasing age and the incidence of the variant. For example, the increasing
use of post-vocalic rfor Anglos in the American South (e.g, bearfor bea'or carfor ca) may show a
linear progression in which increasing age is coterminous with the increasing incidence of post-
vocalic r. This is sometimes referred to as fine stratification. Or the pattern may show sharp or
gradient stratification, in which there is an abrupt change in the relative incidence of a feature at
some point on a continuous social axis. For example, in the African-American community, there is a
sharp decrease in the incidence of subject-verb concord as in She go to the store for She goes to the
store at a midpoint in the social continuum that roughly divides the upper working and lower middle
class (Wolfram, 1969). Finally, a curvilinear pattern might be revealed in the distribution of a socially
diagnostic variable, in which the slope of the correlation line between linguistic variation and social
variation reverses its direction at some point in the correlation. For example, on the island of
Ocracoke, off the coast of North Carolina, the pattern of ay backing and raising of items such as high
tide shows that middle-aged men in the poker game network have a higher incidence of backing and
raising than their older and younger cohorts, thus showing a curvilinear pattern of distribution over
age. The different kinds of covariation patterns between dialect features and social factors are shown
in figure 7.3.
Stable linguistic variables defined primarily on the standard-nonstandard continuum of English tend
to be sharply stratified, whereas linguistic features undergoing change often exhibit fine stratification.
This is due in part to the role of social factors in language change within a community. Change tends
to start in a given social class and spread from that point to other social classes in a diffuse manner.
The kind of correlation that exists between social factors and linguistic variation may thus be a
function of both social and linguistic considerations; there is no single pattern that can be applied to
this covariation.
3.2. The social evaluation of linguistic features.
Although there is no inherent social value associated with the variants of a linguistic variable, it is not
surprising that the social values assigned to certain groups in society will be attached to the linguistic
forms used by the members of these groups. While this general pattern of social evaluation holds, the
correlation of particular linguistic variables with social stratification is not always so direct, as
sociolinguistic history molds the diagnostic role of language structures in various ways.
The use of particular language variants may be evaluated as socially prestigious or socially
stigmatized. Social/yprestigious variants are those forms that are positively valued through their
association with high status groups as linguistic markers of status, whereas social/y stigmatized
variants carry a stigma through their association with low-status groups. It is essential to understand
that stigmatized and prestigious variants do not exist on a single axis in which the alternative to a
socially stigmatized variant is a socially prestigious one, or vice versa. The absence of negative
concord in sentences such as She didn’t 0’0 anything, for example, in standard varieties of English is
not particularly prestigious; it is simply NOT stigmatized. On the other hand, there may be particular
patterns of negative formation that carry prestige in some varieties. For example, the choice of single
negative marking on the post-verbal indefinite negatives (e.g., He’// 0’0 nothing) rather than on the
auxiliary (e.g., He won’t 0’0 anything may be considered a prestigious option in some varieties of
English, but the alternative marking in the auxiliary is not considered stigmatized.
Figure 7.3 Possible relations of linguistic and social variables.
The discussion of sociolinguistic evaluation up to this point has assumed a particular vantage point
about norms of linguistic behavior, namely the perspective of widespread, institutional norms
established by higher status groups. These norms are overtly perpetuated by the agents of
standardization in society - language academies, teachers, the media, and other authorities
responsible for setting the standards of behavior. Community-wide knowledge of these norms is
usually acknowledged across a full range of social classes. Linguistic forms that are assigned their
social evaluation on the basis of this widespread recognition of social significance are said to carry
overtprest/ge. At the same time, however, there may exist another set of norms which relates
primarily to solidarity with more locally defined social groups, irrespective of their social status
position. When forms are positively valued apart from, or even in opposition to, their social
significance for the wider society, they are said to have covert prestige. In the case of overt prestige,
the social evaluation lies in a unified, widely accepted set of social norms, whereas in the case of
covert prestige, the positive social significance lies in the local culture of social relations. Thus it is
possible for a socially stigmatized variant in one setting to have covert prestige in another. A youth
who adopts vernacular forms in order to maintain solidarity with a group of friends clearly indicates
the covert prestige of these features on a local level, even if the same features stigmatize the speaker
in a wider, mainstream context such as school. The notion of covert prestige is important in
understanding why vernacular speakers do not rush to become standard dialect speakers, even when
these speakers may evaluate the social significance of linguistic variation in a way which superficially
matches that of their high-status counterparts. Thus widely recognized stigmatized features in
English, such as negative concord, nonstandard subject verb agreement, and different irregular verb paradigms, may function at the same time in a positive, covertly prestigious way in terms of local
norms.
There are several different ways in which speakers within the sociolinguistic community may react to
socially diagnostic variables. Speakers may treat some features as social stereotypes, where they
comment overtly on their use. In English, items such as ain’t, “double negatives,” and “dese, dem, and
dose”are classic features of this type on a general level, but particular dialects may have stereotypes
on a more local level. Thus the production of hoi toia’ers for high tiders has become a stereotype for
the island community of Ocracoke, the plural youns a stereotype for the city of Pittsburgh, and the
use of habitual be in They be doing itis rapidly becoming a stereotyped form for urban working-class
African-American dialects in the United States. Sociolinguistic stereotypes tend to be overly
categorical and are often linguistically naive, although they may derive from a basic sociolinguistic
reality. For example, the stereotype that working-class speakers ALWAYS use dese, dem, and dose
forms and middle-class speakers NEVER do is not supported empirically, although there certainly is a
correlation between the relative frequency of the nonstandard variant and social stratification.
Furthermore, stereotypes tend to focus on single vocabulary items or selective subsets of items rather
than more general phonological and grammatical patterns.
Another type of sociolinguistic role is assumed by the social marker. In the case of social-markers,
variants show clear-cut social stratification, but they do not show the level of conscious awareness
found for the social stereotype. Various vowel shifts, such as the northern cities vowel shift discussed
earlier, seem to function as social markers. There is clear-cut social stratification of the linguistic
variants, and participants in the community may even recognize this distribution, but the structure
does not evoke the kind of strongly evaluated overt commentary that the social stereotype does. Even
if participants don't talk about these features in any direct manner, there are still indications that they
are aware of their existence. This awareness is often indicated by shifts in the use of variants across
different styles of speaking.
The third possible sociolinguistic role is called the social indicator. Social indicators are linguistic
structures that correlate with social stratification without having an effect on listeners’judgment of
social status. Whereas social stereotypes and social markers are sensitive to situational variation,
social indicators do not show such sensitivity, as shown by the fact that levels of usage remain
constant across formal and informal styles. This suggests that the correlation of socially diagnostic
variables with social factors operates on a more unconscious level than it does for social markers or
stereotypes.
The social recognition and evaluation of dialects does not relate just to particular dialect variables but
to entire dialect communities. Research on perceptual dialectology(Preston, I986b) shows that
overall dialect perception is generated by linguistic differences, popular culture caricatures, and local
identification strategies. For example, caricatures of New York City speech make this a highly
recognized dialect area for virtually all American English speakers, regardless of their geographical
locale. At the same time, the perceptual location of other regional areas may be subjected to a
“proximity factor,” in which the more distant the dialect is geographically, the more likely it is to be
classified globally.
4 Dialects and Social Commitment
The preceding discussion has viewed the role of dialects in society primarily on a micro-level, as we
have examined the relations that exist between language variables and social variables. There are,
however, issues related to the broadly based position of dialects in society. In this final section, we
address some of these broader issues and consider the social role that sociolinguists can assume in
addressing concerns relevant to dialects in society.
We have assumed in our discussion that dialects will continue to flourish in contemporary society, but
many popular accounts of dialects question their enduring vitality. For example, in the United States it
is often reported that dialects are levelling because of the widespread exposure to a standard,
relatively homogenized dialect through the media, the increase in interregional travel and migration,
and ready transportational access to virtually any dialect area of the country within a matter of hours.
Our preceding discussion indicates that the future of dialect diversity is assured on both a linguistic
and social basis. For example, we have seen that the pressures of internally induced linguistic
variation can take dialects in radically different directions once a particular linguistic shift has been
initiated. The vowel changes taking place in the northern cities and southern vowel shifts in the
United States are apparently making these varieties more dissimilar than ever, and some of the
current rotational shifts may lead to changes in the English vowel system that are as dramatic as
those witnessed during the great vowel shift that took place from 1300 through 1500 (Labov, 1994).
On a social level, the persistence of social dissonance of one type or another and the apparent
inevitability of asymmetric social grouping underly the maintenance of socially and ethnically defined
varieties, although some relic dialects preserved historically through their insularity have been levelled
significantly as they emerged from their isolated status. The vertical social axis underlying the
standard-nonstandard continuum remains operative, thus ensuring the continued vitality of
vernacular dialects in most societies. External social conditions added to internal linguistic conditions
assure the future of robust dialect differences.
The fact that so many dialect differences are defined on a vertical social axis naturally leads to a
symbolic sociopolitical and socio-educational role for dialect. The socially constructed perception that
vernacular dialects are deficient linguistic systems carries with it an attendant set of attitudes and
behaviors that can impact significantly on the social and educational lives of underclass, vernacular
speakers. The failure to recognize the legitimacy of dialect differences may lead to a kind of
discrimination that is as onerous as other types based upon race, ethnicity, or class. Unfortunately,
the existence of dialect discrimination is still not considered to be on the same plane as other types of
prejudice.J. Milroy and L. Milroy (1985: 3) note: “Although public discrimination on the grounds of
race, religion and social class is not now publicly acceptable, it appears that discrimination on
linguistic grounds is publicly acceptable, even though linguistic differences may themselves be
associated with ethnic, religious and class differences.” Dialect discrimination cannot be taken more
lightly than any other case of potential discrimination, and there is now a precedent for litigation
based upon such discrimination. A society that assumes responsibility for combating discrimination
on the basis of race, ethnicity, class, sex, and age should feel obligated to extend the same treatment
to dialect differences.
The relatively short history of social dialectology has shown that it is quite possible to combine a
commitment to the objective description of sociolinguistic data and a concern for social issues
relating to dialect. At various junctures over the past three decades, sociolinguists have become
involved in several important sociopolitical and socio-educational issues related to dialect diversity.
According to Labov (1982), there are two primary principles that may motivate linguists to take social
action, namely the princip/e of error correction and the princip/e of debt incurred. These are
articulated as follows:
Principle of error correction.
A scientist who becomes aware of a widespread idea or social practice with important consequences
that is invalidated by his own data is obligated to bring this error to the attention of the widest
possible audience (Labov, 1982: 172).
Principle of debt incurred.
An investigator who has obtained linguistic data from members of a speech community has an
obligation to use the knowledge based on that data for the benefit of the community, when it has
need of it (Labov, 1982: 173).
There are several outstanding instances in the history of social dialectology where these principles
have been applied. In the 19605, sociolinguists in the United States took a prominent pro-difference
stance in the so-called deficit-difference controversythat was taking place within education and
within speech and language pathology (Baratz, 1968; Labov, 1969). Consonant with the principle of
error correction, sociolinguists took a united stand against the classification and treatment of normal,
natural dialect differences as language deficits or disorders. There is little doubt that sociolinguists
played a major role in pushing the definition of linguistic normalcy toward a dialectally sensitive one,
although the practical consequences of this definition are still being worked out in many clinical and
educational settings (Wolfram and Adger, I993).
In keeping with the principle of debt incurred, social dialectologists also rose to the occasion in the
celebrated Ann Arbor decision (I 979). Linguistic testimony was critical toJudge Joiner's ruling in favor
of the African-American plaintiff children who brought suit against the Board of Education for not
taking their dialect into account in reading instruction. In effect, the judge ruled that the defendants
had failed to take appropriate action to overcome language barriers, in violation of Title 20 of the US
Code, Section I703 (f). In compliance with the judge's ruling, a series of workshops was conducted to
upgrade awareness and to apply sociolinguistic expertise in reading instruction.
There is another level of social commitment that sociolinguistic investigators might adopt toward the
dialect communities in which they conduct their research. This level is more positive and proactive, in
that it involves the active pursuit of ways in which linguistic favors can be returned to the community.
Thus I propose an additional principle of social commitment which I call the princio/e of/ingui5tic
gratuity.
Principle of linguistic gratuity.
Investigators who have obtained linguistic data from members of a speech community should actively
pursue positive ways in which they can return linguistic favors to the community.
This principle seems to be a reasonable extension of social commitment on the part of linguists.
However, this level of social responsibility is not restricted by a qualification based on recognized
community needs, as is Labov's principle of indebtedness. Instead, it is committed to a creative search
for a community-based collaborative model to return linguistic favors - favors that accurately reflect
the role of dialect in society vis-a-vis popular stereotypes. This may be done through popular books
and articles about community dialects, work with preservation societies in collecting and archiving
dialect data, and the development of dialect awareness curricula for community schools (see Wolfram
and Schilling-Estes, I995). Language is, in many ways, the most sacred of all cultural traditions and is
the rightful property of its users. Those who study dialect therefore need to be sensitive to the
symbolic role of language and to preserve this unique artifact that has been shared with us by
archiving for present and future generations of speakers the rich legacy of community dialects. While
social dialectologists may not view themselves as agents of social change, they do have a
responsibility to share the truth about dialects and to address the social consequences that derive
from the failure to understand the fundamental nature of dialect in society.
Cite this article
WOLFRAM, WALT. "Dialect in Society." The Handbook of506io/inguistics. Coulmas, Florian (ed). Blackwell
Publishing, I998. Blackwell Reference Online. 28 December 2007
<http://www.blackwellreference.com/subscriber/tocnode?
id=g978063121I938_chunk_g978063121I9389>
Bibliographic Details
The Handbook of Sociolinguistics
Edited by: Florian Coulmas
eISBN: 978063121 I938
Print publication date: 1998
