Exploring variation in language

Given that all living languages are subject to variation, it is surprising that the subject was for so long ignored or downplayed by mainstream linguistics. In idealizing the ‘homogeneous speech-community’ for his own purposes, Chomsky (1965: 3), for example, was merely maintaining the prevailing assumption that variation was of little theoretical interest. Studying the relationship between language and society remained something of a taboo until the 1960s, when researchers in the emergent discipline of variationist sociolinguistics argued that no satisfactory account of linguistic change could be achieved without a proper understanding of how variation was structured.

Our focus in this chapter is on variation within a language, or microvariation, and we look first at the approach of nineteenth- and twentieth-century dialectologists. Their methodology and assumptions differed greatly from that of modern-day sociolinguists, whose work we examine later in the chapter. Armed with modern recording equipment and applying sociological concepts and methodology, sociolinguists have shown the close relationship between language and a range of extralinguistic or social factors.

Dialectology

Dialectology, the study of geographical differences within the same language, has a long tradition that predates modern linguistics. In the nineteenth century in particular, data from the local and regional dialects of Europe were collected with a view to establishing language families and identifying the branches of family trees (sometimes, it must be said, from nationalistic rather than purely scientific motives).

In the absence of reliable recording equipment, obtaining information about how language varied from one place to another was difficult, and researchers were often reliant on impressionistic data collated from non-specialists. Marburg-based dialectologist Georg Wenker’s early attempts to document spoken dialect across Germany, for example, were based on some 45,000 questionnaires returned from schoolmasters between 1877 and 1887: the sheer volume of data meant that only a fraction of Wenker’s corpus was ever properly exploited.

In the last decade of the nineteenth century Jules Gilliéron attempted to obtain more reliable first-hand data by training a grocer, Edmond Edmont, to conduct dialectological interviews with informants in 639 rural French villages, using a simple pre-IPA transcription system to record his results. Edmont’s findings were published as the Atlas Linguistique de la France (ALF; ‘Linguistic Atlas of France’) between 1902 and 1910. Half a century later, a team of researchers led by Harold Orton at the University of Leeds used similar methodology to Gilliéron, and again focused largely (though not exclusively) on rural villages for the Survey of English Dialects (SED), published between 1962 and 1971. The detailed data on local variation in each of these surveys have been presented as dialect maps, and in some cases isoglosses have been drawn, separating areas using one form from areas using a different one for the same referent (see the example for the verb ‘to peep’ in Figure 11.1).

Figure 11.1: Isoglosses for ‘peep’ (vb) based on SED data (Upton, Sanderson & Widdowson 1987: 136)

It is important to remember that the primary aim of the dialectological surveys was to collect and record local variants before they died out, not to provide an accurate snapshot of variation throughout the country. The bias towards rural English villages in the SED, for example, was consistent with the aim of locating conservative speech forms, but entirely unrepresentative of a country which had been predominantly urban since the mid-nineteenth century. The informants selected, similarly, were anything but a representative cross-section of the English population at the time. Because of their associations with traditional (and often dying) trades, and the specialist vocabulary that went with them, NORMs (non-mobile older rural males) were targeted as ideal SED informants, as they had been by the ALF, for which only 60 out of some 700 informants were female.

Key idea: Dialectical studies

Traditional dialectology focused exclusively on the geographical dimension, while urban variationist studies have focused on a single city, and explored correlations between speech and extralinguistic factors (e.g. gender, social class).

It would be unfair to criticize traditional dialectologal surveys for not being representative of the general population: they did not set out to be so. But we do need nonetheless to interpret their findings with caution for other reasons. Firstly, fieldworkers asked informants which forms they used, but self-reporting is notoriously unreliable. The form an informant offers a fieldworker may not in fact be the form he/she uses most often, even though he/she may sincerely believe that it is. The method, moreover, makes little allowance for intra-speaker variation: all speakers vary in their usage, and the language one feels appropriate for answering questions posed by a stranger undertaking fieldwork is likely to be different from that used with intimates.

For these and other reasons, the isoglosses of dialect maps need to be seen as an idealization of data in which there are gradual changes over geographical space rather than abrupt boundaries between uses. Dialectological surveys provided a wealth of information about variation on a single dimension, that of geography, while keeping all other variables (age, sex, education, etc.) broadly constant. Their findings often yielded insights into the direction of change, but sociolinguists seeking to understand how change occurs had to take essentially the reverse approach, i.e. to hold the geographical variable constant by taking informants from a single place, and vary the other social variables. This was the task that two pioneers of variationist studies, William Labov and Peter Trudgill, set themselves in the 1960s and 1970s.

Urban sociolinguistics: methodology and problems

William Labov’s objective of investigating speech variation in a cross-section of speakers from New York City in the 1960s raised an immediate methodological problem: how could one gain access to the natural speech from informants who knew they were the subjects of investigation? How, in other words, could what became known as the Observer’s Paradox be overcome? An ingenious early response can be seen in his pilot study, undertaken with a team of researchers in three New York department stores: Saks, Macy’s and Klein’s, which could be graded as high, middle and low status respectively on a number of external criteria (e.g. prices of similar display goods, range of goods offered, publications in which the stores advertised).

Labov focused on a single linguistic variable (i.e a speech form known to be used variably within a community), namely non-prevocalic /r/, which may be deleted in New York. The variable (r) (sociolinguistic variables are conventionally placed in round brackets) therefore had two variants, labelled (r)-0 and (r)-1:

• (r)-0: \[Ø]

• (r)-1: \[r]

Case study: The speech community

In Labov’s department stores experiment, speakers used more (r)-1 pronunciations as the prestige of the store increased, and individually they used more of these pronunciations when repeating the words fourth floor, i.e. when they were paying more attention to their speech. So while they didn’t all speak in the same way, they did at least agree on how they felt they should try to speak in careful style. Labov says this type of agreement is indicative of a speech community: while usage may vary considerably across its members, the community nonetheless shares speech norms (1972: 120):

The speech community is not defined by any marked agreement in the use of language elements, so much as by participation in a set of shared norms; these norms may be observed in overt types of evaluative behaviour and by the uniformity of abstract patterns of variation which are invariant in respect to particular levels of language.

Interestingly, if a similar experiment were done in a rhotic area of England (i.e. an area in which non-prevocalic /r/ is still pronounced), it is likely that the pattern observed in New York would be reversed, i.e. the high status store would use least (r)-1 and speakers would tend to delete them more in careful speech, because /r/-deletion rather than /r/-retention is prestigious in England. Sociolinguistic variation, then, lends the best support to the dictum that England and America are ‘two nations separated by a common language’.

The method adopted by Labov’s team was to identify in each store items known in advance to be on sale on the fourth floor, and approach shop assistants to ask where they could be found. When the expected answer fourth floor came it duly provided two instances (or tokens) of (r) in two different environments, the first in preconsonantal position and the second word-finally. By feigning not to have heard, the researchers could then elicit a repetition of the two words in what Labov called emphatic style, in which, it is reasonable to presume, informants would be speaking more carefully to ensure they were properly understood. The researcher would then note the four tokens in two environments and two speech styles, from an informant completely unaware of having taken part in a sociolinguistic experiment.

Key idea: Rapid anonymous observation

Ingenious techniques such as rapid anonymous observation were used in early variationist studies to overcome the Observer’s Paradox, i.e. the problem of obtaining ‘natural speech’ from informants who know they are under investigation. Sociolinguists have now largely abandoned the notion of ‘natural speech’, on the grounds that all speech is designed with an audience in mind.

The results, when collated and analysed, showed a remarkable correlation between the status of the store and linguistic behaviour, with most use of the prestigious (r)-1 variants occurring, in both environments, in the high-status Saks store and fewest in the lowest-status store, Klein’s. (r)-1 use increased consistently in the repeated ‘emphatic’ style, suggesting that speakers use more prestige variants when paying more attention to their speech (see Case study on p. 226).

Figure 11.2: (r)-use in three New York department stores (Labov 1972: 52; adapted by Wardhaugh 1998: 162)

Labov’s technique of rapid anonymous observation had overcome the Observer’s Paradox and demonstrated a clear, quantifiable correlation between speech and social status. A question we need to ask here, though, is ‘whose social status?’. For all its advantages, rapid anonymous observation yields very little information about informants themselves, beyond that which can be reasonably guessed, for example sex and approximate age. We therefore know little about the shop assistants’ own socio-economic status but, perhaps surprisingly, there is little reason to suppose that it actually corresponded to that of the stores where they worked: indeed, on one criterion, that of pay level, Macy’s rather than Saks employees were believed to be of highest status. In all likelihood, the shop assistants’ speech was a better reflector of the status of their customers’ status than of the assistants themselves. The assistants, in other words, seemed to be ‘borrowing’ the status of their customers by accommodating to them. As we shall see in Chapter 13, the concept of accommodation has important consequences for our understanding of linguistic change.

Urban surveys: New York and Norwich

The department store study was followed by two major urban surveys on either side of the Atlantic: by Labov himself in New York City, and by Peter Trudgill in Norwich. In both cases, a representative sample of people who had all lived in the city for some time was selected, on the basis of which informants were invited to take part in a sociolinguistic interview. Informants’ age and gender were noted, and index scores for socio-economic class were established for each informant on the basis of scales for a number of criteria, such as education level, occupation and income.

The interview itself was structured in such a way as to elicit a range of speech styles, so that intra- as well as inter-speaker variation could be measured. The early part of the interview in which personal data were gathered, for example, was presumed likely to elicit speech styles at the more formal end of a speaker’s repertoire, but rather less formal than those of reading styles, in which speakers’ attention could be variably drawn to their speech. In Labov’s minimal pair style, the informant is asked to focus very directly on the variables under investigation, e.g. guard and God, which are homophonous for some New Yorkers ([ga:d]); asking informants to read a word list maintained attention on individual words, but their capacity to self-monitor was reduced considerably when they were invited to read a passage of text, in which examples of the key variables had been liberally inserted. But could access to natural vernacular ever truly be obtained in experimental conditions?

Labov and Trudgill were both clear that the Observer’s Paradox could not be overcome as it had been in rapid anonymous observation, but argued nonetheless that it was possible to divert informants’ attention away from their speech and thereby elicit something akin to a natural, or ‘casual’ style. This could be encouraged by interviewing informants with family or friends, allowing digressions or interruptions (e.g. from telephone calls), and by the famous ‘danger of death’ question (see Case study below).

Case study: Dangerous New York, tranquil Norwich?

Towards the end of the interview, Labov would ask his informants whether they had ever been in a situation where they had genuinely feared for their lives (Labov 1966: 107):

Have you ever been in a situation where you were in serious danger of getting killed (where you said to yourself, ‘This is it!’)?

The question subtly diverts speakers’ attention away from their speech and directs it towards the telling of an exciting story: the speaker stands to look ridiculous if it turns out that there was in fact no real danger.

While not all New Yorkers had tales of this kind to tell, this approach generally worked well in New York, but failed dismally in Norwich, leaving Peter Trudgill to wonder whether Norvicensians simply led more uneventful lives than their New York counterparts. Trudgill’s solution – asking informants whether they had had a good laugh recently – worked rather better, while similarly exerting gentle pressure on informants to tell a good story, and thereby diverting their attention away from their own speech. The underlying methodological assumption that formality of speech style increases with attention to speech became known as the audio-monitoring hypothesis, which Labov (1972: 208) sets out thus:

‘There are a great many styles and stylistic dimensions that can be isolated by the analyst. But we find that styles may be ordered along a single dimension, measured by the amount of attention paid to speech. The most important way in which this attention is exerted is in audio-monitoring one’s own speech, though other forms of monitoring also take place.’

The methodology of the structured sociolinguistic interview has been challenged on a number of grounds, including the artificiality of the question-answer format, the non-comparability of scripted and unscripted styles, and the audio-monitoring hypothesis on which many of its assumptions rest. It is simply not true that speakers always use more formal styles when paying attention to their speech, and most sociolinguists would now argue that, since speakers are always tailoring their speech to a particular audience, the very notion of ‘natural vernacular’ is a misnomer. Nonetheless, the controlled experimental data which the New York and Norwich surveys produced are still noteworthy for the insights they yielded, for the first time, about the relationship between language and social factors, the most important of which we review below.

Variation by class and style

A striking finding of both surveys was the very clear pattern of class stratification in speech: use of prestigious or standard variants by each social class mirrors exactly the hierarchy determined by the criteria for the social class index. This is true, moreover, in all styles, though differences are less pronounced in formal styles. Figure 11.3 shows class stratification for the (ng) variable in Norwich, for which variation affects the present participle and gerund suffix -ing. Again, two variants were identified: the standard velar nasal [ŋ] ((ng)-1), and East Anglian vernacular [Ən] or \[

] ((ng)-2) of hunt’n, shoot’n, fish’n stereotype. An index score for each speaker was then calculated on the basis of a score of 1 for tokens of (ng)-1, and 2 for each token of (ng)-2, the total being divided by the number of tokens. The results for each class in each style were then plotted on a graph:

Figure 11.3: Norwich (ng) by class and style (Trudgill 1974: 92)

Aside from the clear pattern of stratification across the five social classes Trudgill identified in Norwich (lower, middle and upper working class; lower and middle middle class), two points are noteworthy from this graph:

1. The fact that the lines slope in the same direction is indicative that, for this variable at least, Norwich forms a speech community on the basis of shared norms (see Case study on p. 226). While the social classes differ in their behaviour, Norwich informants of all classes adjust their speech in the direction of more [ŋ] use in formal styles.

2. The graph lines are quite steep for this variable, indicating a significant difference between formal and informal styles: the Norwich speech community is clearly aware of the social significance of this variable, and when given an opportunity to monitor their speech, informants make an effort to avoid the non-standard variant.

3. A variable like (ng) which shows a high degree of social and stylistic variation is called a marker. In some cases, a vernacular variant of a marker becomes so well known as to be the subject of overt mockery by outsiders: these stigmatized variants or stereotypes (for example ‘Bristol l’: see Chapter 5), are then avoided by all but lowest-status speakers. Indicators by contrast, show class stratification but relatively little stylistic differentiation. A good example, again from Norwich, is the (a) variable in words like after, cart, path for which the standard form is a back vowel, but vernacular variants are fronted:

• (a)-1: [α:]

• (a)-2: [ä: - ä:]

• (a)-3: [a: - ã:]

Figure 11.4: Norwich (a) by class and style (Trudgill 1974: 98)

As we can see from Figure 11.4, the variable (a) shows the familiar pattern of social stratification but the slopes are much flatter, because there is little difference between formal and informal styles. This suggests that, although the Norwich speech community agrees on what the ‘correct’ form is (the lines again all slope in the same direction), they do not perceive the variation to be socially significant in this case.

Case study: Martha’s Vineyard

Labov has sometimes been criticized for viewing speakers as ‘sociolinguistic automata’, whose linguistic behaviour is entirely moulded by extralinguistic factors such as age, gender and social class. But as an early study from 1963 shows, he has always understood that variation and change could equally be driven by subjective factors.

The island of Martha’s Vineyard lies three miles off the coast of Massachusetts, and had a permanent population of around 6,000 people, swelled by some 42,000 summer visitors who, in Labov’s words, ‘flood the island in June and July every year’. These visitors were, for the most part, rather more prosperous than the islanders themselves, whose traditional industries of fishing and agriculture were in decline. Labov describes the island as a very desirable place to live, but life was nonetheless difficult for Vineyarders: their county, Duke’s, was the poorest in Massachusetts, and a fragile economy, coupled with island isolation, had produced high unemployment and living costs. It is hardly surprising, then, that there was resentment of wealthy ‘summer people’ who were buying up property on the island, and that Vineyarders themselves were torn between remaining on the island and seeking better economic prospects on the mainland.

A feature of Martha’s Vineyard dialect had been the use of centralized diphthongs in the NIGHT and HOUSE lexical sets. The local night [nƏIt] and house [hƏɷs] pronunciations appeared to have been losing ground to uncentralized mainland forms ([naIt] and [hƏɷs]) for some time. Labov’s investigation of what he called the (ay) and (aw) variables indicated, however, that the Vineyard forms were undergoing something of a resurgence, particularly among the 31–45 age group. The users of the centralized diphthongs were, however, predominantly people favourable to the island and who intended to remain there, in spite of the more limited opportunities it offered. This clear correlation between centralization and pro-Vineyard orientation can be seen in this table:

Table 11.1: Centralization and orientation towards Martha’s Vineyard (after Labov 1972: 39, Table 1.6)

Key ideas: Stratification and hypercorrection

•  Variation in urban studies has been found to be highly stratified, mirroring the class hierarchy.•  Cases of hypercorrection, in which intermediate social classes use more standard or prestige forms than the classes above them, have been seen to be indicative of change in progress ‘from above’ (i.e. from above the level of consciousness and in the direction of a prestige norm).

Language and gender

Of all the findings of modern sociolinguistics, none can have been more intensely debated than what has become known as the sociolinguistic gender pattern (SGP), set out by Trudgill in the following terms:

In all the cases examined, it has been shown that, allowing for other factors such as social class, ethnic group and age, women on average use forms which more closely approach those of the standard variety or the prestige accent than those used by men.Trudgill (2000: 70)

Spotlight: Hypercorrection

Another meaning of hypercorrection, and a frequent source of language-based comedy in popular drama, is the over-extension of a rule learned by a social climber to a linguistic environment where it does not apply. Before British English speakers were as aware as they are today of each others’ regional accents, it was claimed, for example, that northern English speakers living in the south and aspiring to the prestigious RP accent would pronounce some words like butcher with the southern STRUT vowel, i.e. as [bʌtƏ] rather than [bɷtƏ], because they erroneously assumed that all instances of /ɷ/ could be replaced by /ʌ/ in RP. While this strategy works fine for come and rub, it does not for butcher or pull, where northerners and RP speakers have the same vowel.

Dramatists have seen comic potential in the propensity of Cockneys – traditional London English speakers – to drop h at the beginning of words. Fans of the cult 1960s TV adventure puppet show Thunderbirds will recall, for example, how Parker, an ex-jailbird now working as manservant to the aristocrat Lady Penelope, would attempt to use higher status speech by re-inserting the lost initial h’s of Cockney English, usually in the wrong places, e.g. ‘I must apologize for the hunconventional entrance, m’Lady, but I ‘ad to happre’end ‘im some’ow’.

Trudgill is not, of course, claiming that women speak ‘better’ than men, nor, indeed, that men and women have different languages. Gender-based differences in speech, with the exception of those imposed by the grammar (for example, a male Russian says ja sidjel ‘I sat’ but a female would say ja sidjela), are generally a matter of more or less, with the genders using the same forms in different proportions. But women nonetheless consistently appear to use more prestige forms than men do. Macaulay’s (1976) data for the (i) variable in Glasgow, for example, suggest that women’s use of prestige variants corresponds broadly to that of men in the social class immediately above them:

Figure 11.6: Glasgow (i) by class and gender (data from Macaulay 1976; see Coates 2013: 55, Fig. 4.7)

Explanations for this remarkably consistent finding have appealed variously to women’s traditionally greater role in the rearing and education of children, to their purportedly greater need to assert status through language, given a generally subordinate social position, and to men’s greater subjection to workplace vernacular norms.

A self-evaluation test from the Norwich survey suggested that attitudinal factors may also play a part. At the end of interview, Trudgill told his informants that he would say some words in two different ways, and asked them to identify (a) the ‘correct’ pronunciation and (b) the pronunciation they themselves used most of the time. Informants had no difficulty recognizing that [tju:n] rather than [tu:n] was the standard pronunciation of tune, for example, and were generally accurate in the identification of their own usage (determined by Trudgill on the basis of the form they had used more than half the time in casual style in the recorded interview).

But an interesting pattern obtained among those who, according to the available data, answered the second question wrongly: here the over-reporters – those who thought they used the standard form more than they actually did – were mostly female, while the under-reporters – who used fewer vernacular forms than they thought they did – were mostly male, irrespective of social class. Trudgill suggested that many men genuinely believed they used these variants more than they did because, perhaps at a subconscious level since no deception seemed to be involved, they actually liked them, even though they were stigmatized low-status forms. These variants had covert prestige by virtue of their association with working-class speakers, the stereotypically ‘rough and tough’ nature of whose working lives was arguably more attractive to men than to women, who identified more strongly with overtly prestigious forms.

Key idea: The sociolinguistic gender pattern (SGP)

Women have been found consistently to use more standard or prestige forms than comparable men: this finding became known as the sociolinguistic gender pattern (SGP).

None of the explanations for the SGP is unproblematical, and indeed Trudgill’s original rider ‘allowing for other factors such as social class, ethnic group and age’ raises a number of difficulties. It is particularly difficult to control for social class when, as at the time of the Norwich study, a significant proportion of women were not in paid employment, and were often assigned the same score as their husbands on occupational criteria. The one area of general agreement is that there is no biological basis for gender-based differentiation, and indeed, there is evidence that under certain social conditions, the SGP can be reversed.

Figure 11.7: (a) index scores in three Belfast communities (after Milroy 1987: 124, Fig. 5.4)

In her seminal Belfast study, Lesley Milroy examined the use of the (a) variable, for which a wide range of local vernacular variants had been identified, in three Belfast communities: Ballymacarrett, the Hammer and the Clonard. As the graph above shows, in two of these communities the pattern was as expected for both younger and older speakers, but in the Clonard younger women had higher vernacular scores for this variable than younger men. Milroy explains this unexpected pattern in terms of changing social network structures. Significantly, the Clonard had been affected by unemployment in a different way from the other communities: here, younger women were more likely to be in stable employment than men. The workplace, which requires people to conform and show solidarity, acts as a powerful linguistic norm enforcement mechanism, to which men have traditionally been subjected to a greater degree than women.

…the young women in the Clonard contrast with the men in being fully employed, and have developed solidary relationships of the kind usually associated with men of the same age.(Milroy 1987: 144)

The role-reversal among younger Clonarders left women rather than men subject to the normative pressure of the workplace, with the result that the women’s social networks were more dense and their vernacular accordingly more focused in terms of regular use of ‘broad’ Belfast variants. The men’s vernacular, by contrast, was more diffuse, that is to say they did not use these forms with anything like the same consistency. The importance of social networks for our understanding of linguistic change, to which we return in Chapter 13, cannot be underestimated.

More recent findings have prompted a re-evaluation of the sociolinguistic gender pattern by suggesting that, rather than favouring prestige forms per se, women are more likely than men to avoid highly localized variants, and in fact often lead change in the direction of non-local, non-standard norms. On Tyneside, for example, younger middle-class women were found by Milroy et al. (1994) to be leading change away from the local glottalized (

 variant of /t/, and towards the glottal stop [], which is widely used in many urban British dialects. We are left with what has become known as the ‘gender paradox’, namely that women appear both to favour conservative prestige forms and to lead innovation in the direction of new non-standard forms.

Sociolinguists have, however, become less and less comfortable with viewing gender in a deterministic sense. Penelope Eckert has notably emphasized speaker agency in use of language to create and form identities, and seen gender less as something defining a person than as something which one ‘does’. The relative importance of gender in shaping a person’s identity may well change with age, and may vary for the same person according to situation: the use of particular variants may increase in situations where one wishes to assert a gender identity, and reduce in situations where that gender identity is less important.

Key idea: The ‘gender paradox’

More recent findings have suggested that women are likely to avoid highly localized forms, rather than actively choose prestige or standard ones. The ‘gender paradox’ emerged from evidence that women seem to favour conservative standard forms, while at the same time innovating in the direction of new non-standard ones.

Fact-check

1. [1](answers.mdx#rfn11-1)  Why do dialect maps not provide a representative picture of speech?

1. They ignore variation on the social and stylistic dimensions

1. They are often based on unreliable self-report data

1. They often use unrepresentative informants, e.g. NORMs

1. All of the above

1. [2](answers.mdx#rfn11-2)  What does an indicator variable show?

1. Significant social and stylistic variation

1. Significant variation on the style dimension only

1. Significant variation on the social dimension only

1. Significant variation according to gender

1. [3](answers.mdx#rfn11-3)  What is the Observer’s Paradox?

1. A consistent finding in sociolinguistics concerning class-based variation

1. An explanation for covert attitudes to vernacular speech

1. An occasional finding in sociolinguistics concerning gender differences in speech

1. A methodological problem in sociolinguistics regarding data collection

1. [4](answers.mdx#rfn11-4)  What is covert prestige?

1. Overuse of prestige or standard forms by lower-class speakers

1. A mismatch between the ‘objective’ and ‘subjective’ class position of informants

1. The adoption of older speech variants by upper-class speakers

1. An apparently unconscious preference for low-status forms, which appears disproportionately to affect men

1. [5](answers.mdx#rfn11-5)  When does hypercorrection occur?

1. Speakers over-extend a prestige pronunciation to environments where it is not used

1. High-status speakers use low-status forms

1. Some speakers apply prescriptive rules to correct the grammar of others

1. Speakers revert to old-fashioned or obsolescent standard forms

1. [6](answers.mdx#rfn11-6)  What was the use of centralized diphthongs in Martha’s Vineyard found to be?

1. Obsolescent

1. Restricted to older speakers

1. An assertion of local identity and solidarity

1. Associated with middle-class speakers

1. [7](answers.mdx#rfn11-7)  Why has the audio-monitoring hypothesis been criticized?

1. Speakers don’t actually pay attention to their speech

1. Women audio-monitor more than men do

1. People do not always use more formal variants when monitoring their own speech

1. It is not possible to prevent speakers from audio-monitoring

1. [8](answers.mdx#rfn11-8)  What is the sociolinguistic gender pattern?

1. A robust finding that men and women have different speech forms

1. A consistent finding that women use more standard or prestige forms than men

1. An occasional finding that women use more standard or prestige forms than men

1. A hypothesis that men and women acquire language differently

1. [9](answers.mdx#rfn11-9)  What was Labov’s ‘danger of death’ question designed to do?

1. Scare people into using more formal speech

1. Make sociolinguistic interviews more interesting for informants

1. Gain background information for the New York study

1. Divert informants’ attention from their speech

1. [10](answers.mdx#rfn11-10)  Why did younger women have higher vernacular scores than men in the Clonard?

1. Belfast English is a special case

1. The workplace had become an important vernacular norm-enforcement mechanism for younger women, but not men

1. Men were uniquely subject to pressures in the direction of RP

1. While unemployed, men paid little attention to their speech

Dig deeper

A. Bell (1984) ‘Language style as audience design’, Language in Society 13: 145–204

J. Chambers, Sociolinguistic Theory (revised edition, Blackwell, 2009), Chapters 1 & 2

J. Chambers and P. Trudgill, Dialectology (2nd edition, Cambridge University Press, 1998), esp. Chapters 2–5

J. Coates, Women, Men and Language (3rd edition, Routledge, 2013), Chapters 1, 4 & 5

W. Labov, Sociolinguistic Patterns (University of Pennsylvania Press, 1972), Chapters 1, 2 & 5

J. Milroy, L. Milroy, S. Hartley & D. Walshaw (1994) ‘Glottal stops and Tyneside glottalization: Competing patterns of variation and change in British English’, Language Variation and Change 6: 327–58

P. Trudgill, Sociolinguistics (4th edition, Penguin, 2000), esp. Chapters 1, 2, 4 & 5

R. Wardhaugh, An Introduction to Sociolinguistics (6th edition, Blackwell, 2010), esp. Chapters 2, 6 & 7