Linguistics: Introducing Sociolinguistics-MIRIAM MEYERHOFF

CHAPTER 4

Language attitudes

Key terms introduced in this chapter:
■ semantic shift
■ semantic derogation
■ linguistic relativism
■ deterministic
■ perceptual dialectology
■ social identity theory

■ salient
■ accommodation theory
■ convergence
■ divergence
■ subjective and objective measures

INTRODUCTION

Imagine you are sitting at home and the phone rings. You answer it and ﬁnd yourself talking to a stranger on the other end of the line. What are you thinking as you listen to them?
When you talk to someone, you start to form opinions about them, sometimes solely on the basis of the way they talk (Chambers 2003: 2–11). The last time you rang a service centre to buy something over the phone, or to complain about something, you would have spoken to a complete stranger. And yet, within minutes or even seconds, you probably composed quite a detailed picture of who you were talking to. Were they male, or female? Were they a native speaker of English? Did they have a strong regional dialect, or could you perhaps only say very vaguely where they come from (‘somewhere in Scotland’ or ‘probably the South’)? You might decide that you think they are Asian or a Paciﬁc Islander. You may also have strong ideas about whether they are ‘nice’, ‘friendly’ and ‘competent’, or whether they are ‘rude’, ‘disinterested’ and ‘stupid’.
We draw very powerful inferences about people from the way they talk. Our attitudes to different varieties of a language colour the way we perceive the individuals that use those varieties. Sometimes this works to people’s advantage; sometimes to their disadvantage. For instance, in the university where I work, a number of people speak with the southern British Oxbridge accents that are generally associated with privilege, respect and success. They seem to be found more often in the senior ranks of the university than people who don’t. Of course, there are exceptions – the head of the university college who still speaks a clearly northern variety of English – and the exceptions are as interesting as the rule.
In this chapter we will consider how closely linked language and attitudes are. We will start by looking at examples that show how attitudes towards other people are expressed

through language, by looking at the case of sexist language. We then examine people’s pos- itive and negative attitudes to different language varieties and we will see how these attitudes can shed light on the way people perceive to be organised.

Connections with theory

Language attitudes or language ideologies? The study of language ideologies is related to the study of language attitudes and perceptions about language discussed in this chapter. Woolard (1998) provides a discussion of the different ways in which the term ideology has been used in anthropology and anthropological linguistics; in general, its scope is necessarily broader than the study of language attitudes. The study of language ideologies considers how the beliefs and theories that speakers have about different forms of language help them to rationalise and relate highly complex social systems, such as access to power, and what social processes sustain those beliefs.

GENDER, LANGUAGE AND ATTITUDES

Language provides many windows on speakers’ attitudes to themselves and others. Our everyday speech encodes a surprising amount of information on our attitudes. In this section, we start to investigate attitudes by looking at how attitudes to women and men are reﬂected in language. We will see that synchronic and historical data may provide telling attitudinal data.

Semantic shift and semantic derogation

‘But the longer I live on this Crumpetty Tree, The plainer than ever it seems to me,
That very few people come this way,
And that life on the whole is far from gay!’ Said the Quangle Wangle Quee.
Edward Lear 1877, The Quangle Wangle’s Hat

When Edward Lear wrote The Quangle Wangle’s Hat in 1877, the word gay already had several meanings. The Quangle Wangle Quee meant that his life was lacking in joy and mirth, which in fact is the oldest meaning that the word gay has – and some people still identify it with this meaning. But even by the late nineteenth century, gay had acquired a parallel set of meanings, most of which were decidedly negative and which focused on sexual promiscuity. At this time gay was used to refer to women who were sexually promiscuous; it was only in the early twentieth century that it seems to have started to be used to refer to homosexuals
– probably the meaning we most strongly associate with the word now.
Over time, speakers may begin to use words in slightly different ways, and as these minor changes accumulate a word can end up meaning something very different from what it started out meaning. This process can be called semantic shift (or drift). For instance,

Semantic shift

Incremental changes to the meaning of
a word or phrase. Sometimes included within the scope of grammaticalisation
(or grammaticisation)
theory, but unlike classic grammaticalisation, semantic shift need not entail structural reanalysis of the word/phrase. That is, a verb might stay a verb but its meaning might be severely weakened or altered over time.
facts

the word pretty originally meant ‘cunning’ or ‘skilful’ and then went through a period when it meant ‘gallant’ or ‘brave’. The meanings of ‘pleasing’ or ‘attractive’ that we associate with pretty appear in the ﬁfteenth century, but it took a long time before these meanings edged the others out. Despite the wild trajectory pretty has had over semantic space, it has maintained an essentially positive set of meanings over time.
In this respect it contrasts with the history of gay, which has acquired negative con- notations as it has moved from meaning ‘joyful’ to meaning ‘immoral’. It is true that gay does not have a universally negative meaning now, but this process of reclaiming a positive meaning for gay only began comparatively recently. Even though in some circles it is a positive or neutral term of identiﬁcation, this is not the case for all speakers of English, and some people still consider gay to be a derogatory epithet. (Recent shifts in the colloquial uses of gay
have been decidedly negative.)

No, really?

‘The question is,’ said Alice, ‘whether you can make words mean so many different things.’
‘The question is,’ said Humpty Dumpty, ‘which is to be master – that’s all.’ Martin Gardner’s (1970) classic annotation of Lewis Carroll’s Alice books
notes that Humpty Dumpty’s view on the meaning of words has a long history. He suggests it can be seen as a form of nominalism, which the philosopher William of Ockham defended in the Middle Ages. Ockham argued that the meanings of
words derive from what we use them to signify.

Connections with theory

People sometimes end up in confused arguments about what words ‘really’ mean. Just because one meaning of a word is older than others, this doesn’t make it the ‘real’ meaning of that word, and you would ﬁnd yourself in all sorts of trouble if you tried to enforce this line. For instance, in Old English, man meant ‘human being’, irrespective of sex and age, but I doubt (m)any adult women would use this as grounds to use the ‘Men’s’ room at a movie theatre. Similarly, meat originally meant ‘(solid) food in general’, but this meaning is now wholly lost. You would be considered rather odd if you went around saying things like
‘an egg is full of meat’ (which was ﬁne when Shakespeare wrote it in Romeo and Juliet).
Most linguists ﬁnd the notion of ‘real meaning’ unhelpful. Instead, they ﬁnd it more useful to talk about what is conventionally implied by a word when it is used, what other words it frequently occurs with (i.e., its collocations), and what it implies when it is used in different conversational contexts. In some contexts, such as religion, older meanings are still relevant even if they have fallen out of use in everyday speech. However, it is important to remember that in these cases the repetition of the rituals serves to (re)construct those meanings in just those particular contexts.

A semantics text, like Kearns (2000), will tell you more about these ways of thinking about meaning, and Eckert and McConnell-Ginet (2003) includes excellent discussion showing how the study of semantics can complement sociolinguistics.

In a study in the 1970s (now ripe for updating), Muriel Schultz noticed that when she looked at words used to describe women and men, there was a distinct tendency for words describing women to have acquired negative overtones (bitch, tart, minx), while this was not true for words about men. Moreover, the words for women also linked some kind of sexual activity with the negative attitudes, again in a way that was not paralleled by the words for men. So there is a big difference between the attitudes towards women and men having multiple partners, expressed in the contrast between slut and stud. And although younger speakers of English (especially, younger women) can use the word slut to refer negatively to a promiscuous man, generally speaking there is no way of expressing the kind of disapproval about a man that slut expresses about a woman. When a word’s meaning shifts and acquires more negative connotations, it can be referred to as semantic derogation.

Semantic derogation
exercise
Semantic shift that results in a word acquiring more negative associations or meanings.

Semantic derogation

Do you think it is true that, in general, there is no way of expressing the kind of dis- approval about a man that slut expresses about a woman? If it is true, why do you think this is?
In some varieties of English rake describes a promiscuous male. Do you think slut
and rake differ only in the sex of the person they refer to?

Attitudes and context of use

exercise
Speakers of English sometimes differ in how negative they ﬁnd a word like minx or tart (especially tarty). Are these words ever entirely positive or is their meaning always somewhat ambivalent? What determines how positively you might interpret them? The person who uses them or the actions they describe? Are there other factors?

This process of semantic derogation is seen particularly clearly in male/female pairs that have, as a result of semantic shift and derogation, acquired quite different meanings. So originally courtier and courtesan both simply referred to people attached to a princely court. However, courtesan quickly acquired derogatory connotations and became a euphemism for a mistress or prostitute. You see evidence of a similar process having applied in the different synchronic meanings for master and mistress.
Table 4.1 gives an even more detailed perspective on this process. It tracks the historical trajectory of a number of English words that currently refer, or once did refer, to women. But
exercise

Pretty nice?

When we discussed pretty, I said that it has essentially kept a set of positive meanings throughout its changes. But how positive do you think the word pretty actually is? Would an artist consider it positive if someone complimented their work by saying ‘You paint the prettiest pictures’?

Table 4.1 Historical (diachronic) change affecting words that currently refer, or have at some time referred, to women.

→ Semantic shift over time →

gay (adj.) (persons) full addicted to (woman) leading homosexual stupid, hopeless of joy and social pleasures an immoral life (1935) (1980s)
mirth (1310) and dissipations (1825) (1637)
girl a child of either a female child, a sweetheart, a prostitute or a Black woman sex e.g., knave unmarried lady-love (1648) mistress (1711) (1835)
girl (1290) woman (1530)
harlot a low fellow, a male servant an unchaste knave (1330) (1386) woman, a
strumpet (1450)
hussy a mistress of a a (playfully) rude a female of the household, a term of lower orders, of thrifty woman addressing a low or improper (1530) woman (17th C) behaviour (18th C)
tart a delicious a young woman a female a young favourite baked pastry for whom some prostitute (1887) of an older man, (1430) affection is felt a catamite, a male
(1864) prostitute (1935)
queen a king’s wife, a term of an attractive a male woman of high endearment to a woman, a homosexual rank (893, 900) woman (1588) girlfriend (1900) (1924)
whore a female a woman a general term a male prostitute prostitute committing of abuse (1633) (1968)
(1100) adultery (1440)
wench a female child a wanton a servant a working class
(1290) woman (1362) (1380) girl (1575)

it also shows some of the other directions in which their meanings have developed. The definitions and dates are taken from citations in the Oxford English Dictionary. What overarching generalisation do you think you could make based on this data?
All of these words have undergone a process of semantic derogation. Some start out simply describing femaleness in neutral or positive terms (wench or hussy) and some start

out being ungendered (gay) or referring to males (harlot). In the last case, as the word began to denote women, it also acquired negative connotations, in the same manner that the neutral or positive words shift and acquire negative meanings over time.
Another thing you will notice from Table 4.1 is that the trajectory of these words tells us about more than just social attitudes towards femaleness. Attitudes to homosexual men, and specific groups of women – Black women and working-class women – are also embedded in the changes in meaning. Taken as a whole, even a small sample of words, like those shown in Table 4.2, suggests a picture of society in which the only group of people
immune to this kind of derogation are heterosexual, White, middle-class men.

Connections with theory

White, middle-class, heterosexual males are often treated as the unmarked category in society and in research on language in society (Trechter 2003). This assumption of male unmarkedness also underpins the prescriptive norm of using the masculine pronoun he when referring to a non-speciﬁc person.
Saying that White, middle-class men are immune to this kind of semantic derogation is only true in a particular place at a particular time. There are, of course, derogatory words for them too (Henderson 2003), and some of these also show evidence of semantic drift. Punk, for example, is used to refer almost exclusively to White males on Union Island in the Grenadines, and children in Vanuatu use the word turis (‘tourist’) as a term of abuse to each other. Of course, punk started out negative, and tourist isn’t
necessarily male.

Increasingly, researchers on language and gender are emphasising how important it is to understand gender in relation to sexuality (see Cameron and Kulick 2003, and also Chapter
10). The importance of this is suggested very strongly by the data provided in Table 4.1. As Cameron and Kulick point out, class and race are also important in defining how we understand sexuality and gender. From even this small amount of data it is possible to see how attitudes to women, and the general eroticisation of women, are part of a complex set of links and attitudes to other groups that are candidates as the objects of White, middle- class heterosexual male desires.
In an interesting study, that foreshadows the more recent move linking attitudes to gender and sexuality, the sociolinguist Elizabeth Gordon (1997) found that listeners were highly likely to categorise a young woman with a broad, non-standard accent as (among other things) highly likely to be ‘sleeping around’. By contrast, listeners did not categorise a young woman using a more reﬁned, middle-class accent as so likely to be promiscuous. Gordon traces this association between lower-class varieties of English and sexual promiscuity back into the Victorian era (when something like modern class distinctions started to emerge due to the urbanisation and industrialisation of society). In later chapters we will see that there is a large body of data showing that different ways of speaking correlate with the social class and sex of the speaker. Gordon suggests that the different attitudes people have to women’s use of broad or cultivated accents may play a role in determining the nature of some of these generalisations.

The linguistic derogation of women can be seen in many cultures. For example, Atiqa Hachimi (2001) shows there are aphorisms and sayings in Moroccan Arabic which cover all stages of a woman’s life and which are revealing of social attitudes to women. Examples (1)–(3) are taken from her work:

(1) l-bnat ma-ka-y-str-hum Rir trab.
‘Only death can control girls.’
(2) ʔumm-uk θumma ʔumm-uk θumma ʔumm-uk θumma ʔab-u:k.
‘Your mother, then your mother, then your mother and then your father.’ (3) l -ʕ uz-a ktər mən ʃ-ʃitan.
‘The old woman is worse than the devil.’
(Hachimi 2001: 42–44)

On the basis of a number of other linguistic examples, and an analysis of the sociocultural position of women, Hachimi argues that these kinds of aphorism encapsulate more widely held attitudes. She argues that they show that a Moroccan woman is positively valued only if she is actively producing children. Fulﬁlling the role of mother provides some insulation from the otherwise uniformly negative attitudes to women that are expressed in folk wisdom. A mother is to be treasured beyond all others, as indicated in (2), but before she starts having children (as in (1)) and after she stops (as in (3)), a woman is seen in very negative terms.

LINGUISTIC RELATIVISM

Linguistic relativism
Weaker position than determinism. Holds that the value of one factor is not wholly independent of the value of another factor, but instead is
somehow constrained by it. Associated with the Sapir–Whorf hypothesis which suggests that the way we perceive the world around us is in some way reﬂected in the way we talk. (See also Reﬂexive.)
When people differentiate between groups, they almost inevitably make qualitative judge- ments about the basis of the differentiation. Comparisons between the members of a speaker’s ingroup and members of outgroups tend to be made in such a way that they ensure a positive self-image. It stands to reason, therefore, that where one group holds more social power, the members of that group will be in a position to assert the validity of the way they perceive themselves and others, and they will try to assert the moral or aesthetic superiority of their ingroup.
This is one way of understanding what’s going on with sexist or racist language. In turn, it provides a useful basis for understanding why people find racist or sexist language objectionable. Obviously it is not the words themselves that are objectionable. As virtually every introductory linguistics class tries to stress, words are simply arbitrary signs that communities of speakers use to denote something (that is, to pick out and identify a thing or event in the world). Hence, what people ﬁnd objectionable about sexist or racist language is not the linguistic process of denotation, it is the underlying social and cultural assumptions about the way the world is and how it should be organised.
The term linguistic relativism can be used to refer to the hypothesis that the way we talk about others, and the words we use, does more than simply denote entities or events in the world. Linguistic relativism instead proposes that the way we perceive the world plays a part in how language is structured. Linguistic relativism is sometimes called the Sapir–Whorf hypothesis. Both Sapir and Whorf worked on Native American languages, and Whorf is famously associated with asserting that because the Native American language Hopi does not make the same tense and aspect distinctions that English does, Hopi speakers must perceive the world and the passage of time differently from the way English speakers do.

This argument has often been represented in extreme forms (usually by people who want to make fun of it).
For example, it has been suggested that Whorf was claiming that the grammar of Hopi imposed fundamental cognitive constraints on its speakers. That is, not only did the structure of the grammar mean that they do not perceive the passage of time as English speakers do, but they could never perceive the passage of time the way English speakers do. This would be a deterministic view of the relationship between language and thought because it contends that the shape of the language determines how its speakers perceive and experience the world.
Whorf did not actually make such deterministic claims himself. He argued a weaker, and less deterministic, position, which stressed the important links between how we talk (language), how we think about or perceive things (mind), and what it is that we perceive and have to talk about (the world). This is represented schematically in Figure 4.1.

language

mind world

Determinism/
deterministic

The idea that there is a strong causal relationship between two factors (i.e., one determines how the other will be). The idea that if you know the value for one factor, you can automatically and reliably predict
the value for another. (See also Linguistic relativism.)

Figure 4.1 Mutually reinforcing inﬂuence of language, thought and the world that we perceive and talk about.

Connections with theory

The claim that words are completely arbitrary signs may need to be qualiﬁed a little. Some researchers have found that words that are closely related semantically have closer phonological forms than would be predicted by chance. It seems possible that speakers perceive patterns in the world and subconsciously map these into linguistic patterns.

There is some evidence to support this mutually constitutive model, some of which is particularly relevant to our discussion of sexist language. A number of experimental studies have been done with adults and children showing that the name of a professional occupation is explicitly marked as male or female – people ﬁnd it hard to think of them being ﬁlled by someone of the opposite sex. So, it is much harder for people to think of a ﬁreman as being a woman (small children will often simply reject this as impossible), whereas a ﬁreﬁghter can be imagined as a woman or a man. Similarly, research asking people to ﬁnd images illustrating topics such as ‘Urban man’ versus ‘Urban life’ found that university students were much more likely to produce images that included only men in the ﬁrst case, and images that either included women and men (or consisted of cityscapes) in the latter case.
Opposition to sexist, racist, or heterosexist language (by which we mean the unques- tioning assumption that one sex, or one race, or one sexual orientation is better than another)
exercise
exercise

Generic reference terms

You can easily replicate Martyna’s (1980) or Cassell’s (1994) experiments on per- ceptions of gendered terms. For example, you could ask your subjects to go away and ﬁnd you pictures to use in advertising chapters in a book. You could give half of them gendered titles like ‘Man’s relationship with the environment’ and half of them ungen- dered titles like ‘Relating to the environment’. Or you could provide people with a set of pictures you have chosen and ask some of them to tell you which ones are suitable illustrations of sentences that include gendered job titles, like policeman, waitress, and give some of them ungendered titles like police ofﬁcer, waiting staff.
Once you start thinking about this, you will see that there are a number of interesting permutations you can do using different groups of informants or using different topics
or titles as relatively neutral controls.

very often starts from the weak Whorﬁan position that language, thought and the world are interrelated. For example, people who actively promote language change by providing guidelines for how to avoid sexist language base their arguments on the assumption that if people choose their words more carefully this will in turn affect the way they think about the relationships between women and men. Advocates of non-sexist (non-racist, etc.) language policies hope that this process will destabilise the assumptions that people make about whether or not the group distinctions they are drawing are natural or just.
Similarly, arguments against providing guidelines for language use may dispute the details of this model. Instead of assuming that there is mutual inﬂuence between all three domains, counter-arguments contend that the direction of inﬂuence is asymmetric, and that language itself does not and cannot inﬂuence the way people think or the way they perceive the world. Notice that this seems to be based on a somewhat stronger and more deterministic position than the weaker, relativistic position outlined in the previous paragraph.
However, since a strong deterministic position clearly misrepresents the relationship between language and perception, this line of reasoning concludes that changing the language will make no difference. It is necessary to change the way people think ﬁrst rather than trying to change the way people talk through language policies and publishing guidelines to avoid sexist or racist language. Under this model, once the way people think has changed,
language change will follow.

Address and reference

Many linguists have drawn attention to the social importance of how people address and refer to one another (see Sally McConnell-Ginet 2003, for instance). You can explore this in several ways.
Either: What attitudes to you do you think are expressed in the ways other people address you? Keep a record of all the ways you are addressed over the period of a week (and note who addresses you like this and where). How do you feel about the different

address terms? What would happen if someone at university tried to address you the way your family does? How would you feel?
Or: What kinds of attitudes to members of different social groups are expressed by the way they are addressed in public? Start to keep notes on all the terms you hear used to address customers in, e.g., stores, lunch stands. Is everyone addressed the same way? Do the customers and the servers use the same terms to address each other? It may help to pool your results with some friends for either of these exercises.
Whichever exercise you choose, consider how much your results support the idea that language, thought and the world are intertwined. If you think you have found any evidence for strong associations between attitudes and language, do you think it is possible to destabilise or even break those associations by changing the kinds of words
used?

RECLAIMING DEROGATORY TERMS

Later in the chapter we will look in more detail at the pioneering work of Henri Tajfel in social identity theory. One of the principal motives for Tajfel’s research was his desire to understand prejudice and racism – where does prejudice come from? How is it maintained as a social phenomenon? Tajfel observed that where there is an unequal relationship between groups, this inequality can be perceived as more or less legitimate, or as more or less permanent and stable. He proposed that when people believe they are being treated unequally, their responses to this situation will be constrained by the extent to which they think the current, unfair situation is legitimate and how readily they believe it could be changed.
So, to take a very simple example, people might think it is reasonable and legitimate for social power to be extended to people who possess physical power. If a woman believes this, then she might decide that it is legitimate for men in our society to have more social power than women because many men are bigger and physically more powerful than many women. Furthermore, if she generalises this observation and believes that it is a fundamental biological fact, she may also see the situation as permanent and unalterable. In this case, where the intergroup difference is understood as both legitimate and hard to change, Tajfel suggested that people are unlikely to contest or ﬁght against the situation.
On the other hand, a woman might look around and see that not all men are in fact physically stronger than women. If that happens, she might perceive the larger inequalities between women and men that stem from this generalisation to be unstable. In addition, seeing that the situation is unstable, she might begin to question the fundamental legitimacy of the idea that physical power is a basis for accruing social power. In this case, Tajfel suggested she would be more likely to take active steps to combat the inequalities between the groups. (Tajfel actually works through all the possible combinations of perceived legitimacy and stability and makes even more ﬁne-grained predictions about outcomes and actions than I have outlined here, but this rough summary will do for our purposes.)
Tajfel’s observations about the social outcomes associated with different perceptions of the stability and justice of intergroup differences also have linguistic signiﬁcance. If mem- bers of a social group do not perceive inequalities and biases against them to be legitimate or stable, then members of that group may seek to effect not only social change but language change as well. This occurs when linguistic practices become seen as part of a larger social

Connections with theory

The anthropologist David Aberle distinguished four types of social movement in terms that parallel Tajfel’s continuum of personal–group identities. Aberle (1966) talks about transformative movements (which aim for a total change in supra-individual systems), reformative movements (which aim for partial change in supra-individual systems), redemptive movements (which aim at a total change in individuals) and alterative movements (which aim for partial change in individuals).

matrix. Once that matrix is contested and renegotiated, all practices sustaining the system of inequalities, including linguistic practices, become candidates for renegotiation and contestation. The words used to refer to or address a group are especially likely to be subject to scrutiny and reanalysis.
This is precisely what happened with both nigger and girl as general terms of reference for Blacks and women respectively. The proscription against Whites using nigger to refer to Blacks and against using girl to refer to adult women resulted from Blacks and women questioning the legitimacy and stability of intergroup differences that had been naturalised before then. These intergroup differences and the hierarchy associated with them had been naturalised partly through the repeated use of these words with negative or disdainful connotations (this provides another example of the manner in which language constructs social relations as well as reﬂecting them).
The hierarchies were, of course, constructed in other ways too, and through other social practices, but, crucially, the linguistic practices were seen as part of that broader context. It was because of this that they became targets of contestation and eventual reanalysis. This reanalysis essentially proscribed their use as ways of referring to members of an outgroup in polite social situations. The use of both words as negative and trivialising terms of reference persists in some social contexts, of course. Perhaps of more interest is the fact that they are used with positive connotations among ingroup members. This process of reclaiming what was previously a negative term and redeﬁning it in positive ways was a strategy for dealing with perceived inequalities that Tajfel also discussed.
A particularly successful example of reclaiming a negative word and redefining it positively is the word queer. For centuries, queer had more or less negative meanings in English, and these negative associations carried over into its use as an outgroup description of gays and lesbians. In the 1990s, the word began to be reclaimed and asserted with positive connotations within the lesbian and gay community, and is a relatively neutral term for a lot of speakers of English now. This positive redeﬁnition of queer challenged the legitimacy of negative attitudes towards homosexuals, and it destabilised the privileged position of heterosexuality as an authority against which non-normative practices could be judged. The reclaiming and redeﬁnition of queer was, initially, associated with quite radical attempts to destabilise the power of heterosexual norms (as discussed in Cameron and Kulick 2003:
27–29, 77). But this bold redeﬁnition of the term has been less successful. Queer activists and queer theorists have not (yet) been able successfully to challenge the stability of the dominance of heterosexual norms in all the areas in which they might have hoped.
This discussion has covered some relatively familiar facts about the way speakers use language to express negative or derogatory attitudes to other groups in society. We have also

Reclaiming negative words

exercise
There are a number of examples where groups have reclaimed negative words and given those words a positive sense for ingroup use. Homosexuals reclaimed queer, some women use bitch as a term of strength.
Can you think of any others? (There may have been groups you knew at school who tried to redeﬁne the terms others used to refer to them.) Were these attempts to reclaim a word successful? Did they succeed in questioning the legitimacy or stability of the intergroup differences they were based on? Alternatively, why did they fail?
What other factors were involved that are not covered in the discussion here?

seen that we can learn a lot about social attitudes through historical drift as well as the synchronic uses of a word. We have also seen that the meaning of derogatory terms may be contested and actively redeﬁned by the groups they refer to, often with the express hope that changing how a word is used may change attitudes to the group of people it denotes.
In the remainder of this chapter we consider more subtle relationships between language and attitudes. We start by looking at research that shows people’s perceptions of what different dialects there are are tightly bound up with their perceptions of what different dialects are like. We then return to the phenomenon of accommodation that was introduced in Chapter 3. We will see that accommodation theory is built on the supposition that speakers express their attitudes to themselves and others in the way they speak.

PERCEPTUAL DIALECTOLOGY

In Chapter 2 we looked at variationist sociolinguistics which emerged from the traditions of regional dialectology. Social dialect studies, such as the study of New York City, had very similar goals to the goals of regional dialectology; the chief differences between the two approaches were the kinds of data they collected. Dialectologists working in both of these traditions share the objective of describing language in all its richness and diversity, in order to thereby better understand what language is and how it works as a system.
There is yet another form of dialectology – perceptual dialectology – the methods and goals of which are more closely related to the methods and goals used in surveys of attitudes to language in social psychology. In social dialectology, boundaries between varieties are identiﬁed on the basis of trained linguists’ observations of actual phonetic and grammatical features that constitute salient differences between varieties. In regional dialectology, boundaries are identiﬁed on the basis of what trained ﬁeldworkers are able to elicit from speakers or speakers’ reports of what they usually say. In perceptual dialectology, the beliefs and thoughts that non-linguists have about language are used to distinguish varieties. People’s perceptions about language, whether descriptively accurate or not, are just as important to the researcher as the objective facts about how speakers talk.
In Chapter 1, we determined that sociolinguistics was concerned with the study of speech communities, and the manner in which an individual’s linguistic performance relates to shared community norms. This dual concern means that we cannot focus exclusively on facts about production; that is, only on what people say. In addition, we would like to know

Perceptual dialectology

The study of people’s subjectively held beliefs about different dialects or linguistic varieties. The focus
on lay perceptions about language complements
the regional dialectologists’ more objective focus on the way people are
recorded as speaking.

about perception; that is, how and what people hear. We will see that perception is a more complicated process than simply decoding the sounds and words that someone else has encoded and produced. Non-linguistic factors seem to act as quite strong filters or constraints.
Work in perceptual dialectology has been pioneered by the sociolinguist Dennis Preston, and it is closely linked to what has been called ‘folk linguistics’. Folk linguistics looks more generally at non-linguists’ beliefs and perceptions about language and language use – for example, asking what constitutes ‘good’ or ‘bad’ language. Here, I will only focus on percep- tions about accents and dialect boundaries; more comprehensive resources are given in the
‘Further reading’ section at the end of the chapter.
Preston has developed a number of ways of eliciting people’s perceptions of and attitudes to different varieties of a language. One method is to ask people simply to tell you where they think people speak differently. For example, Preston provided respondents with maps of the United States and asked them to draw lines showing where speakers have different accents. In addition, he invited them to label the areas they had marked off in any way they wanted to. Some people used geographic labels similar to the kind a regional dialectologist would use, e.g., ‘Southern’ or ‘Midwestern’. An example can be seen in Figure 4.2.
Preston notes that even when you combine the responses from a large number of people (which minimises the effect of any one person’s particular idiosyncrasies, such as being very aware of the Rhode Island accent because their boyfriend grew up there), this method produces a dialect map that looks rather different from the classic regional dialectology boundaries in the United States. The geographic labels are generally less detailed and discriminate fewer dialect boundaries than professional dialectologists do in, for instance, the historically complex area of the southern Atlantic states. This is unsurprising since the average person has neither the time to devote to making ﬁne distinctions between varieties,
and nor do they have the technical resources for categorising them at the level of detail that

Figure 4.2 Dialect map of the United States drawn by a Californian respondent showing perceived areas of difference and providing some labels for varieties. (Map courtesy of Dennis Preston.)

a dialectologist does. However, some of the perceptual boundaries between dialects do fall quite close to boundaries derived from linguists’ dialect studies.
In a related study, Preston established that despite a lack of close professional study, the average listener was able to categorise speakers roughly according to where they come from. The method he used for showing this was as follows. He played nine recordings from speakers who came from towns running north–south through the Midwest and Southern US states. He played these in random order to untrained listeners, and asked the listeners to place the speakers in order (most southern to most northern variety). They were very good at differentiating southern varieties from northern varieties.
It is quite interesting that the discriminations respondents make in exercises like the map-drawing task and the accent-ordering task are often similar to the discriminations linguists make between varieties. If there was a mismatch, then this might indicate that perceptual dialectology tests simply measure something completely different to what linguists measure. But because some of the dialect boundaries recognised by linguists and non- linguists are very similar, this suggests that the two measures of dialects are mapping essentially the same thing. This suggests that if we do ﬁnd differences between perceptual and regional dialect boundaries, we might want to pay closer attention to these differences. Preston’s research suggests that lay listeners are ﬁltering what they hear through some kind of social ﬁlter that then maps these phonetic differences on to social dialect boundaries that matter more to them than they do to dialectologists. In other words, these differences may provide us with information that is directly relevant to understanding all the cognitive processes that people use to perceive and classify language.
The ﬁltering role that social information plays can also be seen in the labels for dialect regions that some respondents provide. These sometimes contain evaluative, as well as geographical, information. For instance, the label Californiese Human Growthese (in Figure
4.2) characterises a distinct regional variety, but it also suggests a somewhat dismissive or negative attitude to the variety it labels. One of Preston’s respondents from Iowa labelled Hawai’i as being characterised by Hawaii Intonation Pigeon. This label provides geographic information (Hawaii ), typological information (a kind of pidgin language), and structural information (the variety is distinctive for its intonation).
Linguists who have studied the sociolinguistic situation in Hawii would certainly agree with this respondent that there is at least one language variety spoken on Hawai’i that is unique to the islands. However, linguists would almost uniformly be thinking of the creole spoken there – a much more stable linguistic variety than a pidgin (a creole is a language that evolves from a pidgin as a consequence of contact between several mutually incom- prehensible languages; see Chapter 11). And to a linguist, intonation is only one of many features that distinguish the creole in Hawai’i from mainland US varieties of English, and it is probably not the most salient of those differences. In addition, for linguists it is arguable that a creole based on English is actually a variety of English. So there are several dimensions on which linguists’ and non-linguists’ categorisations might differ. Perceptual dialectology is not so much interested in whether the Iowa respondent’s perceptions about Hawai’i are
‘right’ or ‘wrong’. Rather, it is interested in what those perceptions tell us about which features of language people most readily pay attention to, and how they integrate those features in a socially meaningful way into their further experiences of language.
It is also interesting that non-linguists sometimes perceive dialect boundaries where linguists do not. That is, they believe they hear differences differentiating areas that regional dialectologists do not consider to be distinctive. For example, Eastern New Jersey or (Relaxed) Northwest.

We can ﬁnd examples of such discrepancies outside of the United States as well. In New Zealand, if you stopped people on the street and asked them whether they can tell where someone comes from in New Zealand just by the way they talk, the replies you would get would be different from the opinions you would get from linguists. People will often cite Southland (shown in Figure 4.3), and mention the use of ‘rolled r’s’ there. There is some empirical basis for people’s perceptions of Southland as a distinctive region. Southland had a very high proportion of Scottish settlers in the nineteenth century, and because of this many Pa¯keha¯ (White) speakers of English in Southland had a Scots ‘r’. This is very salient, or noticeable, in New Zealand because the rest of the country was non-rhotic (r-less). It is interesting to note here that even though the so-called ‘Southland burr’ is not widely used in Southland, a lot of New Zealanders still think of it as a distinctive regional feature of New Zealand English. They perceive it to be a distinct dialect area.
In addition to mentioning a variety that is now pretty much obsolete, people often say that the West Coast of the South Island has a distinct manner of speaking. As you can see
in Figure 4.3, the West Coast is relatively isolated: the Southern Alps form a barrier to the

West Coast

Southland

Figure 4.3 Map of New Zealand showing two perceived dialect areas.

east, and there are very few roads entering the region. For a long time, the West Coast made the most of this isolation, and, for instance, West Coast pubs opened on Sundays with apparent impunity while the law against this was policed throughout the rest of the country. Geographic and psychological isolation like this often foster regional linguistic differences, so it would not be surprising if we found that West Coasters in fact do speak differently from the rest of New Zealand, as many New Zealanders perceive them to. But linguists have been unable to find evidence of any clear, systematic basis for this perception of a regionally distinctive variety of English.

Connections with theory

Michael Montgomery (2000: 44–45) discusses several types of isolation that may have a linguistic impact. Isolation may be:

■ physical, or geographic (how remote is a community?)
■ sociological (what types of contact does it have with other communities?)
■ economic (how much external exchange is there of goods, ideas, etc?)
■ psychological (how open is a community to others? what attachments are there to its own culture?)
■ cultural (does a community maintain distinctive practices and beliefs?)
■ technological (are there mediated forms of external contact?)

Similarly, people may perceive social dialects that linguists do not. Many New Zealanders also believe that Ma¯ori speakers can be identified by the way they talk (Ma¯ori are the Polynesians who have lived in New Zealand for about a thousand years). Again, it has proved difﬁcult for linguists to identify reliable, objective criteria that uniquely mark this subjective perception that a Ma¯ori English exists; some possible features are discussed in Holmes (1997), but it is not clear how exclusively these mark a particular ethnic variety.
So, in both of these examples, people have opinions about dialects or varieties for which there is limited objective (linguistic) evidence. What then are people responding to? Some sociolinguists would argue that these subjective perceptions are taking deeply held beliefs about social boundaries and projecting these beliefs into the linguistic system. The argument goes: because people perceive the boundary between Ma¯ori and Pa¯keha¯ ethnic groups to be salient, then there must be a linguistic boundary between those groups; they must speak differently. Because they perceive the West Coast to stand apart from the rest of New Zealand, they believe that people there must speak differently from the rest of the country.
If this is, indeed, the way such perceptions of dialect differences emerge in the absence of objective support, then that is a very strong indicator of the crucial role language plays in reflecting and constituting different social identities. This is one reason why studies of perceptual dialectology can be important data for sociolinguists. They provide an independent measure – perception data, rather than production data – of how central language is to the formation and maintenance of social and personal identities. That is, how people perceive language provides evidence that is just as useful and relevant to the complicated balancing

act between ﬁtting in and being distinctive (introduced in Chapter 2) which may motivate differences in the way speakers use language.

Connections with theory

There is a lot of evidence showing a bleed-through between attitudes to a language and speakers of a language. This research was pioneered by the Canadian social psychologist Wallace Lambert who showed that the same speaker would be ranked very low on some social traits (e.g., power, wealth, trustworthiness) when speaking one language and high on the same traits when speaking another.
Recently, John Baugh and his colleagues (Purnell et al. 1999) have conducted a number of similar experiments and found clear evidence that landlords, for instance, respond negatively to the same speaker when they use features of Hispanic or Black English. Anita Henderson (2001) showed that personnel managers are acutely sensitive to and react very negatively to the presence of either features of AAVE pronunciation or
AAVE syntax.

In subsequent chapters we will see that the social categories that are salient in a particular community may be reproduced in the production of linguistic variation. However, for now, the perceptual dialectology research provides a valuable reminder of the ways that social factors can be part of the process of perception as well as of the process of production.

Social identity theory
A social psychological theory holding that people identify with multiple identities, some of which are more personal and idiosyncratic and some of which are group identiﬁcations. Experimental work
in this framework suggests that people readily see contrasts between groups in terms of competition, and seek to ﬁnd means of favouring the co-members of
the group they identify with over others.
ATTITUDES TO LANGUAGE: IDENTITIES AND ACCOMMODATION

In the last chapter we looked at ways in which speakers’ attitudes to the individual they are talking to can affect the way they talk. Audience design proposed that speakers derive their style shifts to an addressee from the characteristics that they associate with the speech of the group as a whole. This presupposes that speakers perceive their interlocutors to be individual representatives of a group. This presupposition can be traced back through the social psychological theories that underlie audience design; namely, social identity theory (SIT) and communication accommodation theory (CAT), which was also introduced in Chapter 3. The next sections provide a brief outline of some of the features of both SIT and CAT that are salient for sociolinguists. We examine some of the linguistic effects of accommodation, focusing particularly on mismatches between what speakers perceive themselves to be doing, and what objective measures show that they actually are doing.

SOCIAL IDENTITY THEOR Y

Social identity theory (SIT) is a theory of intergroup relations in which language is one of many potent symbols that individuals can strategically use when testing or maintaining boundaries between groups. The theory was proposed by the social psychologist Henri Tajfel

who had a deep interest in understanding the social and psychological processes under- pinning conﬂict between members of different ethnic and religious groups.

Tajfel’s interest in identity and prejudice grew out of his own personal expe- riences. He was born in Poland in 1919, and moved to France in 1937 for university. During the Second World War he served in the French army and after being captured spent ﬁve years in German prisoner-of-war camps. He was able to survive the war by assuming a French identity, masking his Polishness and his Jewishness. He would later point out that regardless of the interpersonal relationships he had developed within the German camps, if his Slavic and Jewish group identities had been discovered they would have completely and unques- tionably determined his fate (Turner 1996).

To this end, Tajfel (1978) distinguishes between identities which are principally personal and identities which are principally associated with a group. SIT recognises that we all identify with many personas at different times and places and in different contexts; nevertheless, it assumes that we simplify away from a lot of this complexity in any given interaction and perceive it as being a more or less intergroup or interpersonal exchange. That is, we generally perceive a particular personal or group identity to be most salient at a particular stage in an interaction (where ‘salient’ here means the identity activated and oriented to by the immediate context of the interaction).
When a personal (rather than group) identity is salient, our behaviour is more likely to be constrained by idiosyncratic aspects of our personality, our mood, or the immediate context. This would predict that when personal identities are the basis for an interaction our behaviour
– including the way we talk – will be subject to more variability. On the other hand, the theory predicts that if group identities are more salient, the way we behave and talk will tend to accentuate uniformity within groups. In other words, we would gravitate towards what we consider the normal or typical way of talking for a member of that group (and abstract away from the internal differences we know all groups have). In addition, SIT holds that when a contrast is made between the groups, our behaviour will accentuate the differences between the groups as well.
Recall that Bell proposed that intraspeaker stylistic differences derive from, and are therefore less than, differences between groups (Chapter 3). Bell’s proposal follows from these last two points about intergroup communication. The accentuation of uniformity or similarities within a group would have allowed Foxy Boston to identify an appropriate level of, e.g., third person -s presence or absence, given the group identities she perceives to be salient. The accentuation of differences between groups might ensure that whatever the range of intraspeaker variation there is, it will not outstrip differences between the groups that are presently salient. (However, we must remember Baugh’s suggestion that these prin- ciples of interaction and identity management may be subordinate to linguistic constraints.) The salience of interactants’ identities is not determined and ﬁxed from the outset of
an interaction, though it is certainly possible for speakers to go into an interaction with very ﬁxed ideas of who they are talking to. Salience can also be negotiated during an interaction and emerge collaboratively, a point that is important in linking SIT with accommodation and, ultimately, sociolinguists’ interests in style-shifting (as represented by audience design). Both

Salient/salience

A maddeningly under-deﬁned
term when used in sociolinguistics. Sometimes refers to how readily a particular variant
is perceived/heard (this may be due to physiological factors affecting perception, or social and psychological factors that affect prime speakers and make them attend to
a form). Sometimes refers to a
non-linguistic factor that the context or participants appear
to have foregrounded in discourse.

Accommodation

The process by which speakers attune or adapt their linguistic behaviour in light of their interlocutors’ behaviour and their attitudes towards their interlocutors (may
be a conscious or unconcious process). Encompasses both convergence with or divergence from interlocutors’ norms. (See also Social identity theory.)
personal and group identities can be made more salient by others or selected on our own initiative.
Tajfel suggested that personal and group interactions fell at the opposite ends of a scale or continuum, though he was quick to note that this was something of an explanatory idealisation of the way interactions really take shape. A purely personal identity might not actually exist outside of the framework since even when we seem to be acting just as indi- viduals, our behaviour may be interpreted as more or less consistent with the group identities we also possess. For instance, suppose you ﬁnd yourself comforting a child who is crying hysterically after falling off some playground equipment. You would certainly be focused on meeting the immediate needs of the individual child before you, but your response as an individual may well be coloured by group identities, e.g., ‘I hope a parent arrives soon, I’m hopeless with kids.’
SIT also supposes that we have different feelings about and attitudes to the social groups we differentiate. As a general rule, group identities are presumed to stabilise in contrast with other groups’ identities, and this element of comparison or contrast translates into some groups being seen more positively than others. It is also a general rule that we try to ﬁnd some basis for seeing groups we identify with in a better light than the ones we are contrasting them with. So the differentiation between groups has a useful social function. In order to feel good about Us, we need a Them to compare ourselves to.
In the previous chapter we examined two frameworks for analysing speaker style: attention to speech and audience design. We noted that one important dimension on which the approaches can be distinguished is the extent to which the speaker is portrayed as an active participant in the construction and negotiation of a speech event. However, it is also worth noting that both approaches share Tajfel’s insight that individual and group identities are linked.
This was especially clear in Bell’s audience design framework. As we saw in the previous chapter, he attributed some style-shifting to the effects of more personal relationships (i.e., design for an addressee) and some style-shifting to the effects of groups (i.e., design for what Bell called reference groups). In addition, as we have already noted, the mechanisms of audience design are presumed to operate with individuals standing in for a group.
It will be less clear at present how Labov’s attention to speech framework also relates the group to the individual. However, this connection should become clearer in Chapter 8 when we consider parallels between the frequency of speciﬁc variants in different styles and the frequency of those variants in speakers from different social classes.

ACCOMMODATION THEORY

Accommodation theory has much in common with the tradition of social identity theory: accommodation theory is a bundle of principles that are intended to characterise the strate- gies speakers use to establish, contest or maintain relationships through talk. The original statement of the theory by Howard Giles (1973) focused on speech behaviours alone, but developments following in Giles’s footsteps have expanded the scope of the research so as to include strategies in non-verbal communication behaviours as well. The ﬁeld is, therefore, sometimes referred to as speech accommodation theory and sometimes as communication accommodation theory.
Regardless of its scope, accommodation theory rests on one pivotal process: attunement. The idea is that we all tailor, or attune, our behaviours according to the interaction, and this

process of attunement involves a range of communicative behaviours, like speech styles. Attunement renders the addressee(s) as equally important as the speaker and it also presents communicative behaviours as elements in a dynamic system. Drawing on the personal/group distinction of SIT, accommodation theory allows for attunement to attend primarily to very personal or very immediate factors, or else to occur in the context of intergroup contrast. Where an interaction is perceived in terms of group identities and group contrasts, accom- modation theory also proposes that affective factors enter into the dynamic.
This, too, builds on the principles of SIT. An interaction that is perceived to be taking place between ingroup members (or between people who would like to negotiate a common group identity) will foster strategies that accentuate internal commonalities. This strategy, it is assumed, contributes to the social function of generating positive feelings about ourselves and the co-members of that group. This is often accompanied by a downgrading of the out- groups we might be contrasting ourselves with.
The two main strategies used in the process of attunement are convergence and divergence. Convergence involves a speaker altering the way they talk so that it approaches the norms of their interlocutor and accentuates commonality between the interlocutors (as discussed above). As we will see shortly, convergence can entail approaching the actual norms of the addressee, or it may involve approximating norms that the speaker believes (incorrectly) are characteristic of their addressee. On the other hand, divergence involves accentuating differences between the speaker and their addressee(s). Speakers may con- sciously undertake either strategy, but it is important to note that accommodation may occur well below the speaker’s level of conscious awareness (this is sometimes misunderstood by linguists, who think that attunement and accommodation are consciously controlled moves in a conversation). In particular, it is important to note that the speaker may not be able to describe or identify the precise linguistic features that are altered through the attune- ment processes of accommodation. The next two sections provide examples of linguistic convergence and divergence.

Conver gence

When the attunement involves increasing similarities between the speaker and their addressee, Giles called this convergence. This may happen at the level of very marked linguistic differences, such as the choice of language, or it may occur more subtly at the level of features such as pitch and speech rate. Speakers are generally reasonably aware of what motivates them to alternate between languages depending on the context and their addressee (and we return to switching between languages in Chapter 6). However, they may be quite unaware of changes that take place in their prosody, and their realisation of phonological or morphosyntactic variables.
Convergence with the addressee in choice of language is something that is learnt quite early, and there are obvious functional reasons for this. There’s not much point talking to your Mandarin-speaking grandfather in English if he isn’t going to understand a word you say, and vice versa with your Canadian cousins. However, children also seem to learn that alternating their dialect or accent may make for more effective communication, depending on their addressee. A little boy growing up in Scotland, with non-Scottish parents, was heard to do just this as early as 19 months. Sam was dropped off by a parent at kindergarten one morning and decided to go and look at the books. He walked across the room saying ‘Book, book, book’. The vowel he used in ‘book’ when his parent ﬁrst put him down was relatively centralised

Convergence

Accommodation
towards the speech
of one’s interlocutors. Accentuates similarities between interlocutors’ speech styles, and/or makes the speaker sound more like their interlocutor. It is assumed to be triggered by conscious or unconcious desires to emphasise similarity with interlocutors we like, and to increase attraction. (See also Divergence; Social identity theory.)

Divergence

Accommodation away from the speech of one’s interlocutors. Accentuates differences between interlocutors’ speech styles, and/or makes the speaker sound less like their interlocutor. It is
assumed convergence is triggered by conscious or unconcious desires to emphasise difference and increase social distance. (See also Convergence; Social identity theory.)

[bυək] – similar to what he would hear at home – but by the time he had crossed the ﬂoor of the nursery to the reading corner, he was using a backed and rounded vowel more like the one used by his Scottish caregivers, [buk]. Sam’s kindergarten teachers would certainly understand [bυək], just as his parents would understand [buk], so in this case his convergence on the Scottish norms in his daycare and his parents’ norms at home is unlikely to be motivated by comprehension problems. Accommodation theory would suggest that his behaviour shows he associates other social and interactional beneﬁts with speaking more like the different groups of people he moves in and out of.
Studies have also shown that people are quite quick to attune their speech rate to their addressee’s. Generally, if we are talking to someone who talks more slowly than we do, we converge by slowing down our own rate of speech. Our interlocutor may also converge by speeding up slightly. This kind of mutual accommodation – some give and take by both parties – is an integral part of the theory.

Diver gence

Attunement doesn’t always entail convergence. Depending on the circumstances, speakers may decide that their interests are best served by maintaining, or even accentuating, distinctions between themselves and their interlocutors. This strategy is called divergence. Just as convergence in choice of language can facilitate comprehension, divergence in language choice can serve as a shield. For instance, in a report that tourists were being ripped off on visits to Prague, the journalist mentioned waiting staff who ‘suddenly lose their ability to speak previously excellent English when questioned by foreigners about what they paid for’ (Krosnar 2005).
Divergence at the level of accent can be equally functional. An American who has lived outside of the United States for many years says that she plays up her American accent, diverging from the locals, when she wants sympathy, or sometimes when she wants better service. So, for instance, if a police ofﬁcer challenges her for stopping in a ‘No Parking’ zone, she replies in a broad accent suggesting she is perhaps a tourist and hopes it will make the police ofﬁcer decide giving her a ticket isn’t worth it. Similarly, she trades off the stereotype of Americans being vociferous complainers if service isn’t good by accentuating her accent when she feels that the service she is getting isn’t efﬁcient or prompt.
And there are less Machiavellian functions to divergence. People may diverge lin- guistically from their interlocutors in order to accentuate differences if the comparison will foster positive feelings about their ingroup. Jokes are often made about how touchy Canadians and New Zealanders are if they are mistaken for Americans or Australians (respectively). A strong reaction accentuating their pride in being a Canadian or a New Zealander can be strengthened by the use of marked or unique features of their accent.
In the previous chapter we considered some examples of divergence, and these showed that the reasons why individuals might diverge are often related to their perceptions of and attitudes towards a group, as well as to individual members of that group. Our discussion of divergence illustrates the point made by social identity theory, namely that personal and group identities fall on a scale and are inherently blurred. We will return to this point in Chapter
6 when we look more speciﬁcally at accommodation in language choice.

Asymmetric convergence and divergence

As we have said, accommodation theory is a theory about interaction, and as such it is concerned with the negotiation of perceptions and identities between interlocutors in conversations. The examples given in the introduction to convergence and divergence are fairly straightforward ones, and they avoid dealing with disputes or contestation.
However, the theory allows for the possibility of an interaction in which one person converges and the other person diverges. These examples can be particularly enlightening, as they show how complicated and important people’s attitudes towards others are and how these attitudes can be played out in language use. One such example is found in the debate about how to write Hawaiian.
Hawaiian is spoken in the US state of Hawai’i, where the dominant language is English. In Hawaiian, vowel length is phonemic; this means that a difference in vowel length alone can change the meaning of a word. So, for example, kau means (among other things) ‘to place something’ and ka¯u means ‘your(s)’. The only difference is that the word meaning ‘your(s)’ has a long /a/ vowel and the verb meaning ‘place’ has a short one. Hawaiian also has a phonemic glottal stop /ʔ/ (the sound in the beginning and middle of the word marking surprise, ‘uh-oh’), so the words ulu (‘to grow’) and ´ulu (‘breadfruit’) are only distinguished in meaning by the presence of the glottal stop in ´ulu.
Now, because these elements aren’t phonemic in English, neither the glottal stop nor vowel length has any obvious way of being written in a spelling system that is based on the English alphabet. There are two options. You can omit them, or you can use orthographic conventions that are not used in English: you can write a line over a long vowel and you can write the glottal stop with an apostrophe or a single open quote (as in the word ‘Hawai’i’). These are called the ´okina and kahako¯, respectively.
Generally, it is preferable to use these symbols – if you leave them out it would be a bit like skipping the ﬁnal ‘e’ in English words like bake or garbage. That is to say, you could still read it, but it’s just not standard spelling. So a lot of people in Hawai’i, even if they are not speakers of Hawaiian, try to learn where the kahako¯ and the ´okina belong. This is seen as a gesture of respect for the language and its speakers. In other words, their attunement takes the shape of convergence, similar to the case of the little boy who uses Scots vowels when speaking to his Scottish daycare workers.
But the situation is complicated by the larger relationship between Hawaiian and English. Some people who speak Hawaiian are concerned about the inﬂuence that English is having on the language, and they would prefer to foster features that might create obstacles that would prevent further English-influenced incursions on it. So some Hawaiian language activists have argued that leaving the long vowel and glottal symbols out of Hawaiian is a good idea, because it makes the language more opaque to English speakers, and helps to maintain it as an ingroup code. In other words, by arguing in favour of making the spelling less transparent to people used to English norms, some speakers of Hawaiian advocate divergence. Interestingly, the ´okina and kahako¯ are the linguistic focus for both attempted convergence and divergence.
Similar cases of asymmetric convergence and divergence can take place between individuals. The sociologist Ben Rampton provides some interesting examples (1998). In one recording, ﬁve teenage girls are talking and listening to music together. One of the Anglo girls starts talking about and expressing a passion for bhangra (a Punjabi music style). Her Indian friends give her very minimal feedback and encouragement to keep talking about it –

Rampton found in interviews that many of the Punjabi teenagers were quite unenthusiastic about their Anglo peers adopting ‘their’ music.
These examples show that convergence and divergence need not be symmetric. They
can be asymmetric, with one group or person converging and the other group or person diverging.

Connections with theory

There have been a lot of experimental studies that show strong relationships between positive attitudes to an interlocutor and convergence in choice of language and some aspects of speech styles. People who are well disposed to each other have been found to converge on how often they interrupt each other, how long a pause they leave between turn, length of turns they take, and non-linguistic aspects of communication, like laughter. But we still lack a lot of work on more detailed aspects of the linguistic system, such as variables like (r).
Also, it is not clear how accommodative attunement relates to or complements priming. Priming is when a speaker follows the form or content of a preceding speaker’s turn and it has been studied by a number of psychologists. They ﬁnd that if I say, ‘Why did you lend her your car?’, you are more likely to use the same sentence structure in your reply. That is, you will probably say something like, ‘But I didn’t lend her my car’, repeating my order of the verb, goal and object, rather than the equally used grammatical alternative,
‘But I didn’t lend my car to her’. It’s not clear whether priming depends on speakers’
attitudes to each other or to the task at hand. As far as I know, such questions have not been explored in the experiments on priming.

Subjective and objective measures
A speaker’s perceptions of their own performance and their performance evaluated by some external measure.
Subjective and objective measures of convergence

The business of measuring convergence and divergence is complicated even further by the fact that interactants may believe they are converging or diverging, but they fail to achieve their goal. This may be because they misanalyse or misjudge their goal, or it may be because they do not have the necessary resources or skills to reach their goal accurately.
In Chapter 3 we mentioned the case of Peter Trudgill’s convergence with the speakers he was interviewing in Norwich. In this case, he was not aware that he was converging even when he did. That is, there was a mismatch between his subjective perception of what was happening and the objective reality. He thought he was using the same interviewer style in every case, so from his subjective perspective there was no convergence to the norms of his interlocutors. But an objective measurement of what was going on after the fact showed that in fact he had converged.
If we distinguish between subjective and objective levels of convergence and divergence, there are four logical possibilities. These are shown in Table 4.2.
In cells A and D, there is a match between what the speaker subjectively believes is going on and what any objective observer would discover if they examined the interaction. In A, speakers believe they are converging and they succeed in doing so. In the other case, they

Table 4.2 Four possible kinds of interaction according to whether a speaker converges or diverges on subjective or objective measures. (Source, Thakerar et al. 1982: 238.)

subjective

Divergence

Objective
convergence divergence
A
C
B
D

believe they are diverging and again they succeed in this. When the subjective and objective measures of attunement coincide it is fairly easy for the researcher to invoke speakers’ attitudes as an explanation for the behaviour observed, as I did in the examples in the previous sections. However, in the other two cells, B and C, there is a discrepancy between the strategy speakers believe they are employing and the actual details of their performance. We will look more closely at such situations in the next two sections.

Subjective convergence and objective divergence

Cell C represents the case where a speaker may be trying to converge with their interlocutor, but in the process of trying to converge they actually end up diverging. This seems to happen if the speaker:

■ incorrectly judges the situation, and
■ converges to the way they perceive their interlocutor to be talking (rather than to the way their interlocutor really does talk).

A study in Thailand found Thai children doing this when they were talking to ethnically Chinese speakers of Thai (Beebe 1981). There is a stereotypical Chinese accent associated with Chinese speakers of Thai, but the Chinese subjects in Beebe’s study did not use this, they spoke standard Thai. Nevertheless, when the children in Beebe’s experiment were talking to an ethnically Chinese experimenter, they began to use features stereotypically associated with Chinese pronunciations of Thai in their own speech, even though these features were absent in the speech of their interlocutor. In other words, the children seemed to be con- verging to what they (erroneously) perceived their interlocutor to be doing, and they were effectively unable to ‘hear’ what their interlocutor really was doing.
In this case, it appears that the children were converging not to their interlocutor’s individual norms but rather to the norms widely associated with the group that they perceived their interlocutor to belong to (Chinese speakers of Thai). Notice that in this case, the motives for such behaviour are more complicated than they are in the more straightforward cases of cells A and D. Because the children’s objectively divergent behaviour seems to be based on subjective notions of convergence, we have to analyse their unintentional divergence as an attempt to seem agreeable – that is, as if it were objective convergence.

Subjective divergence and objective convergence

An even more interesting example of a discrepancy between the strategy that the speaker perceives they are using and the strategy that an objective measure shows them to be using is found in cell B where there is subjective divergence but objective convergence. I am not aware of any studies that illustrate this process taking place in interactions between individuals, but work by Nancy Niedzielski on the perceptions and attitudes of groups of speakers seems to suggest that it is possible.
Niedzielski (1997) devised a simple but effective experiment. She had already deter- mined that speakers of US English in Detroit (on the Canadian border) generally now pronounce words like MOUTH with a raised onset. However, Detroiters still perceive this raising of (aw) to be characteristic of Canadian English and not their own.
Previous perceptual dialectology studies in the Detroit area had shown that Detroiters are convinced that they speak ‘Standard’ American English, and it appears that they are completely unaware that a growing number of them use these raised, Canadian-like variants of (aw). Niedzielski drew on principles of SIT and accommodation theory and predicted that given Detroiters’ social perception of themselves as speakers of ‘Standard’ American English they would be likely to perceive a recording of Detroit speech as sounding like General American, even if the speaker uses raised Canadian-like variants.
Niedzielski made up a tape with some sentences in which Detroiters used raised variants of (aw), e.g., south, house and out. These became the target words for the experiment. She then synthesised different versions of the target words and asked a number of listeners from Detroit to tell her which one of the options was closest to the variant they heard in the original sentence. One of the synthesised versions was identical to what the speaker actually said, and one was close to the historical norms for the Detroit accent. One was even more open, and in fact more typical of what is considered General American.
Niedzielski led half the people listening to the tapes to believe that the speaker was Canadian, and half of them to believe that the speaker was from Detroit, but in all other respects each respondent heard exactly the same thing. So half of the subjects thought they were trying to match synthesised versions of a local Michigander’s vowels (i.e., a member of their ingroup) and half thought they were listening to a Canadian’s vowels (i.e., that they were listening to someone from an outgroup).
Even though the objective facts of the case were that everyone heard a Detroiter using raised variants of the (aw) diphthong, Niedzielski found that respondents were much more likely to report hearing the diphthongs as raised if they thought the speaker was Canadian than they were if they thought the speaker was from Detroit. This is shown in Figure 4.4, which combines the results for several sentences.
When her subjects thought the speaker was from Detroit, they said that what they had heard was closer to the synthesised variants with lower onsets – either the ones people in Detroit traditionally used, or even an ultra-low variant that had never been used in Detroit. It seems that even when Detroiters heard a local who was objectively converging with their neighbouring Canadians’ pronunciation of these words, their subjective perception was that the speaker uses the traditional open Detroit variant.
Since this study does not combine production and perception data from the same people, it does not technically fill out cell B in Table 4.2. However, Niedzielski’s results suggest that it will be possible in fact, not just in theory, to ﬁnd instances of subjective divergence with objective convergence.

60
55

50 47

40 36

Percent
30 26

19
20

‘Canadian’ voice
‘Michigan’ voice

17

10

0
Ultra-low onset Standard /a/ onset Actual token

Synthesised diphthong selected

Figure 4.4 Inﬂuence of perceived nationality of speaker on choice of token in matching task. Pooled results for south, about and out. (Source, Niedzielski 1997: 71–72.)

The possibility of mismatches between speakers’ perceptions of what they are doing and their objective performance raises some additional problems for the sociolinguist work- ing from spontaneous speech. In recordings made from spontaneous conversation, the researcher doesn’t have access to the privileged information about a speaker’s attitudes to their interlocutor and the context is not controlled for such factors as it is in a social psychology experiment. This deﬁciency cannot be remedied by asking people afterwards either, because, as we have just seen, people may believe they are doing one thing and actually do something quite different.
However, we don’t want to ignore these beliefs and perceptions even when they do not seem to be based on objective reality. They may tell us things about the structure of a speech community that even a trained linguist cannot detect. For example, Naomi Nagy has been part of a large team doing ﬁeldwork in Montreal, Canada for some years now. This city is part of the French-speaking state Quebec, with an English-speaking minority. The Anglophone residents of Montreal all have to be (more or less) bilingual in French since that is the dominant language in the city. Many of the Anglophone Montrealers in the study claim that they have a distinctive way of speaking French which marks them as a group. However, to date, Nagy hasn’t been able to find any linguistic features that distinctively mark the French of the Anglophone bilinguals as a group. This may also be a case of subjective divergence with objective convergence. But it is also possible that the speakers are attuned to differences that are very subtle and infrequent indeed, and which have so far escaped the attention even of researchers who know the community very well.

CHAPTER SUMMARY

This chapter has reviewed several rather different approaches to the study of language attitudes. It has tried to make the case for including attitudes about language and about

different users of language as an important part of sociolinguistics, including the study of variation which generally relies almost exclusively on data from production and avoids issues related to perception.
It began by taking a look at how attitudes to others are revealed through language, speciﬁcally focusing on the paths through which word meanings change over time. This linked with subsequent sections of the chapter in which people’s perceptions about different language varieties were brought back into focus, and strategies used to contest and renegotiate the meaning of words were related to principles of group and personal identity theory.
We saw that non-linguists’ perceptions of dialect boundaries largely match the boundaries linguists draw, and suggested that this indicates that the variables being analysed by linguists and attended to (often quite subconsciously) by non-linguists are fundamentally similar. Where there are differences between the two, I have suggested that this may be an important indicator of social factors relevant to that speaker or their speech community.
The chapter has also looked in some detail at accommodation theory, which was introduced in the discussion of speakers’ style-shifting in Chapter 3. The roots of audience design were traced back through accommodation theory to social identity theory in this chapter, and a wider range of strategies – symmetric and asymmetric convergence and divergence – were discussed. Accommodation theory stresses the importance of speakers’ attitudes to their addressee, and the resulting dynamism in interactions. It also provides us with a context for comparing what speakers think they are doing with what they actually are doing.
In the next chapter, we continue to look at the way speakers balance between inter- personal and intergroup needs in conversation. The lens there is on the social signiﬁcance of alternations between different languages, both at the national or institutional level and at the personal or individual level.

FUR THER READING

Pauwels (1998) – on non-sexist language.
Hellinger and Bussmann (2001–2003) – three-volume collection of articles on language and gender, many of which deal with attitudes to women expressed in a variety of languages.
Lucy (1992) – a more recent view on linguistic relativity.
There is an extensive literature on the analysis of racist language, especially among researchers coming out of the social psychology tradition, e.g., Wetherell and Potter (1992), Billig (1995), and see also Wodak and Reisigl (2001).
Niedzielski and Preston (2000), Preston and Long (1999) – folk perceptions of language and perceptual dialectology.
Gallois et al. (1995) – summary of theory and principles in communication accommodation and attunement.
Trudgill (1983) – case studies of accommodation in British pop and punk rock.
Warner (1999) – on the politics of Hawaiian language revitalisation; Schütz (1994) on the development of orthographic standards in Hawaiian.
Niedzielski (1999) – more accessible source of data on mismatches between perception and reality.

Contents

Introducing Sociolinguistics-MIRIAM MEYERHOFF - ,CHAPTER 4 Language attitudes

No comments:

Post a Comment