LANGUAGE ATTITUDES, NORM AND GENDER A … · LANGUAGE ATTITUDES, NORM AND GENDER A PRESENTATION OF...

27
LANGUAGE ATTITUDES, NORM AND GENDER A PRESENTATION OF THE METHOD AND RESULTS FROM A LANGUAGE ATTITUDE STUDY By Marie Maegaard Department of Scandinavian Research University of Copenhagen Abstract This paper deals with the more traditional sociolinguistic area of language attitudes. Based on data from the city of Århus, Denmark, I discuss the methods used in speaker evaluation experiments. The paper presents both a methodological discussion and the results of a study using the described techniques. The method is a mainly qualitative version of ‘the verbal guise technique’, and involves young people in Århus. Twelve speech samples (six female and six male voices) were used for this experiment, representing three different accents. The aim is to investigate the possibility of the Århus accent being a linguistic norm ideal among the young people in Århus. The results show that this does not seem to be the case, and that speakers are evaluated differently according to both accent and gender. 1. Why study language attitudes in Århus? Denmark is a relatively homogenous society, linguistically speaking. The Copenhagen dialect(s) has spread throughout the country, and today the local dialects are levelled (Pedersen 2003). Thus, the different ways of speaking that I refer to in this paper are probably best described as different accents of Danish since they only differ phonetically (Wells 1982). The Copenhagen accent is widely perceived as standard Danish (Kristiansen 2001). However, it is possible that in some areas of Denmark other linguistic norms exist. If they do, we may expect the Copenhagen accent to be challenged and the local accent to persist. The obvious places to look for such an alternative norm are in the larger cities in Denmark. Århus is with its 300,000 citizens the second largest city in Denmark and it is at the same time “the metropolis of Jutland” as the mayor explains on the city’s website (www.aarhus.dk) . Århus (and Jutland) is often thought to be in opposition to the national capital of Copenhagen, and this might apply to language as well as to other matters. So, one might say that if the Copenhagen accent proves to be the norm among adolescents in Århus then it is unlikely that

Transcript of LANGUAGE ATTITUDES, NORM AND GENDER A … · LANGUAGE ATTITUDES, NORM AND GENDER A PRESENTATION OF...

LANGUAGE ATTITUDES, NORM AND GENDER A PRESENTATION OF THE METHOD AND RESULTS FROM A

LANGUAGE ATTITUDE STUDY

By

Marie Maegaard Department of Scandinavian Research

University of Copenhagen

Abstract This paper deals with the more traditional sociolinguistic area of language attitudes. Based on data from the city of Århus, Denmark, I discuss the methods used in speaker evaluation experiments. The paper presents both a methodological discussion and the results of a study using the described techniques. The method is a mainly qualitative version of ‘the verbal guise technique’, and involves young people in Århus. Twelve speech samples (six female and six male voices) were used for this experiment, representing three different accents. The aim is to investigate the possibility of the Århus accent being a linguistic norm ideal among the young people in Århus. The results show that this does not seem to be the case, and that speakers are evaluated differently according to both accent and gender.

1. Why study language attitudes in Århus? Denmark is a relatively homogenous society, linguistically speaking. The Copenhagen dialect(s) has spread throughout the country, and today the local dialects are levelled (Pedersen 2003). Thus, the different ways of speaking that I refer to in this paper are probably best described as different accents of Danish since they only differ phonetically (Wells 1982). The Copenhagen accent is widely perceived as standard Danish (Kristiansen 2001). However, it is possible that in some areas of Denmark other linguistic norms exist. If they do, we may expect the Copenhagen accent to be challenged and the local accent to persist. The obvious places to look for such an alternative norm are in the larger cities in Denmark. Århus is with its 300,000 citizens the second largest city in Denmark and it is at the same time “the metropolis of Jutland” as the mayor explains on the city’s website (www.aarhus.dk). Århus (and Jutland) is often thought to be in opposition to the national capital of Copenhagen, and this might apply to language as well as to other matters. So, one might say that if the Copenhagen accent proves to be the norm among adolescents in Århus then it is unlikely that

other norms than the Copenhagen one are accepted among young Danes anywhere else in the country. If, on the other hand, it turns out that the Århus people orient themselves towards the Århus accent, then it is possible that there are other places in Denmark where the dominance of Copenhagen is absent, and the local accents are regarded as the norms. This in turn means that the local accents might resist the pressure from Copenhagen and the levelling of dialects will come to a stop. On the other hand, the situation might turn out to be far more complicated than this. It may be the case that several parallel norms exist and that they carry positive values in different domains or dimensions. Thus, Kristiansen (2001) argues that in the Næstved area young people orient themselves towards two standards, one that has to do with dynamism, and one that is related to school, career etc. However, none of these standards involve the local or regional accent – both of them are Copenhagen accents. I will return to a discussion of the problems of interpreting one accent as more positively evaluated than another later in this paper. Thus, the aim of the study is to determine whether or not Århus functions as a regional norm centre. The term norm should in this context be replaced by the term norm ideal (Kristiansen 1992) to avoid confusion. As the above indicates, I am not concerned with the notion of norm as the normal language use, but of norm as a psychological construct. One could say that in Århus the language norm is the Århus accent since this is what most people in Århus speak, but their norm ideals might be different. The norm ideals consist of two parts: the abstract language system and the attitudes associated with the linguistic variation. In the case of language change, the relevant part of the concept is the attitudes, as I will argue below.

2. Attitudes and stereotyping The theoretical starting point for this study is social identity theory and its relation to language attitudes. The assumption is, as in all of the papers in this collection, that subjective processes, and here specifically language attitudes, play a very influential role in the processes of language variation and change. Consequently, if we want to understand such processes, we need to focus on the subjective processes underlying them. It is widely recognised that ‘subjective attitudes towards language […] are important, for example, in the study of linguistic change, and [they] can often help to explain why a dialect changes when and how it does’ as Trudgill puts it (1995: 10). Or even stronger: ‘the underlying cause of sociolinguistic differences, largely beneath consciousness, is the human instinct to establish and maintain social identity’ as Chambers states (1995: 250). Thus, it is of great importance to investigate the social identities that are connected to different kinds of language use. This means that understanding the process of language change requires insight into the categorisations and stereotypes connected to the linguistic variation.

In language attitude research researchers are often concerned with the stereotypes connected to different kinds of spoken language. The concept of stereotypes was mainly developed by Tajfel and is part of the social identity theory (Tajfel 1981, Hogg and Abrams 1988). In the words of Hogg and Abrams: “Stereotypes are generalisations about people based on category

membership. They are beliefs that all members of a particular group have the same qualities, which circumscribe the group and differentiate it from other groups. A specific group member is assumed to be, or is treated as, essentially identical to other members of the group, and the group as a whole is thus perceived and treated as being homogenous.” (1988: 65)

What the respondent does in speaker evaluation experiments is stereotyping: on the basis of language use the respondent assigns specific characteristics to the speaker, e.g. ’tall’, ’nice’, ’trustworthy’, etc. It is important to note that these characteristics are generally not idiosyncratic, they are part of a shared and agreed-upon characteristic of a social category, which the speaker is believed to belong to. The claim that these beliefs are shared among members of a society or a group is supported empirically by the results from studies that used the speaker evaluation technique (e.g. Garrett et al. 2004; Kristiansen 1991, 1999; Maegaard 2001), and theoretically by the social identity theory (e.g. Tajfel and Turner 1986; Hogg and Abrams 1988). By use of the speaker evaluation technique (i.e. the matched guise technique and its derivatives, see Garrett et al. 2003 or Ryan et al. 1988 for an overview) the researcher elicits stereotypical descriptions of what is perceived to be a distinctive group associated with a particular language use. 3. Overt and covert attitudes The concept of covert attitudes is highly debated. It is often stated that covert norms, covert prestige or covert attitudes exist, but as Labov puts it: “Though investigators speak freely of “covert norms”, the main evidence for their existence is simply the fact that non-standard forms persist” (2001: 512). Other scholars, however, do see these covert norms as evidenced in research. In his famous study from Norwich, Trudgill found that “’covert prestige’ can be associated with certain linguistic forms, and that it is possible in some cases to provide evidence to show that this is in fact the case. […] Covert prestige […] appears to lead to linguistic changes ‘from below’” (1983: 184-5). The discussion of overt vs. covert attitudes is linked to the division of direct vs. indirect methods for collecting data on language attitudes. While the use of direct measures (e.g. interviews or questionnaires asking subjects to rank different

dialects) elicits overt attitudes, it is not necessarily the case that indirect measures like verbal guise experiments elicit covert attitudes. Nevertheless, in Denmark it has been shown in several studies that the attitudes elicited without the respondents knowing what they are participating in, are very different from the attitudes they display when asked directly. The former, indirect, approach yields attitudinal patterns which correspond well with the changes in language use in the Danish speech community. The second, direct, approach results in attitudes which reproduce the commonly established norms that one finds in the media, in institutions, in political arguments etc. This overt evaluative pattern corresponds less well with the change in the pattern of language use (Kristiansen 1991, Maegaard 2000). The fact that these two kinds of attitudes differ is the reason why it is important to collect data without the subjects knowing about the purpose of the investigation, particularly from a language change perspective.

4. The method Data was collected using the speaker evaluation technique. Because my main focus is on language change, the data collection was carried out among young people in different types of schools in Århus. One reason for focusing on young people in a study like this is that young people’s language use might to some extent become standard language use in the future (see Røyneland 2001 for a discussion of this). Another reason is that the responses in a speaker evaluation experiment are social categorisations, and being at a transitional time of life in which the process of categorisation is very important and salient, adolescents are very likely to react more profoundly in this kind of experiment.

4.1 Respondents and speech samples All in all there were 170 respondents, 84 girls and 86 boys, 115 from the academically oriented high schools, the ’gymnasium’, and 55 from the trade-oriented high schools, the ’teknisk skole’. Data from the latter was collected in two kinds of schools, one that educates pupils to become builders, carpenters, painters etc, and one that educates pupils to become chefs, waiters, bakers etc. It was intended that the number of respondents from the two types of high schools should be almost the same, but practical problems resulted in the quite low number of respondents from the ‘teknisk skole’. The age of the respondents ranged from 16 to 18 in the ‘gymnasium’, and 16 to 21 in the ‘teknisk skole’ with 51 of them being between 16 and 18. 168 of the respondents reported that they spoke the Århus accent themselves. The last two reported that they were born in Copenhagen/Sealand, and that they did not speak the local accent, but they had both lived in Århus for more than ten years, so they were included in the study. The speech samples were selected to represent three different accents: 1) Århus, 2) conservative Copenhagen and 3) modern Copenhagen. The speakers

were 16-17 years old, six boys and six girls, two representing each accent. I recorded 55 persons from four ‘gymnasiums’ in Århus and Copenhagen, and chose the twelve speakers from among these, based on their accent and the content of the recordings, which was supposed to be relatively neutral. The idea was that if two speakers of the same accent were evaluated similarly and speakers of different accents differently, it is likely that the evaluation differences would be due to accent differences, and the evaluations of the speakers could be grouped according to accent. I included the gender difference because language use quite possibly is evaluated differently according to the gender of the speakers. The gender difference is not regularly included in verbal guise experiments, but since we know gender to be an important factor in the processes of language variation and change (e.g. Eckert 1989, 2000), it was included in this study. The samples were 20-30 seconds long and the speakers all talked about the same topic: ‘What’s a good teacher like?’ with almost the same content. This subject was chosen because it is something all the young people (both the speakers and the respondents) could relate to since they were all at school and had been for at least ten years. It turned out to be an easy topic for the speakers to talk about, and furthermore, they had roughly the same comments about it. This way the samples consisted of natural speech and the content was kept almost constant.

Apart from the intonation pattern, the Århus samples contain very few differences from the Copenhagen samples. I could have chosen a ‘broader’ variant of the Århus accent but then the possibility of it being a regional norm would have been quite small. This way, I aimed at eliciting the attitudes towards a language use that is recognisably Juttish but still very close to standard Copenhagen.

The two types of Copenhagen speech do not represent a ‘high’ and a ‘low’ accent. I have chosen to use the term ‘conservative’ instead of ‘high’ because the division between ‘high’ and ‘low’ implies a class division connected to the two types of Copenhagen that we have no evidence for. However, the conservative Copenhagen samples are very close to what has been referred to as “high Copenhagen” (e.g. Brink and Lund 1975; Kristiansen 1999). The modern Copenhagen samples, on the other hand, are not just “low Copenhagen”. They share some of the features with this accent but they also contain a few other items. Table 1 is a brief description of the samples and the main differences between them. The description is based on Brink and Lund (1975), Brink et al. (1991), Nielsen (1959), Nielsen (1998), and Nielsen and Pedersen (1991).

4

Copenhagen intonation. One instance of the Copenhagen/Sealand discourse particle ‘ik’. instead of before labials and velars (4/5). No shortening of the vowel preceding (4/4). instead of for ‘jeg’ (1/1)

CCG

10

Copenhagen intonation. No shortening of the vowel preceding (1/1). instead of for ‘jeg’ (2/2)

1

Copenhagen intonation. No shortening of the vowel preceding (2/2). instead of in ‘tøjler’(2/2)

CCB

7 Copenhagen intonation. No shortening of the vowel preceding (1/1). instead of in ‘fremmest’ (1/1)

2

Copenhagen intonation. instead of before labials and velars (2/2). instead of in ‘resten’ (1/1). instead of for ‘jeg’ (1/1)

MCG

8

Copenhagen intonation. One instance of the Copenhagen/Sealand discourse particle ‘ik’. Lengthening of stressed short vowels (3/7). instead of before labials and velars (2/2)

5

Copenhagen intonation. Staccato rhythm. instead of (5/5). Velarised and shortening of the preceding vowel (2/2)

MCB

11

Copenhagen intonation. Staccato rhythm. instead of in ‘interesse’ (1/1). Velarised and shortening of the preceding vowel (1/1)

6

Århus intonation. instead of (1/1) in ‘hård’. instead of for ‘jeg’ (3/3)

ÅG

12 Århus intonation. instead of for ‘jeg’ (1/1)

3

Århus intonation. Realisation of the vowel preceding (1/1). The ending -et pronounced instead of . instead of for ‘jeg’ (1/1)

ÅB

9

Århus intonation. instead of (2/4). instead of for ‘tager’ (2/2). The ending -et pronounced instead of (1/2). instead of for ‘jeg’ (2/2)

Table 1. Linguistic description of the speech samples. For each variable the number of occurrences, and the number of manifestations of a certain variant is stated in parenthesis. For instance ‘The ending -et pronounced instead of (1/2)’ means that there are two occurrences of the variable -et and one of them is manifested as . For each of the speakers, accent and gender is stated (CCG meaning Conservative Copenhagen, Girl etc.).

The main difference between the Århus and the Copenhagen samples is the intonation patterns. In simplified terms, the difference is that, in the Copenhagen accents, the tone rises on the low-stressed syllables, whereas, in the Århus accent, the tone rises on the high-stressed syllables. The intonation patterns are probably the most obvious characteristics of the different Danish accents, but they are poorly described (Grønnum 2001: 305). Another marker of the speakers’ Århus origin is the pronunciation of the ending –et, which in the Copenhagen accents is pronounced and in the Århus accent . Furthermore, one of the Århus speakers has instead of Copenhagen or , and another one has instead of Copenhagen . Two of the Århus speakers also differ with respect to the pronunciation of jeg (1. pers. sing.) using a Juttish variant instead of Copenhagen .

The two Copenhagen accents differ especially with respect to some vowels, the pronunciation of t and !, and a certain type of monophthongisation. The modern accent has before labials and velars where the conservative has . The modern accent has lengthening of stressed short vowels. It has where the conservative has . The modern accent also has velarised and furthermore the vowel preceding is very shortened. Something similar happens to semi-vowels like in words like lavet which in the conservative accent is pronounced . In the modern accent the diphthong merges with the which means that the word is pronounced with a very short vowel. In some descriptions it is even transscribed as in order to stress this fact. In the modern accent the conservative is pronounced .

One of the conservative copenhagen speakers has instead of twice in the word tøjler which is a more conservative manifestation of -øj.

Furthermore, the modern accent has a sort of staccato rhythm that has not yet been formally described. It could be something like the staccato rhythm reported from different studies of language use in multicultural and multilingual youth communities in European cities, including Copenhagen (e.g. Hansen and Pharao in press, Keim 2004, Quist 2000,), though not as strongly manifested.

Though the differences between the speech samples are small, as can be seen from the above description, it is quite easy for the respondents to categorise the speakers as speakers of different accents. This was tested after the speaker evaluation test, and the results showed that the respondents had very clear impressions of the speakers’ accents (see table 2). In this test, the informants were asked to categorise the speakers’ language into the categories århusiansk (Århus), københavnsk (Copenhagen), or rigsdansk (the common term for the standard accent), or any combination of the three. The wording in the questionnaire was not the same as I use in this paper because I chose to use the

folk terms københavnsk and rigsdansk instead of moderne københavnsk and konservativt københavnsk, which are labels that no non-linguist would use. Rigsdansk is a broader term than konservativt københavnsk, in the sense that rigsdansk is not tied to a particular geographical place, and, especially outside of Copenhagen, is ideally considered a non-Copenhagen way of speaking (Ladegaard 2001). However, in practice, it is the conservative Copenhagen accent that is referred to when rigsdansk is discussed (Kristiansen 2001).

Å K R Å-K Å-R K-R Å-K-R

sp. 1 CCB 12.2 26.2 42.1 0.6 4.3 14.6 - sp. 2 MCG 11.7 76.5 4.9 2.5 0.6 3.7 - sp. 3 ÅB 68.7 3.7 16.6 1.2 9.2 0.6 - sp. 4 CCG 6.1 61.0 17.1 4.3 2.4 9.1 - sp. 5 MCB 15.2 53.7 17.1 3.0 1.8 7.3 1.8 sp. 6 ÅG 87.7 5.5 3.1 - 2.5 0.6 0.6 sp. 7 CCB 10.4 18.4 50.3 2.5 4.3 13.5 0.6 sp. 8 MCG 6.1 84.8 3.7 1.8 - 3.7 - sp. 9 ÅB 58.3 2.5 22.7 1.2 14.7 0.6 - sp. 10 CCG 16.0 46.6 28.2 1.8 1.2 6.1 - sp. 11 MCB 22.8 45.1 16.0 3.7 6.2 4.3 1.9 sp. 12 ÅG 78.4 4.9 8.6 1.9 5.6 0.6 -

Table 2. Perceptions of the speakers’ accents as either Århus (Å), københavnsk (K) or rigsdansk (R) or both Århus and rigsdansk (Å-R) etc. Numbers are percentages of the total number of evaluations for each speaker. For each of the speakers, accent and gender is stated (CCB meaning Conservative Copenhagen, Boy etc.). The shaded numbers show which category was the most frequent one. Table 2 shows that the speech samples are generally considered as expected: 1 and 7 are mainly rigsdansk, 2 and 8 are mainly københavnsk etc., but there are important exceptions. Speech samples 4 and 10, the conservative Copenhagen girls, are considered more københavnsk than rigsdansk. This could result from phonetic differences between girls and boys, i.e. boys using more conservative variants of the variables, but the phonetic analysis did not reveal any such differences. Speaker 4 uses the discourse particle ‘ik’, which is considered a Sealand/Copenhagen feature, and this could explain why 61% labelled her speech københavnsk. However, there is no obvious explanation why almost 47% labelled sample 10 københavnsk. Overall, the Århus girls are more often than the Århus boys considered to speak Århus; the same is true for modern Copenhagen girls versus modern Copenhagen boys with regard to københavnsk. In the case of the modern Copenhagen samples this could be explained by actual linguistic differences, since girls and boys differ somewhat, especially with respect to the

‘staccato rythm’, which could be what causes the different perceptions. However, there is no obvious linguistic explanation in the case of the Århus speakers. The reason behind these differences in perceptions of the speech samples might be sought in the different expectations the respondents have about girls and boys. Labov has described the overall pattern in sociolinguistic research on gender differences as a pattern showing that female speakers tend to lead in “changes from above” whereas male speakers tend to lead in “changes from below”. According to Labov, a similar pattern is seen when it comes to stable sociolinguistic variables, where female speakers tend to show a lower rate of stigmatised variants and a higher rate of prestige variants than male speakers (Labov 2001: 266 pp). This has also been shown to be the case in Denmark with respect to the Århus/standardvariation among Århus youth (Nielsen 1998, see also Nielsen and Nyberg 1992) and with respect to the high/low Copenhagen variation in Copenhagen (Gregersen and Pedersen 1991). In the case of Århus the language use can be said to be changing ‘from above’, while the situation in Copenhagen is partly a case of ‘change from below’, partly a case of stable sociolinguistic variables. If it is a common experience that girls tend to speak more standard than boys, then listeners are likely to expect girls to speak more standard than boys, and they will react when they are presented with girls who do not. As a consequence they are likely to evaluate the girls’ speech as less standard than the boys. This could explain why the perceptions differ according to the speakers’ gender. Still, the overall pattern is that the speech samples are perceived as intended.

4.2 The questionnaire Usually researchers using speaker evaluation experiments use semantic differential scales like ’interesting-boring’, ’trustworthy-untrustworthy’, etc. The use of scales like these has advantages as well as disadvantages. It is possible to collect large amounts of data and to analyse them quite easily with the use of statistic analysis. On the other hand, using predefined scales simply taken from other studies with other respondents can also prove to be problematic. This has been shown especially by Garrett, Coupland and Williams in their attitude studies from Wales (see Garrett et al. 2003, ch. 3, or Garrett et al. 2004). First of all, the scales presuppose the relevance of certain traits and the irrelevance of others. In some studies a pilot study of a comparable group of informants has been carried out before the main study, in order to determine which traits are actually relevant to the informants. However, in most studies this has not been done. Secondly, many relevant categorisations might prove to be impossible to make into scales (e.g. categorisations like ‘farmer’ or ‘wears glasses’). They are not necessarily constructed in terms of ’more or less’, that is, in terms of scaleable entities. Thirdly, the evaluation scales are often adult-like versions of something which researchers feel is unsuitable for print (see Garrett et al. 2003:

64). But if the items have been translated into adult-like language then they may not carry the same meaning to the younger informants anymore, and this is of course problematic.

An alternative is to use open-ended questionnaires, but there are problems connected with this too. Using this type of questionnaires certainly implies limitations to the number of respondents used in the study. The coding of qualitative data is always very time consuming and therefore the amount of data is often smaller in this kind of studies than in strictly quantitative studies. Moreover, the coding often tends to be less objective, and this means that the reliability of the results is not as strong. Despite these weaknesses open-ended questionnaires seem very useful when we want to study the stereotypes connected to different accents because they offer a more detailed but also a more complex picture of the stereotypes.

I designed four types of questionnaires for the study, but in this paper I will mainly focus on the first, which is quite simple. There is one question for each speech sample: Hvad er dit umiddelbare indtryk af denne person? (‘What is your immediate impression of this person?’). This is practically the same question that Garrett et al. asked in one of their studies of language attitudes in Wales (2004). The respondents in the Århus study had no idea that they were taking part in a language attitude study. They were told nothing about the reason for the experiment, but afterwards we talked about it, and they were free to withhold their responses if they decided that after all they did not want to participate. However, none of them did so. I also asked them what they thought it was all about. None of them suspected it to have anything to do with language. This is only possible because of the relatively small differences between the three accents. In a bilingual community like the Welsh/English in Wales or the French/English in Canada this could probably not be done. The risk that the informants realise that the study is about language is bound to be greater when the speech samples are actually in different languages. However, this may not affect the results of the study, judging from a study by Giles (1970) who compares the results from a matched guise experiment with the results from a study of attitudes towards the same accents using a direct method. The results are very similar in Giles’ study, but this might be due to the design of the study. In the study the same subjects were used in the matched guise experiment and in the direct study (ranking of dialect labels), and they were given the questionnaires for both studies at the same time. Some subjects did the label ranking before and others after the matched guise experiment (Giles and Powesland 1975: 29). In any case the possibility that the subjects were actually aware of the nature of the experiment when they evaluated the guises is quite high and the result possibly a consequence of this.

As has been shown by Kristiansen (e.g. 1991), the results from indirect and direct studies of language attitudes are not always similar. There is a

possibility that the differences between Giles’ and Kristiansen’s results have to do with methodological differences and not necessarily with actual differences in the attitudinal patterns in the communities under study. In this context, though, the important thing is that it was actually possible to collect the data without the respondents realising that they were taking part in a language attitude study. This is important because it seems that the responses in this kind of experiment can be interpreted as covert attitudes, as argued earlier.

5. The data When asked for their immediate impression of the speaker, the respondents write down all sorts of things, describing various parts of the personality and life they imagine the speaker has. Sometimes they even describe the speaker’s hair colour, details about his or her childhood, favourite band etc. However, most of the responses are simpler than that. Examples of responses are: (1) lidt oprørsk (‘a little rebellious’)

‘rebel’ (‘rebel’) egen mening (‘own opinion’) ikke påvirket af andre (‘is not influenced by others’) selvstændig (’independent’) måske flippet (’perhaps alternative style of clothing’) (response to speaker 8, modern Copenhagen, girl)

(2) irriterende (’annoying’)

kedelig (’boring’) ikke særlig livlig (’not very lively’) fra landet (‘from the countryside’) (response to speaker 3, Århus, boy)

(3) han er rolig (‘he is calm’)

nede på jorden (’down to earth’) lidt gammeldags (‘a bit old-fashioned’) principfast (‘high-principled’) og stolt (‘and proud’) (response to speaker 7, conservative Copenhagen, boy)

I have listed the above responses so that each line contains one item, that is one word, phrase, sentence, etc. with a specific semantic content. In order to systematise the responses, I grouped them together according to their semantic content. I set up larger categories like ’rebellious’, ’intelligent’ and ’boring’ to contain all items with a similar meaning. This means that e.g. the ’intelligent’ category would not only contain items like ’smart’, ’clever’ and ’intelligent’, but

also items like ’she is certainly not stupid!’. The method used in the construction of categories is based on grounded theory (e.g. Strauss and Corbin 1998). Grounded theory offers a method in which the coding of data is constantly revised and the items are repeatedly compared. I found it useful for this kind of data coding. Another option would be not to construct larger categories at all and work with the literally written items (as in Garrett et al. 2004). A total of 70 categories obtained (see table 3). Sometimes one response contains the same item expressed in different ways (e.g. ‘a little rebellious’ and ‘rebel’ in the first response above), and in these cases it is only counted once. frequency evaluative dimension 1 sympatisk (nice) 133 attractiveness 2 intelligent (intelligent) 131 superiority 3 rolig (calm) 119 dynamism 4 kedelig (boring) 114 dynamism 5 ung (young) 110 dynamism 6 social (social) 107 attractiveness 7 frisk (lively) 101 dynamism 8 selvsikker (self-confident) 101 dynamism 9 usikker (insecure) 90 dynamism 10 ambitiøs (ambitious) 83 superiority 11 selvstændig (independent) 80 superiority 12 autoritetstro (establishment type) 78 superiority 13 genert (shy) 70 dynamism 14 almindelig (ordinary) 67 - 15 selvhøjtidelig (self-important) 64 attractiveness 16 medmenneskelig (humane) 57 attractiveness 17 dum (stupid) 51 superiority 18 fornuftig (rational) 49 superiority 19 ligeglad (careless) 49 superiority 20 stræber (creep) 46 superiority 21 smart (chic) 44 attractiveness 22 engageret (committed) 44 dynamism 23 pligtopfyldende (consciencious) 44 superiority 24 glad (cheerful) 40 attractiveness 25 doven (lazy) 38 dynamism 26 moden (mature) 34 superiority 27 velformuleret (articulate) 30 superiority 28 festmenneske (partygirl/boy) 29 dynamism 29 eftertænksom (thoughtful) 28 superiority 30 sjov (funny) 25 dynamism 31 bestemt (firm) 23 superiority

32 københavner (Copenhagener) 23 - 33 snobbet (snobbish) 23 superiority 34 ærlig (honest) 22 attractiveness 35 initiativrig (enterprising) 21 dynamism 36 forvirret (confused) 20 superiority 37 ansvarlig (responsible) 18 superiority 38 overfladisk (superficial) 18 attractiveness 39 politisk interesseret (interested in politics) 18 - 40 nørd (nerd) 17 superiority 41 alvorlig (serious) 16 attractiveness 42 flot (good looking) 16 attractiveness 43 jyde (Jut) 16 - 44 overklasse (high class) 15 superiority 45 kreativ (creative) 14 dynamism 46 spændende (interesting) 14 dynamism 47 flippet (alternative type) 13 attractiveness 48 ryger (smoker) 13 - 49 sej (tough) 12 attractiveness 50 uselvstændig (dependent on others) 12 superiority 51 fantasifuld (imaginative) 11 dynamism 52 populær (popular) 11 attractiveness 53 bøsse (gay) 10 attractiveness 54 intellektuel (intellectual) 10 superiority 55 usympatisk (unpleasent) 10 attractiveness 56 poppet (mainstream style of clothing) 10 attractiveness 57 fra byen (from the city) 9 - 58 fra landet (from the country) 8 - 59 fræk (cheeky) 7 attractiveness 60 langsom (slow) 7 dynamism 61 naiv (naive) 7 attractiveness 62 sjællænder (Sealander) 7 - 63 skater (skater) 7 attractiveness 64 ledertype (leader type) 6 superiority 65 maskulin (masculine) 6 attractiveness 66 feminin (feminine) 5 attractiveness 67 hiphopper (hip hopper) 5 attractiveness 68 rebelsk (rebellious) 5 attractiveness 69 underklasse (lower class) 5 superiority 70 århusianer (Århusian) 5 - Residual categories: summary 114

pedagogical matters 60 uncategorisable items 516 Table 3. Total list of coding categories listed according to frequency in the data, and distributed into three evaluative dimensions. I only maintained categories if the number of items belonging to them is at least 5. Out of 3241 items 690 did not fit into any of the 70 categories. This number corresponds to 21% of the items and may seem overwhelming. However, in a number of these instances the response simply reiterates what the speaker just said and should be subtracted from the number of uncategorisable items. Another group of respondents believed the speakers to be teachers and wrote about pedagogical matters. All of these responses are included as ‘residual categories’ in table 3. This leaves 516 uncategorisable items which amounts to 16% of the total. Many of these uncoded items consist of descriptions of the lives of the speakers, and though very interesting, these items are quite difficult to code in a meaningful way. These are items like: (4) Opvokset i trygt miljø med mor og far. Nu lettere provokerende

med hang til at gå med nogle udlændinge. ‘Grew up in a safe environment with mom and dad. Now slightly provocative with a bent for hanging out with foreigners.’ (response to speaker 1, conservative Copenhagen boy)

(5) Bor evt. i boligblok. Mangler evt. tryghed. ‘Possibly, lives in an apartment building. Possibly, needs security.’ (response to speaker 2, modern Copenhagen girl)

(6) Pigen med de mærkeligste venner. ‘The girl with the weirdest friends.’ (response to speaker 8, modern Copenhagen girl)

These items are all legitimate responses to the question the respondents were asked (‘What’s your immediate impression of this person?’), but they are very difficult to code into categories. It is obvious that a lot of interesting data gets disqualified using this kind of coding procedure, and this shows the problems with ‘translating’ qualitative data into quantitative.

Similar problems arise in connection with the responses containing drawings, such as (7) A drawing of a cardiogram fading out, with the text: ingen respons

(‘no response’)

(response to speaker 9, Århus boy). The rest of the uncategorisable items are responses like ‘red-haired’, ‘polite’, ‘football player’ that are not difficult to make into categories, but were not sufficiently repeated to constitute a category.

Some of the labels in table 3 might need a few comments. The term Danish word stræber (literally meaning ‘striver’) is very frequent in the data and is very commonly used among Danish adolescents. Used by young people stræber is a derogatory word, meaning someone who works hard in school, is eager to please the teacher, does all his/her homework, etc.

Københavner (‘Copenhagener’) is quite frequent in the data as well (rank 32). Apart from meaning ‘someone from Copenhagen’ it carries other meanings. In many responses it is supplemented by comments like ‘Copenhagener, you know, someone who thinks she’s something, but really she isn’t’ or ‘Copenhagener: very self-satisfied’.

Also the label ung (‘young’) is very frequent (rank 5). This was unexpected in a way because the respondents were told before the experiment that they were about to listen to 12 young speakers their own age. The frequent use of the label ‘young’ shows that ‘young’ is part of the respondents’ immediate impressions of the speakers, i.e. part of the stereotype. Seen this way, the high frequency of ‘young’ is not surprising.

6. Results 6.1 Overview of the frequency of the labels The results can be viewed from different perspectives. Firstly, table 3 provides information on the frequency of labels and the distribution of them into the three dimensions suggested by Zahn and Hopper (1985): superiority (intellectual status and competence, social status, and speaking competence), attractiveness (social and aesthetic appeal) and dynamism (social power, activity level, and the self-presentational aspects of speech). Most of the labels relatively unproblematically fall into one of the dimensions, but for some of them a discussion is needed. ‘Young’ has been classified as a dynamism label, mainly because of its distribution in the data, where it often appears together with other dynamism labels. In further support of this classification, it can be mentioned that a matched guise study of young Danes’ attitudes towards speakers with some versus no English colouring of their speech showed results for the ‘young-old’ scale to be similar to those for the dynamism scales (Andersen 2004: 180).

The nine labels that failed to fit into any of the three dimensions are, apart from ‘interested in politics’, ‘ordinary’ and ‘smoker’, all labels that have to do with the speaker’s origin.

The list of labels in table 3 looks very different from the findings in Garrett et al. (2004). The keyword categories reported here are: ‘taboo/pejorative

items’, ‘boring’, ‘Welshness/Englishness’, ‘farming and other activities’, ‘sexuality’, ‘rich/posh’, ‘drugs’, ‘boasters and bullshitters’, ‘troublemaker and funny’. Apart from the different coding procedures used in this study and in Garrett and colleagues’, the studies seem to be comparable. Nevertheless, there are important differences. The age of the respondents is almost the same (15-16 years in Garrett and colleagues’, 16-18 (with four older respondents) in the present study). However, the fact that the respondents in the Welsh context had not yet finished secondary school and moved on to high school might be of some importance here, since the two groups of respondents then are in different stages in the educational system. Also, the content of the speech samples is very different in the two studies. In the Welsh context, the speakers were telling stories, and in the Danish context they talked about good teachers. Telling a story to your class mates and giving your opinion on good teaching in an interview with a stranger are two very different speech events, and they may influence the data in very different ways. Finally, in the Welsh context one item on the questionnaire referred to issues of ethnicity (with the question ‘how Welsh do you think this speaker sounds?’) which may have drawn some attention to the fact that the respondents were participating in a language attitude study.

These methodological differences could account for the different results but on the other hand it should also be noted that young people in Århus and in Wales are not the same, and they might use language and categories differently.

Out of the 70 coding categories, 22 belong to the superiority dimension, 23 to the attractiveness dimension, 16 to the dynamism dimension, and 9 did not fit into any of these dimensions. The distribution of labels is skewed, with few dynamism labels compared to the number of superiority and attractiveness labels but the distribution of actual items into the different dimensions looks rather different. Here, 870 fall into the superiority dimension, 625 into the attractiveness dimension and 908 into the dynamism dimension. How these dimensions are in fact unequally represented for the different speakers can be seen from table 4.

Speaker Superiority Attractiveness Dynamism Total # % # % # % # 1 (CCB) 133 52 56 22 68 26 257 2 (MCG) 51 27 76 40 64 34 191 3 (ÅB) 51 29 28 16 99 56 178 4 (CCG) 69 36 43 22 82 42 194 5 (MCB) 36 16 98 43 95 41 229 6 (ÅG) 70 43 33 20 58 36 161 7 (CCB) 138 58 34 14 68 28 240 8 (MCG) 45 25 63 35 71 40 179 9 (ÅB) 74 38 40 20 82 42 196

10 (CCG) 73 39 60 32 56 30 189 11 (MCB) 41 18 68 29 123 53 232 12 (ÅG) 71 51 26 19 42 30 139

Table 4. The distribution of labels into the three dimensions for each speaker. The letters in parenthesis indicate the accent and gender of the speaker (e.g. ‘CCB’ means ‘conservative Copenhagen, boy’). The table shows, for instance, how superiority is a very frequent dimension in the responses to speakers 1 and 7 (conservative Copenhagen boys) whereas this is not the case in the responses to speakers 5 and 11 (modern Copenhagen boys). These numbers can be interpreted as indications that different dimensions are relevant in responses to different speakers. However, the numbers in table 4 do not reveal anything about the actual semantic content of the items used. Thus, the many dynamism items in the responses to speaker 3 (Århus boy) and speaker 11 (modern Copenhagen boy) are surely not of the same semantic content. This can be seen when presenting the data in another way. 6.2 Speaker profiles The analysis provides relatively clear evaluation profiles of the twelve different speakers with regard to highly frequent categorisations. This gives a less broad and more specific picture of the attitudes, compared to the presentation above, but at the same time this picture is based on only the highly frequent categories and thus it leaves out a great deal of other categorisations. Table 5 shows the most frequent categorisations for each speaker.

The degree of agreement among the respondents regarding the stereotypes is quite high. It could be expected that some of the profiles were contradictory, i.e. contained labels with opposing semantic content. This is not the case for any of the speaker profiles (except perhaps for the evaluations of the conservative Copenhagen girls as both ‘calm’ and ‘lively’), a fact that indicates that the respondents share interpretive norms. If two opposing norms were present, one would expect contradictory labels to occur within the same profile.

The profiles seem relatively distinct but at the same time it is seen that the same labels appear in several of the evaluation profiles. This way, it is not so much the ascribing of a certain label to the speakers, as it is the combination of the labels that is important. In table 5 the speaker profiles have been grouped pairwise according to accent and gender, and the labels in italics represent the categories that appear in both speaker profiles. Thus, both of the conservative Copenhagen speaking girls are evaluated as ‘intelligent’, ‘ambitious’, ‘calm’ and ‘social’, whereas speaker 4 is in addition considered ‘lively’ and speaker 10 ‘nice’ and ‘self-confident’. The conservative Copenhagen boys are considered ‘intelligent’, ‘calm’, ‘nice’, ‘self-confident’, ‘conscientious’ and ‘boring’, and

speaker 7 is in addition an ‘establishment type’, ‘ambitious’ and ‘independent’, whereas speaker 1 is ‘humane’.

The modern Copenhagen speakers are ‘young’, ‘social’, ‘independent’ and ‘lively’. The girls are furthermore considered ‘self-confident’ and speaker 2 is also a ‘smoker’. The boys are both considered ‘chic’ and speaker 5 is additionally regarded as ‘happy’, ‘humane’, ‘committed’, ‘nice’, ‘calm’ and ‘funny’, while speaker 11 is ‘enterprising’ and ‘creative’.

Sp. 4

lively (20), ambitious (14), calm (11), intelligent (11), social (10)

Girls

Sp. 10 nice (22), ambitious (17), intelligent (17), social (16), calm (14), self-confident (13)

Sp. 1

intelligent (31), nice (20), self-confident (15), conscientious (13), humane (12), calm (10), boring (10)

CC

Boys

Sp. 7

intelligent (24), establishment type (22), boring (16),calm (16), self-confident (16), ambitious (15), independent (12), nice (11), conscientious (10)

Sp. 2

independent (16), social (16), young (15), self-confident (13), lively (11), smoker (10)

Girls

Sp. 8 young (12), social (12), self-confident (11), independent (10), lively (10)

Sp. 5

social (26), lively (24), happy (13), humane (13), chic (12), committed (12), nice (12), young (11), independent (11), calm (10), funny (10)

MC

Boys

Sp. 11

young (22), lively (13), enterprising (13), chic (12), creative (12), independent (11), social (10)

Sp. 6

stupid (13), insecure (10), boring (13), ordinary (11), self-important (10), establishment type (10), snobbish (10)

Girls

Sp. 12

insecure (23), shy (15), stupid (15), boring (14), establishment type (14), ordinary (11), careless (11), ‘creep’ (10), lazy (10), self-important (10)

Sp. 3

boring (24), shy (18), calm (15), insecure (12), nice (12), intelligent (11), careless (10), lazy (10)

Å

Boys

Sp. 9 nice (20), calm (19), insecure (18), boring (15), intelligent (15), shy (11), establishment type (10)

Table 5. The most frequent categorisations. At least 10 categorisations per speaker. The labels are listed according to frequency which is indicated by the numbers in parenthesis. The labels in italics show which labels appear in the

profiles of both speakers of a certain accent and gender. CC, MC and Å mean conservative Copenhagen, modern Copenhagen and Århus, respectively. Using the terminology from Zahn and Hopper (1985), the conservative Copenhagen speakers are categorised highly in the superiority dimension: ’ambitious’, ‘conscientious’, ‘independent’, ’intelligent’, and an ‘establishment type’, whereas the modern Copenhagen speakers are evaluated highly in the dynamism dimension: ‘committed’, ‘creative’, ‘enterprising’, ‘funny’, ‘lively’, ’self-confident’, and ’young’. It seems, then, that conservative and modern Copenhagen carry very different values. Even though it is not always obvious to tell if a label carries negative or positive meaning, it seems that the conservative and the modern speech samples are evaluated positively, but in different dimensions. Actually, this is not a new finding; it has been reported several times in previous attitude studies from other parts of the country (e.g. Kristiansen 2001, Kristiansen et.al. 2002, Maegaard 2000). The new thing is that this result is also found in Århus, and by the use of a slightly different method. As mentioned earlier in this paper, it has been interpreted by Kristiansen (2001) as an indication of the co-existence of two different norms for two different domains among Danish youth. Kristiansen suggests that young people in Denmark orient themselves to two standards, one that has to do with the school and the business world, and one that has to do with the media. Thus, the conservative Copenhagen accent is seen as ‘the language of the school’ while the modern Copenhagen accent is seen as ‘the language of the media’.

As mentioned before, distinguishing between positive and negative evaluations is very difficult. Some labels are quite obviously either positively or negatively loaded (e.g. ‘nice’ or ‘boring’), but for others the loading is much more unclear. For example, is ‘ambitious’ always a positive label? Is ‘cheeky’ always negative? This is probably not the case at all. The focus point here must be the stereotypes the labels are part of. By focussing on the stereotypes – in this study represented by the speaker profiles – the connection between labels becomes clearer and their interpretation easier.

It is interesting that the labels ascribed to the Copenhagen speakers are almost entirely positive. The speakers are not at the same time evaluated negatively in other dimensions. This indicates that it is not important to the respondents whether or not e.g. the modern Copenhagen girls are ambitious, intelligent, etc. or not. The important thing seems to be that they are self-confident, independent, etc. This illustrates why it may be problematic to use the same evaluation scales for all samples in a study of language attitudes, because such a procedure implies the relevance and importance of all the presented personality traits to the evaluation of all the speech samples. This assumption is not supported empirically by the results of this study. The respondents will probably be able to evaluate the speakers with respect to many traits, but if half

of the traits are irrelevant to them then using them as scales in a questionnaire will give a wrong impression of the attitudes. The methodological consequence of this must be not to rely solely on the use of semantic differential (or other types of) scales.

The Århus profiles are a different story. Both girls and boys are evaluated as ’boring’ and ’insecure’. These are quite negative labels, and they are, for the girls’ part, supplemented by characterisations like ‘careless’, ‘creep’, ‘lazy’, ‘ordinary’, an ‘establishment type’, ‘self-important’, ‘shy’, ‘snobbish’, and ‘stupid’. The boys are on the other hand considered: ‘calm’, ‘careless’, ‘intelligent’, ‘lazy’, ‘nice’, ‘shy’ and an ‘establishment type’, which shows at least some positive evaluations. Judging from many studies of attitudes towards dialect and standard, which operate with a division into the evaluative dimensions of sociability/solidarity versus competence/status, one would expect the dialect speakers to be evaluated positively in the sociability dimension (e.g. Giles and Coupland 1991: ch. 2). That is not the case here. Notwithstanding the ‘nice’ and ‘calm’ attributions accorded to the boys, the overall picture is that the positive reactions towards the Århus accent are practically non-existent. The respondents seem to favour the Copenhagen accents and evaluate the local accent quite negatively. Judging from the present study then, the Århus accent is not a norm ideal among Århus youth. This result is very clear, but as it builds on a relatively small data set, more work needs to be done to make the claim more secure.

6.3 Girls and boys As can be seen in table 5, the evaluations are somewhat gender specific. The evaluations of the conservative Copenhagen girls and boys show very little difference. As mentioned above, for both girls and boys most of the shared labels clearly belong to the superiority dimension. It is interesting, though, that the boys are labelled ‘boring’, while the girls are labelled ‘social’. It seems therefore that the use of conservative Copenhagen might carry somewhat different values according to gender. The boys are considered to possess the traditional traits needed in order to be successful in school, and in addition they are boring. The girls on the other hand, share some of the success-in-school labels, but are at the same time considered ‘social’. Thus, there is some indication that it might be more attractive for a girl to speak with this accent than for a boy.

The difference according to gender is interesting within the modern Copenhagen speakers too. Except for ‘independent’, all labels assigned to modern Copenhagen speakers belong to the dynamism or the attractiveness dimension. However, girls’ and boys’ profiles are quite different. All speakers are evaluated as ‘young’, ‘social’, ‘lively’, and ‘independent’ but the girls are also ‘self-confident’, and speaker 2 is in addition a ‘smoker’. The boys on the other hand are ‘chic’, and speaker 5 is furthermore ‘calm’, ‘happy’, ‘funny’,

‘nice’, ‘humane’ and ‘committed’, while speaker 11 is ‘enterprising’ and ‘creative’. Thus, there seem to be some difference in the evaluations of the modern Copenhagen girls and boys, where girls are seen as social individuals who are managing on their own (‘independent’) and believe they are able to (self-confident), while boys are also seen as social and independent, but at the same time as positive (‘happy’ and ‘funny’), creative and enthusiastic (‘committed’ and ‘enterprising’). This is even more pronounced when analysing the data in another way. Instead of focusing on the highly frequent categorisations, one could choose to focus on the other categorisations ascribed to only certain speaker types. For instance, the labels ‘rebellious’ and ‘smoker’ are not very frequent (5 and 13 occurrences, respectively), but they are found solely in the evaluations of the modern Copenhagen girls, and seem to fit well with the main profile of these girls, which is why the ascription of these labels to them do not seem to be coincidental. The labels ‘smoker’ and ‘rebellious’ twist the profile in a new direction. Smoking and being rebellious is not just being independent but also being in opposition. Being teenagers, what the young girls are thought to be opposing is probably the adult norms. This would mean that speaking the way they do gives the respondents an impression of someone opposing the adult norms.

If one applies the same kind of analysis to the responses to the modern Copenhagen boys, the profile is supplemented by the labels ‘cheeky’, ‘feminine’, ‘interesting’, and ‘skater’ (with 7, 5, 14 and 7 occurrences, respectively). ‘Cheeky’ gives the impression of someone quite provocative but not as strongly opposing the norms as the ‘rebellious’ girls. The label ‘feminine’ is remarkable since it is often stated that the low variety typically carries connotations of masculinity. The modern Copenhagen speakers in this study do have some features which are traditionally associated with low social status, but also new features which apparently do not carry such associations. Labelling the speakers ‘feminine’ might be a symptom of that. The two modern Copenhagen boys were also the speakers most frequently referred to as homosexuals. ‘Interesting’ fits quite well with the main profile, and ‘skater’ is a style that is usually associated with young male urban life, and therefore an obvious label to assign solely to the modern Copenhagen boys.

No other labels apply to only certain speaker types, so this type of analysis can not be extended. This could be an indication that the modern Copenhagen speakers are sometimes seen as very different from the others.

The Århus profiles display a most remarkable gender difference. The girls are evaluated much more negatively than the boys. While the boys do attract a few positive evaluations like ‘calm’, ‘intelligent’ and ‘nice’, the girls are nothing but ‘ordinary’, ‘stupid’ and ‘self-important’. This is probably a typical pattern, and it corresponds to Trudgill’s (1983) findings from Norwich, where the non-standard vernacular carried positive values for male speakers, but not for female.

Generally, the evaluation differences according to gender show tendencies towards stronger reactions when a girl speaks the non-standard accent than when a boy does. It is most clearly seen with respect to the Århus accent, but the modern Copenhagen girls are also seen as opposing the norms. On the other hand the conservative Copenhagen girls are evaluated slightly more positively than the boys. This pattern could be seen as connected to the often reported finding that female speakers tend to lead in the standardisation processes (Eckert and McConnell-Ginet 1999; Labov 2001). If the wish to construct a positive social identity can be seen as the underlying cause of linguistic change, then the Århus girls might be expected to change their language. There seems to be nothing positive associated with the Århus accent when it comes to the girls, and knowing this the girls might gradually change their language in a more valuable direction, i.e. closer to the standard. The boys, on the other hand, are likely to maintain their Århus accent longer, according to the more positive reactions towards the Århus boys in this study. This is very much in line with the general findings in language and gender research. Generally speaking, female speakers are more extreme, both when it comes to the degree of standardness and of non-standardness. However, it seems that this is the case only when the non-standard variables are part of an ongoing linguistic change, whereas male speakers are more extreme when the non-standard variables are ‘old’, as is the case with the Århus accent (Eckert and McConnell-Ginet 1999).

Judging from the present study, it is clearly more accepted when a girl speaks the standard than when she speaks non-standard. With respect to the modern Copenhagen girls, however, it is not obvious how to interpret the issue of being in opposition. To many teenagers perhaps being in opposition is in fact very desirable. After all, adolescence is about growing up and partly taking over, partly rejecting the norms set out by institutions, parents or other adults. Eckert’s findings from Belten High in Detroit support this idea. According to Eckert, language use is very much part of the social practices that constitute the particular social groups. This means that for the pupils at Belten High speaking the way they do is as much part of the identity construction as the clothes they wear, the music they listen to, the activities they engage in. Being a burnout is largely about being in opposition to the institutions and in fact even being stigmatized by them, and this applies to language as well. For Eckert, the crucial point is that “linguistic variables that are prestigious or stigmatized in the abstract are full of positive meaning in the concrete everyday” (2000: 226). It is likely that the evaluations of the modern Copenhagen girls testify to something similar to this. The girls are considered independent and rebellious but still the evaluations are very positive, e.g. ‘lively’ and ‘social’. Thus, they are seen as being in opposition to the adult norms, and perhaps not in spite of that, but rather because of that, they are evaluated quite positively. This is very much in line

with the findings reported by Williams et al. (2004). In that study, it was shown that emergent adults actually associate self-promotional features like cheekiness with good communication.

6.4 Frequent categories The data also allows insight into the categories used by adolescents when doing stereotyping as in this experiment. In table 6 the most frequent categories are listed. They all have a frequency of at least 80 which means that on average around half of the respondents have used these categories in their evaluation. Category # items Evaluative dimension

sympatisk (nice) 133 attractiveness intelligent (intelligent) 131 superiority rolig (calm) 119 dynamism kedelig (boring) 114 dynamism ung (young) 110 dynamism social (social) 107 attractiveness frisk (lively) 101 dynamism selvsikker (self-confident) 101 dynamism usikker (insecure) 90 dynamism ambitiøs (ambitious) 83 superiority selvstændig (independent) 80 superiority

Table 6. Most frequent categories. The striking thing about this list is that in a way it is not striking at all. It is very similar to the traits that have been used in traditional speaker evaluation experiments. If this list can be interpreted as expressing the most relevant evaluation traits for young people in peer evaluation, then dynamism traits seem to be extremely relevant to them. However, it should be noted that in fact some of these dynamism traits come in pairs. This way, ‘lively’ could be combined with either ‘boring’ or ‘calm’ in a semantic differential scale, and ‘insecure’ could be combined with ‘self-confident’. Seen this way, the traits are almost equally distributed in the three dimensions.

7. Concluding remarks The methodological approach which was used in this study combined aspects from different speaker evaluation experiments in order to study the evaluative patterns associated with language variation among young people in Denmarks second largest city, Århus.

There are no indications that the Århus accent functions as a regional norm ideal. The young Århusians seem to favour the Copenhagen accent over

the Århus one. This means that over time, one would expect the Århus accent to be further levelled, especially in the girls’ language, since the Århus-speaking girls are remarkably negatively evaluated. The possible gender differences that the study was designed to discover turned out to be present for all three accents. This shows the importance of including gender as a variable in verbal guise studies in order to get a broader picture of the attitudes.

References Andersen, Margrethe Heidemann. 2004. Engelsk i dansk: Sprogholdninger i

Danmark. Helt vildt sjovt eller wannabeagtigt og ejendomsmæglerkækt? Dansk Sprognævns skrifter 23.

Brink, Lars and Jørn Lund. 1975. Dansk Rigsmål. Copenhagen: Gyldendal. Brink, Lars, Jørn Lund, Steffen Heger and Jens Normann Jørgensen. 1991. Den

Store Danske Udtaleordbog. Copenhagen: Munksgaard. Chambers, Jack K. 1995. Sociolinguistic Theory. Oxford: Blackwell. Eckert, Penelope. 1989. Jocks and Burnouts. Social Categories and Identity in

the Highschool. New York: Teachers College Press. Eckert, Penelope. 2000. Linguistic Variation as Social Practice. Oxford:

Blackwell. Eckert, Penelope and Sally McConnell-Ginet. 1999. New generalisations

and explanations in language and gender research. Language in Society 28, 185-201.

Garrett, Peter, Nikolas Coupland and Angie Williams. 2003. Investigating Language Attitudes. Social meanings of dialect, ethnicity and performance.Cardiff: University of Wales Press.

Garrett, Peter, Nikolas Coupland and Angie Williams. 2004. Adolescents’ lexical repertoires of peer evaluation: boring prats and English snobs. In Adam Jaworski, Nikolas Coupland and Dariusz Galisinski (eds) Metalanguage. Social and Ideological Perspectives. Berlin: Mouton de Gruyter, 193-225.

Giles, Howard. 1970. Evaluative reactions to accents. Educational Review 22, 211-227.

Giles, Howard and Peter F. Powesland. 1975. Speech Style and Social Evaluation. London: Academic Press.

Giles, Howard and Nikolas Coupland. 1991. Language: Contexts and Consequences. Milton Keynes: Open University Press.

Gregersen, Frans and Inge Lise Pedersen (eds). 1991. The Copenhagen Study in Urban Sociolinguistics. Copenhagen: C.A. Reitzel.

Grønnum, Nina. 2001. Fonetik og Fonologi. Almen og Dansk (2nd edition). Copenhagen: Akademisk Forlag.

Hansen, Gert Foget and Nicolai Pharao. In press. Prosodic aspects of the Copenhagen Multiethnolect. In Gösta Bruce and Merle Horne (eds) Proceedings of Nordic Prosody 9. Lund: University of Lund.

Hogg, Michael A. and Dominic Abrams. 1988. Social Identifications. A social psychology of intergroup relations and group processes. London: Routledge.

Keim, Inken. 2004. Linguistic variation and communication practices in migrant children and youth groups. Presentation on the conference: Languaging and Language Practices: Multilingual Turkish Speakers in North Western Europe. April 29-30, University of Copenhagen.

Kristiansen, Tore. 1991. Sproglige normidealer på Næstvedegnen. Kvantitative sprogholdningsstudier. Ph.D.-dissertation. University of Copenhagen.

Kristiansen, Tore. 1992. Har regionalsprog en fremtid i Danmark? Danske folkemål 34. Copenhagen: C.A. Reitzel, 203-247.

Kristiansen, Tore. 1999. Unge sprogholdninger i Næstved 89 og 98. Danske folkemål 41. Copenhagen: C. A. Reitzel, 139-162.

Kristansen, Tore. 2001. Two standards: one for the media and one for the school. Language Awareness 10/1, 9-24.

Kristansen, Tore, Tina Bruun Clausen and Merete Havgaard. 2002. Sprogholdninger hos unge i Nakskov. Danske Talesprog 3. Copenhagen: C.A. Reitzel, 17-70.

Labov, William. 2001. Principles of Linguistic Change – Social Factors. Oxford: Blackwell.

Ladegaard, Hans J. 2001. Popular Perceptions of Standard Language: Attitudes to ’Regional Standards’ in Denmark. Language Awareness 10/1, 25-40.

Maegaard, Marie. 2001. ’Jeg er da stolt af at jeg er sønderjyde – altså sådan forholdsvis’. Om sprogbrug og sprogholdninger hos sønderjyske unge. Danske Talesprog 2. Copenhagen: C.A. Reitzel, 77-166.

Nielsen, Bent Jul. 1998. Talesprogsvariationen i Århus – en sociolingvistisk redegørelse og en sammenligning med sproget i Odder. Danske folkemål 40. Copenhagen: C. A. Reitzel, 51-78.

Nielsen, Bent Jul and Magda Nyberg. 1992. Talesprogsvariationen i Odder kommune I. Lokalsprog og rigsmål i sociolingvistisk belysning. Danske folkemål 34. Copenhagen: C. A. Reitzel, 45-201.

Nielsen, Bent Jul and Karen Margrethe Pedersen.1991. Danske talesprog. Copenhagen: Gyldendal.

Nielsen, Niels Åge. 1959. De jyske dialekter. Copenhagen: Gyldendal. Pedersen, Inge Lise. 2003. Traditional dialects of Danish and the de-

dialectalization 1900-2000. International Journal of the Sociology of Language 159, 9-28.

Quist, Pia. 2000. Ny københavnsk multietnolekt. Om sprogbrug blandt unge i sprogligt og kulturelt heterogene miljøer. Danske Talesprog 1. Copenhagen: C.A. Reitzel, 143-207.

Ryan, Ellen B., Howard Giles and Miles Hewstone. 1988. The Measurement of Language Attitudes. In Ulrich Ammon, Norman Dittmar and Klaus J. Mattheier (eds) Sociolinguistics. An International Handbook of the Science of Language and Society. Berlin: Walter de Gruyter, 1068-1081.

Røyneland, Unn. 2001. Aldersavgrensning og ungdom som språkendringsagenter. In Gunnstein Akselberg (ed) Målbryting 5. Ungdom og språk. Bergen: University of Bergen, 85-112.

Strauss, Anselm and Juliet A. Corbin. 1998. Basics of Qualitative Research: Grounded Theory Procedures and Techniques (2nd edition). London: Sage.

Tajfel, Henri. 1981. Social Stereotypes and Social Groups. In John C. Turner and Howard Giles (eds) Intergroup Behaviour. Oxford: Blackwell, 144-167.

Tajfel, Henri and John C. Turner. 1986. The Social Identity Theory of Intergroup Behaviour. In Stephen Worchel and William G. Austin (eds) Psychology of Intergroup Relations. Chicago: Nelson-Hall, 7-24.

Trudgill, Peter. 1983. Sex and Covert Prestige. Linguistic Change in the Urban Dialect of Norwich. In Peter Trudgill On Dialect. Oxford: Blackwell. 169-185.

Trudgill, Peter. 1995. Sociolinguistics: An Introduction to Language and Society (3rd edition). London: Penguin Books.

Wells, John. 1982. English Accents. Cambridge: Cambridge University Press. Williams, Angie, Peter Garrett and Nikolas Coupland. 1996. Perceptual

dialectology, folklinguistics, and regional stereotypes: Teachers’ perceptions of variation in Welsh English. Multilingua 15/2, 171-199.

Williams, Angie, Peter Garrett and Rosalind Tennant. 2004. Seeing the Difference, Feeling the Difference: Emergent Adults’ Perceptions of Communication and “Good” Communication with Peers and Adolescents. In Sik Hung Ng, Christopher N. Candlin and Chi Yue Chiu (eds) Language Matters. Communication, Culture, and Identity. Hong Kong: City University of Hong Kong Press. 111-136.

Zahn, Christopher J. and Robert Hopper. 1985. Measuring Language Attitudes: The Speech Evaluation Instrument. Journal of Language and Social Psychology 4/2, 113-123.

About the author Marie Maegaard Department of Scandinavian Research University of Copenhagen

Njalsgade 136 DK – 2300 Copenhagen S Denmark e-mail: [email protected]