7KLVDUWLFOHZDVGRZQORDGHGE\>'())@ 2Q'HFHPEHU $FFHVVGHWDLOV$FFHVV'HWDLOV>VXEVFULSWLRQQXPEHU@ 3XEOLVKHU5RXWOHGJH ,QIRUPD/WG5HJLVWHUHGLQ(QJODQGDQG:DOHV5HJLVWHUHG1XPEHU5HJLVWHUHGRIILFH0RUWLPHU+RXVH 0RUWLPHU6WUHHW/RQGRQ:7-+8.

$FWD/LQJXLVWLFD+DIQLHQVLD 3XEOLFDWLRQGHWDLOVLQFOXGLQJLQVWUXFWLRQVIRUDXWKRUVDQGVXEVFULSWLRQLQIRUPDWLRQ KWWSZZZLQIRUPDZRUOGFRPVPSSWLWOHaFRQWHQW W

7KHPDFUROHYHOVRFLDOPHDQLQJVRIODWHPRGHUQ'DQLVKDFFHQWV 7RUH.ULVWLDQVHQD D7KH/$1&+$57&HQWUH8QLYHUVLW\RI&RSHQKDJHQ

2QOLQHSXEOLFDWLRQGDWH'HFHPEHU

7RFLWHWKLV$UWLFOH.ULVWLDQVHQ7RUH   7KHPDFUROHYHOVRFLDOPHDQLQJVRIODWHPRGHUQ'DQLVKDFFHQWV $FWD ٢LQJXLVWLFD+DIQLHQVLD/ 7ROLQNWRWKLV$UWLFOH'2, 85/KWWSG[GRLRUJ

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. Acta Linguistica Hafniensia Vol. 41, November 2009, 167–192

The macro-level social meanings of late-modern Danish accents Tore Kristiansen*

The LANCHART Centre, University of (Received 31 July 2008; final version received 8 September 2008)

The paper reports on the main findings of the LANCHART language attitudes studies. These studies were designed to falsify (or modify) the picture of adolescent language ideology – and its role in language change – that had emerged from previous sociolinguistic studies in . This picture is formulated as three hypotheses: (1) There are two value systems at two levels of consciousness, (2) Language change is governed by subconscious values, (3) Copenhagen is Denmark’s only linguistic norm centre. Following strict guidelines for data collection among 9th graders (aged 15–16) in Copenhagen, Næstved, Vissenbjerg, Odder, and Vinderup we obtained subconsciously offered attitudes that could be compared with consciously offered attitudes. The results neither falsify nor modify the established picture but strongly confirm it. Keywords: language attitudes; social meaning of variation, conscious and subconscious attitudes; Danish speech community

1. Introduction Denmark entered the 20th century as a ‘normally’ diverse European dialect society and left it, arguably, as Europe’s most homogenized speech community. The role of language ideology in this radical change is a main concern of the Downloaded By: [DEFF] At: 13:28 13 December 2009 LANCHART centre and its endeavours to describe how and explain why the metamorphosis happened. The centre deals with ‘language attitudes’ in a many- faceted way, from both macro and micro perspectives, involving both quantitative and qualitative approaches to data gathering and analysis. This paper reports on the macro-level, quantitative, studies. The LANCHART centre studies what its name stands for: language change in real time. This is done by replicating a series of sociolinguistic studies previously carried out in the communities of (listed from east to west; see map on p. 7) Copenhagen, Køge, Næstved, Vissenbjerg, Odder and Vinderup. However, as language attitudes were not a major object of independent empirical investigations in any of these communities – with the exception of Næstved – what replication in this connection means is actually application of (the basics of) the Næstved design to all LANCHART communities (except for Køge where the general research focus is different, see Jørgensen and Møller this volume).

*Email: [email protected]

ISSN 0374-0463 print/ISSN 1949-0763 online q 2009 Taylor & Francis DOI: 10.1080/03740460903364219 http://www.informaworld.com 168 T. Kristiansen Næstved had its ‘attitudinal temperature’ measured twice before (Kristiansen 1991, 1999) and this allows for real time comparisons. A time depth perspective to our LANCHART findings is also obtained by referring to results from language attitudes studies accomplished in other Danish communities where (more or less modified versions of) the Næstved design was applied.

2. Background Thus, the design of the LANCHART language attitudes studies builds on experiences and evidence from previous research in the domains of both language ideology and language use. In this first section, we outline the Danish language situation as it emerges from this research with regard to use (2.1), ideology (2.2), and the role played by Copenhagen (2.3).

2.1 Use: The CONSERVATIVE, MODERN and LOCAL accents of late modern Danish As the rising generations abandoned the traditional dialects, in a kind of mass- movement during the 1960s and 1970s, they apparently looked to no other place than Copenhagen for new ways of speaking. The result is that the Danish language of today is more homogeneous than maybe any other language with millions of speakers. There is very little variation in grammar, and close to all existing variation in segmental phonology develops within and spreads from Copenhagen speech. To the extent that ‘local’ speech (outside of Copenhagen, that is) differs from Copenhagen speech, this is mainly in virtue of some suprasegmental (prosodic) colouring. We shall refer to such ‘locally-accented’ speech as LOCAL. The segmental variation within Copenhagen speech used to be associated

Downloaded By: [DEFF] At: 13:28 13 December 2009 with high versus low social status, both in the Danish population in general and in linguistic descriptions (Brink and Lund 1975, Jørgensen 1980, Gregersen and Pedersen 1991). Accordingly, the two Copenhagen speech varieties, commonly known as rigsdansk and københavnsk, have been referred to in sociolinguistics as HIGH COPENHAGEN and LOW COPENHAGEN (or just H- and L-COPENHAGEN). Studies have shown that more innovation originates in and spreads from L-speech than from H-speech; furthermore, as Copenhagen speech spreads throughout the entire country, its segmental phonological variation has been shown to spread with it (Brink and Lund 1975, Nielsen and Nyberg 1993, Jørgensen and Kristensen 1994). (For further descriptions of the Danish language situation at the level of use, see Pedersen 2003, Kristensen 2003. Note that in this paper we leave out of consideration the ways of speaking Danish that are found among adolescents in several multiethnic urban communities, Quist 2000, 2005). Since many LOW variants appear to be more ‘vital’ than their HIGH counterparts – in the sense of being more and more generally used in the younger generations – it may nowadays be more appropriate to leave the status- related terms behind and instead refer to this distinction in terms of an ‘older’ Acta Linguistica Hafniensia 169 version of Copenhagen speech versus a ‘younger’ version. These are relative terms of course. As new ways of speaking appear continually, the involved social meanings, including representations/evaluations like ‘older’ and ‘younger’, are likely to change accordingly: What is ‘modern and in’ today, may be ‘outdated and out’ tomorrow. While keeping this in mind we will deal with this difference – at the level of use – in terms of CONSERVATIVE versus MODERN.

2.2 Ideology: Two value systems at two levels of consciousness The 1960s and 1970s were pivotal decades for the relationship between Copenhagen speech and the traditional local dialects. Not only did the power balance between varieties in terms of numbers of speakers change radically in those years (the traditional dialects turned moribund as they were no longer transferred to the new generations, and the speech that took over was increasingly characterized by LOW features) but also the attitudinal situation changed in important ways. In Labovian sociolinguistics, theorizing about where language change comes from includes a distinction between ‘above’ and ‘below’ consciousness (Labov 1972). People are conscious of some variation in language use and unconscious of some other variation. If speakers change their usage as a conscious choice, the change is said to come ‘from above’. If speakers change their usage with no awareness of the existing variation and the choice options it offers, the change is said to come ‘from below’. However, if one believes, as we do, that language variation and change is driven by ‘social meaning making’, the distinction between ‘above’ and ‘below’ may be better conceived of as a distinction between two layers of consciousness. For speech differences to ‘make social meaning’, they have to be recognized and evaluated, and we prefer to see those processes as processes of consciousness, whether speakers are aware of them – or not aware of them. In the latter case, we shall say that the ‘social meaning making’ happens subconsciously.

Downloaded By: [DEFF] At: 13:28 13 December 2009 Now, the more interesting issue is whether ‘social meaning making’ is one and the same ‘thing’ at the two levels of consciousness? Or, is it rather the case that speech-related value systems differ depending on the consciousness condition; do people hold one set of overt values when language attitudes are consciously ‘performed’, and quite another set of covert values when the ‘performance’ is offered subconsciously? If so, what are the implications for our interpretation and understanding of variation and change in language use?

2.2.1 Overt ideology Prior to the1960s, the overt ideology of Danish society concerning non-standard ways of speaking appeared in elite discourse without any embellishing reticence: It seems to have been totally unproblematic to treat the difference between the standard and non-standard in moral and aesthetic terms like ‘good’ and ‘proper’ versus ‘bad’ and ‘sloppy’ (Kristiansen 1990). No doubt, this kind of elite discourse was essential to the construction, from the 17th century on, of the (idea of a) standard spoken language. However, it had very little impact on peoples’ 170 T. Kristiansen speech in general. Hence, we are entitled to assume (if we suspect that ideology plays a role in language variation and change in the first place) that non-standard speech – both traditional LOCAL DIALECTS and traditional LOW COPENHAGEN – could resist centuries of derogatory attacks from standard ideology (Milroy and Milroy 1985) because they were protected by covert positivity. Beginning with the 1960s and the introduction of sociological viewpoints into the humanities, the standard ideology was reformulated. It could no longer be built on a blatant dichotomy between ‘good’ and ‘bad’ language in the traditional aesthetic and moral sense. Instead, it was construed around a new dichotomy between contextually defined ‘appropriate’ and ‘inappropriate’ language use – the explicitly stated assumption being that only rigsdansk is appropriate in public contexts. However, the loss of the local dialects was now deplored. Certainly, they could still not pretend to any right of admittance in the public domain of society, but they were to be held in general respect – according, for instance, to the governmental guidelines for Danish mother-tongue education – and to be treated with affection and love in the private spheres of life. In contrast, many of the phonetic characteristics of LOW/MODERN have remained the overt target of elite discourse complaints about bad language (Kristiansen 1990). This overt hierarchization of speech varieties appears not only in elite discourse but is also accurately reproduced in carefully collected research data. When ‘speaker evaluation experiments’ (more about such experiments in section 5.1 below) are administered to subjects who are aware of their disclosing language attitudes, Danes in general – i.e. regardless of age, or other background factors – reproduce an evaluative pattern which fit in with representations of ‘appropriate’ language in private versus public contexts of use: CONSERVATIVE speakers are assessed more positively than LOCAL speakers, who in turn are assessed more positively than MODERN speakers. Although CONSERVATIVE often Downloaded By: [DEFF] At: 13:28 13 December 2009 beats LOCAL on all measurements (usually scales measuring ‘personality traits’), and most certainly on traits relating to ‘personal competence’ and ‘social status’, LOCAL may beat CONSERVATIVE on traits relating to ‘personal integrity’ and ‘social solidarity’, in particular if the CONSERVATIVE speakers to be judged are markedly ‘posh’ (Ladegaard 1992, Pedersen 1986). In contrast, MODERN can always be sure of keeping its bottom position. The same is true of evaluative hierarchies obtained in ‘label ranking tasks’ (more about these tasks in section 5.2 below): Outside of Copenhagen, københavnsk is always downgraded relatively to the local dialect name and rigsdansk. Moreover, in such ranking tasks, adolescents (and only adolescents) express ‘local patriotism’ by ranking the local dialect name in first position, above rigsdansk (Kristiansen 1991, Kristiansen, Clausen and Havgaard 2002, Clausen 2003, Andersen 2004). Thus, the situation as to language use and conscious attitudes seems to be the following: LOCAL DIALECTS fade away and die wrapped in relative positivity, while MODERN spreads and takes over surrounded by relative negativity. This situation has been repeatedly commented on as a bewildering paradox. Acta Linguistica Hafniensia 171 A straightforward and common way out of the paradox consists in concluding that attitudes do not play any role in language variation and change.

2.2.2 Covert ideology Another way out of the paradox consists in conceding that conscious attitudes apparently play no role in processes of variation and change, and then go on to investigate the possible existence and role of other – subconscious – attitudes. This issue was confronted and developed theoretically and methodologically in a series of studies by Kristiansen and colleagues in Næstved (Kristiansen 1991, 1999), Nakskov (Kristiansen, Clausen and Havgaard 2002), Tønder (Maegaard 2001), A˚ rhus (Maegaard 2005), (Clausen 2003), and Ærø (Andersen 2004). By conducting speaker evaluation experiments in a way that secures a ‘catch’ of subconscious attitudes, we have been able to establish the existence of another language-related value system: Speakers with even the faintest LOCAL accent are strongly downgraded on all evaluative traits in comparison with both CONSERVATIVE and MODERN speakers. Furthermore, the C/M-variation has been found to be involved in the construction of a distinction between ‘superiority’ and ‘dynamism’, two evaluative dimensions that cut across the traditional dimensions of ‘status’ and ‘solidarity’: While MODERN speakers do particularly well on ‘dynamic’ personality traits, CONSERVATIVE speakers defend their traditional top position in the evaluative hierarchy much better on ‘superiority’ traits. Most importantly, when we tap into covert ideology – and attitudes are subconsciously offered – the evaluative hierarchization of varieties is no longer in opposition to, but in accordance with, the changes in language use. In consequence, the impact of ideological forces on the relative strength of Downloaded By: [DEFF] At: 13:28 13 December 2009 Danish varieties no longer constitute a necessarily paradoxical situation – not if changes in use are governed by forces of covert ideology rather than by forces of overt ideology. (A more comprehensive presentation of the Danish language attitudes situation can be found in Kristiansen 2003a; other presentations in English focusing on the Næstved community in particular include Kristiansen 1997, 1998, 2003b, 2004, Kristiansen and Giles 1992, Kristiansen and Jørgensen 2005).

2.3 Copenhagen: Denmark’s only linguistic norm centre? As far as we know, the picture of Copenhagen as Denmark’s one and only linguistic norm centre is disturbed, at the level of use, only by a handful of innovative ‘western’ phonetic features which seem to spread independently of the Copenhagen impact. As of yet, the significance of this ‘disturbance’ is debatable. No convincing evidence is to be found in patterns of use (as established by relevant variationist studies) to the effect that these features spread from a 172 T. Kristiansen linguistic norm centre, with the city of A˚ rhus as the most likely candidate (for further discussion, see Kristiansen 2003c). Moreover, although a ‘speaker evaluation experiment’ that was conducted in A˚ rhus (Maegaard 2005) does not allow for conclusions about the spreading ‘western’ features as such, we may mention that the LOCAL A˚ rhus speakers in that study were downgraded relatively to the CONSERVATIVE and MODERN Copenhagen speakers, just like LOCAL speakers everywhere else. It may also be mentioned that the ‘western’ features in question are of a kind that makes it plausible that they are linked to the development of general literacy (their origin may be ‘reading pronunciations’), in which case their appearance and spread are likely to represent a subscription to ‘school’ and ‘superiority’ values rather than a manifestation of positive ‘local’ self-consciousness (Kristiansen 2003c). In brief, the nature of the spreading ‘western’ features is a matter for further investigations. In the meantime, we can safely sum up the general Danish situation, outside of Copenhagen, as follows: (i) the traditional local dialects have been replaced either by (more or less CONSERVATIVE vs. MODERN) Copenhagen speech or by LOCAL; (ii) in comparison with ‘genuine Copenhagen Speech’ (as produced by Copenhageners), ‘locally-accented Copenhagen speech’ (i.e. LOCAL) is subconsciously downgraded by local adolescents – notwithstanding the fact that they thereby, in most cases, downgrade their own language. So indeed, Copenhagen seems to be – judged from the situation at the levels of both use and attitudes – the only linguistic norm centre in Denmark.

3. Hypotheses At the macro level of ‘things’, the background description above allows us to deduce the following three hypotheses about important ideological aspects of the

Downloaded By: [DEFF] At: 13:28 13 December 2009 late modern Danish speech community: (1) There are two value systems at two levels of consciousness (2) Language change is governed by subconscious values (3) Copenhagen is Denmark’s only linguistic norm centre

4. Subjects The subjects selected for LANCHART’s macro-level quantitative attitudes studies were adolescents – limited to 9th grade students, who typically are 15 years old. The main reason for this was that we wanted to maintain the theoretical assumption of our previous language attitudes research in Denmark, namely that adolescence is a stage in life which deserves particular interest when the aim is to understand more about the role of language ideologies (representations and values) in processes of language variation and change. In addition, as 9th grade is the last class of compulsory education, we will be entitled to claim social representativity for our data. Acta Linguistica Hafniensia 173 5. Design The LANCHART language attitudes studies were designed to test the above- mentioned hypotheses. As already said, the design was basically the same as the one that had been developed and used in previous studies, thus allowing for broader comparisons in both space and time than between the LANCHART communities alone. As we were out to obtain both consciously and subconsciously offered attitudes, this objective determined the approach in many ways. In particular, it meant that the data collection procedure was of the uttermost importance (see further in section 5.3), and we will follow its order as this section takes us stepwise through the design in some detail. In Table 1, a dotted line is drawn across the second and third columns in order to indicate that the data collection session consisted of two experiments or tasks, first a ‘speaker evaluation experiment’ (5.1) and then a ‘label ranking task’ (5.2).

5.1 The speaker evaluation experiment The speaker evaluation experiment (SEE) elicited three kinds of answers to the same stimulus material – in terms of (1) personality traits, (2) standardness, and (3) geographical affiliation. The grey shading in Table 1 indicates that the first kind of answer was offered subconsciously by subjects, whereas the remaining answers were offered consciously, after the experimental condition had been changed from non-awareness to awareness.

5.1.1 Speaker evaluation in terms of personality traits If informants are to remain unaware of the purpose of the experiment, both (i) the stimulus material and (ii) the measuring instrument (response format) must be designed with great care. Downloaded By: [DEFF] At: 13:28 13 December 2009 (i) The stimulus voices The accent variation represented by the stimulus voices had to be ‘natural’ – in the sense of belonging to the every-day linguistic experiences of young people in the local community under study. There are two reasons for this requirement. The first reason derives from our fundamental research interest: We want to learn more about the role of language ideology in the processes of social meaning making which imbues social group and identity formation and, arguably, is the driving force behind variation and change. At the macro level of things, which is the perspective of these studies, we have every reason to believe that the language variation which is pertinent to social identification processes among young Danes in their local communities include CONSERVATIVE, MODERN and (their own) LOCAL. The second reason for delimiting the stimulus variation to these three accents is that only so can we hope to accomplish the SEE without subjects becoming aware of its purpose. The number of stimulus voices included in the LANCHART SEEs was twelve (eight in Copenhagen). We arrived at this rather high number for Downloaded By: [DEFF] At: 13:28 13 December 2009 174

Table 1. LANCHART language attitudes studies: Data collection design.

Condition Task Stimulus material and Response format non-awareness SPEAKER EVALUATION (i) stimulus material sound-recorded clips ¼ 12 speakers (1) personality traits [2 F and 2 C for each of the 3 accents C/M/L] – superiority each speaking for c. 30 seconds about ‘what a good teacher – dynamism (ii) response format 7-point adjective scales ¼ ma˚lrettet–sløv [goal-directed–dull] .Kristiansen T. til at stole pa˚ –ikke.. [trustworthy–untrustworthy] seriøs–ligeglad [conscientious–happy-go-lucky] spændende–kedelig [fascinating–boring] selvsikker–usikker [self-assured–insecure] klog–dum [intelligent–stupid] flink–usympatisk [nice–repulsive] tjekket–utjekket [cool–uncool] awareness (2) standardness stimulus material same as above – rigsdansk? response format ¼7-point [yes – no] scale (3) geographic affiliation stimulus material¼ same as above ¼ – Copenhagen? response format categorical choice [Copenhagen – nearby bigger town] ¼ ......

LABEL RANKING stimulus material list of ‘dialect names’ always including, among others, – like better? rigsdansk, københavnsk¼ ,[local dialect name] response format numbering ¼ Acta Linguistica Hafniensia 175 two reasons. Firstly, we found it important to adopt the ‘two-voices-per-accent’ approach that had been an unconditional requirement in previous studies of this kind in Denmark. Using two speakers to represent each accent has been our way of controlling for effects from differences in content and voice quality. Thus, we accept that results are accent-related evaluations only if the representatives of the accents are assessed similarly in a pair-wise way. As the number of relevant accents in our communities is assumed to be three – namely LOCAL, CONSERVATIVE and MODERN – the required number of stimulus voices thus far is six (four in Copenhagen). Secondly, as we suspected ‘the gender issue’ to be of importance to the assessments we want to study (following-up on Maegaard 2005), we needed to include both male and female speakers and thus double the six voices to a total of twelve (thereby increasing the danger of ‘overloading’ the data collection process quite considerably, as the subjects listened to the voices three times; more about that below). With the exception of the first accomplished studies (in Næstved and Tønder), all previous studies (in A˚ rhus, Nakskov, Odense, Bornholm, and Ærø) used the same four male stimulus voices to represent the CONSERVATIVE and MODERN accents, originally recorded and prepared for use in SEEs by Maegaard in 2001 in connection with her A˚ rhus study (Maegaard 2005). Maegaard’s CONSERVATIVE and MODERN voices, both genders, were also used in the LANCHART studies. In the Odder study, we also made use of her four A˚ rhus speakers as our LOCAL voices (see further about LOCAL voices below). In the other local communities, the LOCAL voices were found by following Maegaard’s procedure: Some 20–30 students at local gymnasiums, 16–17 years old, were audio-recoded in 3–5 minutes individual interviews about what’s a good teacher like? – a topic that these students easily responded to and talked about. From each of these local collections of student tales about ‘the good teacher’, we selected two boys and two girls who in our judgement had a local touch to their language. For each of Downloaded By: [DEFF] At: 13:28 13 December 2009 the four selected voices, clippings of speech were edited (in Cool Edit) so as to be quite similar as to length (c. 30 seconds), as well as content and fluency. In three of the non-Copenhagen communities, the LOCAL voices in the SEEs were from the nearby bigger town: Odense in the case of Vissenbjerg, A˚ rhus in the case of Odder, and in the case of Vinderup. The reason for this is that we want to investigate the role of these bigger towns as possible local linguistic norm centres. As Næstved is the only plausible pretender to the role as a linguistic norm centre in its region (Southern Sealand), the LOCAL voices in the Næstved SEE were Næstveders. The following point should be stressed: Our problem in the selection-of- voices process has never been to avoid local voices that might be suspected of being ‘too local’ (in the sense that they could direct judges’ attention to dialectal variation and thereby ruin the experimental aim of accessing a subconscious level of evaluation). On the contrary, our anxiety has always been whether even the most locally accented among the audio-recorded speakers were ‘local enough’ (to be subconsciously recognized as local speakers). This fact nicely 176 T. Kristiansen illustrates a most crucial point in our argumentation – namely the nation-wide homogeneity of adolescent Danish speech and the ensuing possibility of conducting speaker evaluation experiments with unaware judges. (ii) The adjective scales The measurement instrument had to take care not to ask questions that directed subjects’ attention to the evaluation task as a ‘dialect thing’. Our choice of evaluative items in terms of personality traits, as well as the particular adjective pairs we chose to represent these traits, was based on experiences and results from our previous research in Denmark, which has allowed us to collect subconscious attitudes from a large number of audiences. The stimulus speakers were assessed in all five LANCHART communities on the same eight personality traits. These were presented to subjects as ‘adjective scales’ with seven points. Table 1 lists the Danish adjective pairs, and gives English translations. The two adjectives forming a pair are assumed to make up a positive and a negative end, so to speak, of the personality trait they represent or construct (e.g. ‘intelligent’ vs. ‘stupid’), thus allowing for additions of results from individual scales, and, hence, for more general statements about assessments in terms of their ‘more or less’ on a few underlying ‘dimensions’ (‘superiority’ and ‘dynamism’ as suggested in Table 1 on the basis of previous research, see 1.2.2). In addition to the scales, an open-ended question – what is your immediate impression of this person? – urged subjects to add comments about each of the speakers in their own words.

5.1.2 Speaker evaluation in terms of standardness and geographical affiliation These two tasks were solved simultaneously. As indicated in Table 1, in this part of the SEE, subjects were aware of the real purpose of the tasks (more about the

Downloaded By: [DEFF] At: 13:28 13 December 2009 procedure in section 5.3 below). Subjects were asked to judge, while listening to the voices again, whether they spoke rigsdansk or not (on a 7-point scale for each voice), and whether the voice belonged to a person from Copenhagen or from the ‘local town’ – i.e. Næstved (in Næstved), Odense (in Vissenbjerg), A˚ rhus (in Odder), Holstebro (in Vinderup). The obtained data will allow us to shed light on our research questions in two ways. Firstly, the data will allow us to estimate whether the attitudinal patterns obtained on the personality scales can be ascribed to subconscious recognition/evaluation or not. If subjects (as a whole and on average) are not able to differentiate between locals and Copenhageners in a consciously performed recognition task – at a level of correctness that is significantly above random – there is little reason to believe that such recognition has been involved subconsciously in the evaluation task. Inversely, if the consciously offered recognition is significantly better than fifty-fifty, we may argue that recognition was also in play in the subconscious phase of the experiment. Secondly, we want to shed light on how language-related notions of ‘excellence’ (as measured in terms of personality traits), ‘rigsdanskhed’ Acta Linguistica Hafniensia 177 (‘standardness’), and ‘Copenhagenness’ are linked in young Danes’ represen- tations. The relationship between these notions is an always ongoing ‘negotiation’ in the Danish speech community. While it seems safe to claim that most learned people understand rigsdansk either as ‘a non-localizable way of speaking’ or as ‘the high status variety of Copenhagen speech’, lay people often seem to think of rigsdansk as either ‘their own language’ or ‘the speech of a nearby bigger town’ (Ladegaard 2001, 28). Results from previous SEEs conducted according to the design described here indicate that rigsdansk is associated with the way of speaking we call CONSERVATIVE, more than with MODERN and LOCAL. It is essential to shed light on how ‘dialect names’ and ‘ways of speaking’ are linked in our subjects’ representations if we want to argue that results from speaker evaluation and label ranking can be compared.

5.2 The label ranking task The label ranking task (LRT) is a simple way of collecting large amounts of consciously offered and readily quantifiable data that can be analysed into evaluative patterns for comparisons with the subconsciously offered SEE patterns. Subjects were given a list of some 7–10 ‘dialect names’ (labels) covering all of Denmark and were asked to compose a ‘dialect chart’ by giving number one to the dialect they like the best, number two to the dialects they like next best, and so on. The speech variation assumed to be relevant to social identification processes among adolescents in their local community was always included – in terms of københavnsk, rigsdansk, and the local dialect name. Our interest is to see how these three ‘ways of speaking’ are assessed relatively to each other, and to compare consciously and subconsciously offered patterns (LRT and SEE) – on the assumption that labels and accents correspond to each other as follows: [rigsdansk CONSERVATIVE][københavnsk MODERN][local dialect name

Downloaded By: [DEFF] At: 13:28 13 December 2009 $ $ $ LOCAL]. The additional dialects are of lesser interest and will not be dealt with. (They are not excluded from the results table, though, and thus their names and numbers as well as their relative evaluative scores can be studied in Table 2).

5.3 Data gathering procedure All local people (local governmental authorities, school administrations and teachers) who received information about our visit before it took place were carefully instructed not to reveal anything about its purpose. In the cases where two or more school classes were involved at the same school, the experiment was run simultaneously in all classes in order to avoid devastating rumours about our enterprise during breaks. All fieldworkers conducted the experiment in accordance with detailed written guidelines. To start with, the students were told that they were going to take part, anonymously, in an experiment, and that they would be told more about it and be given the opportunity to ask questions afterwards. The fieldworker distributed a booklet with one sheet for each voice containing the adjective scales 178 T. Kristiansen and explained what was going to take place by sticking closely to the text printed on the first page of the booklet (these booklets, both in Danish and English translation, can be found as appendices in Kristiansen and Monka 2006). There was a number written on the first page of each booklet, and the students were asked to remember this number and use it as their signature on a second questionnaire (in order for us to be able to relate consciously and subconsciously offered attitudes at the level of individuals). The booklets were produced with the adjective scales arranged in two different orders, and were distributed in such a way that persons sitting next to each other got different scale arrangements. The students were made aware of this and the fact that squinting at what the neighbours did was meaningless. The audio-recording with the stimulus speakers was played to the students a first time when they were told to just listen in order to get an overall impression of the gamut of speakers to be evaluated. Then, during the replay of the recording, the students ticked off their assessments while listening; a 15 seconds’ pause between each speaker allowed the slower students to finish off and the quicker ones to volunteer additional comments. Now the filled-in booklets were collected and the fieldworker asked the students for suggestions as to ‘what the experiment was about’. (This is ‘the moment of truth’. Did we succeed in collecting subconsciously offered data? – The truth is that no one ever came up with ‘attitudes towards dialects’). At this stage, the students were told that our interest is with how language is perceived and that the audio-recorded voices represented different ways of speaking Danish. A second booklet, with three pages, was distributed and signed by the students with their personal number. While listening to the voices a third time, the students gave their opinions about whether each of the voices was rigsdansk or not, and whether the person behind the voice was from [local town] Downloaded By: [DEFF] At: 13:28 13 December 2009 or Copenhagen. Finally, the students were asked to complete the two last pages, containing the label ranking task and requests for some personal background information. After collection of these booklets, the fieldworker explained the purpose of the experiment and initiated an audio-recorded discussion with the students about different perceptions and attitudes in connection with different ways of speaking Danish.

6. Results The results will be presented in the opposite order of the data collection order presented above: We start with the consciously offered hierarchizations in the LRT (6.1), continue with the SEE results concerning Copenhagenness (6.2) and standardness (6.3), and end with the subconsciously offered hierarchizations in terms of personality traits (6.4). Downloaded By: [DEFF] At: 13:28 13 December 2009

Table 2. Label ranking in five LANCHART communities. Copenhagen Næstved Vissenbjerg 1 københavnsk 1,57 sjællandsk 1,50 fynsk 2,09 2 sjællandsk 2,53 københavnsk 2,67 odenseansk 2,09 3 rigsdansk 3,28 rigsdansk 3,72 rigsdansk 3,54 4 fynsk 4,78 lol-falstersk 4,14 jysk 4,48 5a˚rhusiansk 5,12 fynsk 4,50 sjællandsk 5,00 6 jysk 5,13 jysk 5,39 københavnsk 5,02 7 bornholmsk 5,59 bornholmsk 6,02 bornholmsk 5,89 caLnusiaHafniensia Linguistica Acta p , 0,001 p , 0,001 p , 0,001 n 135, chi2 412 n 163, chi2 502 n 54, chi2 151 df¼ 6 ¼ df¼ 6 ¼ df¼ 6 ¼ ¼ ¼ ¼ Odder Vinderup

1 østjysk 2,26 midtjysk 3,00 2 a˚rhusiansk 2,53 vestjysk 3,52 3 rigsdansk 4,91 rigsdansk 4,86 4 københavnsk 5,41 nordjysk 5,01 5 nordjysk 5,57 a˚rhusiansk 5,56 6 vestjysk 5,86 østjysk 5,60 7 sjællandsk 5,95 sønderjysk 6,91 8 fynsk 6,73 fynsk 7,21 9 sønderjysk 7,09 sjællandsk 7,27 10 bornholmsk 8,73 københavnsk 7,63 11 bornholmsk 9,32 p , 0,001 p , 0,001 n 172, chi2 645 n 81, chi2 261 df¼ 9 ¼ df¼ 10 ¼

¼ ¼ 179 Figures are means on a 7-point scale (Copenhagen, Næstved, Vissenbjerg), a 10-point scale (Odder), and an 11-point scale (Vinderup). Significance test: Friedman 180 T. Kristiansen 6.1 Consciously offered hierarchizations of ‘dialect names’ Table 2 gives the ranking list for each of the five communities. In each case, we are particularly interested in the ranking of the local dialect name(s) (underlined) in comparison with rigsdansk and københavnsk (in bold). The local names (of regions and cities) appear in the two top positions in all five lists, followed by rigsdansk in third position. Thus, in conscious performance there clearly seems to be a strong local patriotism at work. Interestingly, however, the adolescents’ conscious support of local speech is no less regarding their ‘own’ local speech than regarding local ‘bigger city’ speech. In Vissenbjerg on Funen, fynsk does as well as odenseansk. In the Eastern Jutlandish community of Odder close to A˚ rhus, østjysk does better than a˚rhusiansk (p , 0,05), and the adolescents of further-away Vinderup clearly downgrade a˚rhusiansk relatively to the two local dialect names of midtjysk and vestjysk (Vinderup being situated in the border area between Middle and West Jutland). In other words: There is no sign in these data of a positive role for either Odense or A˚ rhus as linguistic norm centres in their regions. This issue is left more open in the case of Næstved, where sjællandsk does better than københavnsk (p , 0,001), as this town might in itself pretend to the role of linguistic norm centre in Southern Sealand. As to the relative ranking of rigsdansk and københavnsk, it generally works against københavnsk in the western part of Denmark (Funen and Jutland), and more markedly so in the less urban communities of Vissenbjerg and Vinderup than in the more urban community of Odder – a pattern which might be seen as an indication that the more urban adolescents have another ‘sense of’ københavnsk. On Sealand, københavnsk beats rigsdansk not only in Copenhagen but also in Næstved. As Næstved is the more urban of the non-Copenhagen communities, one might again see this relative upgrading of københavnsk as an

Downloaded By: [DEFF] At: 13:28 13 December 2009 indication of a possible representational link between københavnsk and modern big-city urbanity. It may be more likely, however, that the relative upgrading of københavnsk in Næstved is to be seen as an effect of the local patriotism that seems to be the general governing principle of the adolescents’ performances in the LRT: københavnsk comes out second in Næstved just as odenseansk comes out second in Vissenbjerg and a˚rhusiansk comes out second in Odder. In any case, these results confirm our previous findings (as described in 2.2.1): When consciously reporting their preferences, it holds generally true that Danish adolescents claim to like their local speech better than the alternatives offered by the larger society. We now proceed to try and establish whether ‘names’ correspond to ‘ways of speaking’ in adolescents’ representations in a way that will allow us to compare the consciously offered label rankings with the subconsciously offered speaker evaluations. In order to do this, we will look at how our differently C/M/L- accented stimulus speakers have been assessed in terms of geographical affiliation, and degree of standardness. Downloaded By: [DEFF] At: 13:28 13 December 2009

Table 3. Allocation of voices to Local city or Copenhagen. NÆSTVED (Local Næstved) Mg2¼ Mb5 Mg8 Lb9 Cb7 Cg4 Mb11 Cg10 Cb1 Lg12 Lg6 Lb3 Local 24,2 35,4 35,9 40,7 43,4 47,8 48,1 58,6 65,7 68,0 68,1 72,9 Cph. 75,8 64,6 64,1 59,3 56,6 52,2 51,9 41,4 34,3 32,0 31,9 27,1

VISSENBJERG (Local Odense) Hafniensia Linguistica Acta Mg8 Cg4¼ Cg10 Mb5 Mg2 Mb11 Cb1 Lg12 Cb7 Lb3 Lb9 Lg6 Local 22,2 35,2 44,4 46,3 48,1 51,9 55,6 55,6 57,4 63,0 74,1 75,9 Cph. 77,8 64,8 55,6 53,7 51,9 48,1 44,4 44,4 42,6 37,0 25,9 24,1 ODDER (Local A˚ rhus) Mg8¼ Mb5 Cg4 Mb11 Mg2 Cg10 Cb1 Cb7 Lg6 Lg12 Lb3 Lb9 Local 17,8 27,6 31,2 34,5 35,1 39,5 59,8 66,1 71,1 76,4 89,1 91,4 Cph. 82,2 72,4 68,8 65,5 64,9 60,5 40,2 33,9 28,9 23,6 10,9 8,6 VINDERUP (Local Holstebro) Mg8¼ Cg4 Mg2 Mb11 Cg10 Cb7 Mb5 Cb1 Lg12 Lb9 Lb3 Lg6 Local 20,0 28,2 30,6 41,5 42,9 42,9 47,1 60,0 72,6 81,2 83,5 91,8 Cph. 80,0 71,8 69,4 58,5 57,1 57,1 52,9 40,0 27,4 18,8 16,5 8,2

TOTAL n 496 (with 1–5 missing for each of the voices) ¼Mg8 Mg2 Mb5 Cg4 Mb11 Cg10 Cb7 Cb1 Lb9 Lg12 Lg6 Lb3 Local 25,3 31,7 35,8 37,2 42,6 47,7 52,8 61,5 69,1 70,4 74,1 79,4 Cph. 74,7 68,3 64,2 62,8 57,4 52,3 47,2 38,5 30,9 29,6 25,9 20,6

Voices are ranked according to increasing/decreasing percentage of ‘Local/Copenhagen’ C CONSERVATIVE (white), M MODERN (boxed), L LOCAL (shaded) b boy, g girl; numbers are the voices’ order of appearance on the stimulus tape ¼ ¼ ¼ ¼ ¼ 181 182 T. Kristiansen

6.2 Evaluation of differently C/M/L-accented speakers in terms of geographical affiliation Table 3 shows the percentages of subjects that have judged the voices to come from either Copenhagen or the Local potential norm centre. (In Copenhagen, the task was to propose a Copenhagen neighbourhood for each of the voices, without any given suggestions; these data are not discussed here). In all four communities, the LOCAL voices are recognized as locals by a majority of the informants. The only exception is Næstved Lb9 who is thought to be a Copenhagener by 6 out of 10 (detailed phonetic analyses by Nicolai Pharao suggest that the perceptually potential local features in this clip are limited to one single instance of a vigorously pronounced stød, which occurs at the end of the clip when most subjects may already have given their answer). As to the MODERN and CONSERVATIVE voices, it is true of all four communities that the percentage of ‘Copenhagen’ is higher, on average, for the MODERN voices than for the CONSERVATIVE voices. This pattern is made clear if we consider accents instead of voices, i.e. we add the four C-voices into the CONSERVATIVE accents, the four M-voices into the MODERN accent, and the four L- voices into the LOCAL accent. Table 4 gives the percentage of ‘Copenhagen’ attributed to each of the three accents in each of the four communities, and in total. In total, MODERN is judged to be ‘Copenhagen’ by two out of three, while CONSERVATIVE yields a 50–50 distribution of ‘Copenhagen’ and the name of the ‘Local bigger town’. LOCAL is judged to be ‘Copenhagen’ by one out of four; or put the other way round, three out of four have recognized the local colour of the LOCAL voices. On the basis of these results we feel entitled to postulate a fairly general representational link in Danish adolescents between MODERN and københavnsk, and between LOCAL and the local dialect name(s). If CONSERVA- TIVE shows a similar representational link to rigsdansk (we return to that issue

Downloaded By: [DEFF] At: 13:28 13 December 2009 in 6.3 below), its distribution here in terms of geographical affiliation may be said to sustain the claim that rigsdansk is perceived as a non-localizable accent, as 50–50 seems the likely result when subjects are asked to categorize non-localizable speech in geographical terms. There clearly is an Islands/Jutland distinction to be noticed as far as LOCAL is concerned. The ‘recognition’ of LOCAL speech is considerably better in Jutland than on the islands (Funen and Sealand). The community ranking in this regard is:

Table 4. Assessment of geographical affiliation at the level of ACCENTS.

MODERN CONSERVATIVE LOCAL Næstved 64,1 46,1 37,6 Vissenbjerg 57,9 51,9 32,9 Odder 71,3 50,9 18,0 Vinderup 65,2 56,5 17,7 Total 64,6 51,4 26,6

Figures are percentages that have answered ‘Copenhagen’ Downloaded By: [DEFF] At: 13:28 13 December 2009

Table 5. Evaluation of VOICES in terms of standardness. COPENHAGEN Cb7 Cg10 Cg4 Mb5 Cb1 Mb11 Mg2 Mg8 3,22 3,27 3,48 3,58 3,71 3,82 3,92 4,17 NÆSTVED Cb7 Cg4 Cb1 Cg10 Mb5 Lb9 Lg6 Mb11 Mg2 Mg8 Lg12 Lb3 2,93 3,19 3,28 3,29 3,65 3,65 3,69 3,90 3,92 4,03 4,04 4,14 VISSENBJERG Cb7 Cg4 Cg10 Cb1 Mb5 Mb11 Mg2 Lb9 Lb3 Lg6 Mg8 Lg12 3,46 3,48 3,56 3,67 3,96 4,00 4,09 4,11 4,43 4,44 4,48 4.69 Hafniensia Linguistica Acta ODDER Lb9 Cb7 Cb1 Cg10 Lb3 Cg4 Mb5 Lg6 Mb11 Mg2 Lg12 Mg8 2,84 2,96 3,29 3,34 3,47 3,49 3,64 3,76 3,87 3,92 3,95 4,07 VINDERUP Cb1 Cg10 Cb7 Cg4 Mb11 Mg2 Mb5 Lg6 Lg12 Lb9 Lb3 Mg8 3,08 3,12 3,14 3,40 3,46 3,50 3,52 3,54 3,56 3,64 3,82 3,90

TOTAL(1) Cb7 Cg10 Cb1 Cg4 Mb5 Mb11 Mg2 Mg8 3,08 3,30 3,38 3,38 3,65 3,82 3,88 4,09 TOTAL(2) Cb7 Cb1 Cg10 Cg4 Lb9 Mb5 Lg6 Mb11 Mg2 Lb3 Lg12 Mg8 3,04 3,29 3,31 3,37 3,41 3,66 3,77 3,83 3,87 3,88 4,00 4,07

The voices are ranked according to means on 7-point scales (lower values are higher standardness) C CONSERVATIVE (white), M MODERN (boxed), L LOCAL (shaded) b boy, g girl; numbers are the voices’ order of appearance on the stimulus tape¼ Friedman Test: Copenhagen p¼, 0,001 (n 132, chi¼2 50, df 7); Næstved¼ p , ¼0,001 (n 181, chi2 188, df 11); Vissenbjerg p , 0,001 (n 54, chi2 51, df 11); Odder p , 0,001 (n ¼ 173, chi2 ¼ 151, df¼ 11); Vinderup p , 0,001 (n¼ 84, chi2¼ 35, df ¼ 11); ¼ ¼ ¼ ¼ 2 TOTAL¼ (1) includes¼ Cph. and excludes LOCAL voices:¼ p ,¼0,001 (n¼ 627, chi 291, df 7)

2 183 TOTAL (2) excludes Cph. and includes LOCAL voices: p , 0,001 (n ¼ 492, chi ¼ 325, df ¼ 11) ¼ ¼ ¼ 184 T. Kristiansen Vinderup / Odder . Vissenbjerg . Næstved. In contrast, the Island/Jutland distinction seems to be of no importance to how MODERN and CONSERVATIVE are geographically categorized. A ranking of the communities with regard to ‘recognizing’ the MODERN speakers as Copenhageners would look like this: Odder . Vinderup / Næstved . Vissenbjerg – whereas a ranking with regard to ‘recognizing’ the CONSERVATIVE speakers as Copenhageners would look like this: Vinderup . Vissenbjerg / Odder . Næstved. It may in fact be more interesting to compare the percentage differences in the four communities between those who say ‘Copenhagen’ to CONSERVATIVE and MODERN, respectively. This difference is bigger in Næstved and Odder than in Vissenbjerg and Vinderup (18–20 percentage points against 4–9). This may be an indication that representations of geographical affiliation in connection with our CON- SERVATIVE/MODERN distinction reflect and construe an Urban/Rural distinction rather than an Islands/Jutland distinction.

6.3 Evaluation of differently C/M/L-accented speakers in terms of standardness In Table 5, the voices are ranked according to mean scores for standardness on a 7-point scale. The lower scores indicate the higher degree of assessed standardness. The general result is that the CONSERVATIVE voices are judged to be the more rigsdansk while the MODERN and LOCAL voices come out fairly even as less rigsdansk. This pattern appears clearly as the community results are added for each voice to a total result (1) and (2), including and excluding Copenhagen, respectively. This is also clear from the presentation in Table 6, where the voices are added to form the accents CONSERVATIVE, MODERN and LOCAL. Odder is an exception to the general picture in that the LOCAL accent is judged to be more standard than the MODERN accent. The Odder results in Table 5 show Downloaded By: [DEFF] At: 13:28 13 December 2009 that this upgrading is reserved for the LOCAL boys (Lb9 and Lb3). Odder Lb9, in

Table 6. Evaluation of ACCENTS in terms of standardness.

mean rank N chi2 sign

Copenhagen C *** M 1,31 1,69 132 22 *** Næstved C *** M/ L 1,48 2,26 2,26 181 81 *** Vissenbjerg C *** M/ L 1,46 2,12 2,42 54 28 *** Odder C ** L *** M 1,63 2,00 2,37 173 51 *** Vinderup C * L/ M1,76 2,11 2,13 84 8 *

Total(1) C *** M 1,27 1,73 627 145 *** incl. Cph Total(2)¼ C *** L/ M1,58 2,16 2,26 492 147 *** excl. Cph ¼ C CONSERVATIVE,M MODERN,L LOCAL; ranking based on mean rank values from ¼ ¼ ¼ Friedman Test; differences between tested with Wilcoxon Signed Pair Test. *** p , 0,001 ¼ ** p , 0,01 * p , 0,05 / n.s. ¼ ¼ ¼ Acta Linguistica Hafniensia 185 particular, does extremely well as he obtains the lowest score not only in Odder but in the whole sample. This is no big surprise, as the association of rigsdansk with A˚ rhus or the A˚ rhus region is a fairly well-known and often reproduced stereotype in general public discourse about where to search for rigsdansk (Ladegaard 2001, 35). What is important to us in this connection is the fact that CONSERVATIVE beyond any doubt is generally perceived of as more rigsdansk than MODERN and LOCAL. We may now proceed to compare consciously offered preferences in terms of rigsdansk, københavnsk, and local dialect names with subconsciously offered preferences in terms of CONSERVATIVE, MODERN, and LOCAL.

6.4 Subconsciously offered hierarchizations of differently C/M/L-accented speakers We will limit ourselves here to present results at the level of accents, which means that the scores for the four voices representing each accent have been added. Such additions make sense, of course, only if the four speakers have obtained similar scores. Indeed, for us to conclude to a decisive dialect features impact on valuations, the stimulus speakers should be assessed, by and large, in a quadruple-wise fashion. In order to give an impression of the kind of regularity that is involved, the result pattern at the level of voices is shown for the trait self- assured–uncertain in Table 7 (these analyses for each voice are reported in Kristiansen 2007). The perfect four-by-four pattern which is shown for Odder is rare, but by and large the voices group together in a way that makes it sensible to add them as four representatives of an accent. One voice does clearly not satisfy our loose four-by-four constraint, namely Næstved Lb9. In fact, the particularly positive

Downloaded By: [DEFF] At: 13:28 13 December 2009 treatment of Næstved Lb9 reappears on all eight scales, and our first speculation should of course be whether we here have found evidence that LOCAL speech may be seen as ‘the best way of speaking’ among Næstved adolescents. However, it is obviously pertinent to the interpretation of the personality traits result for Næstved Lb9 that this voice was the only LOCAL voice not to be recognized as a local by the majority of subjects (see Table 3). Six out of ten Næstved adolescents assessed Lb9 to be a Copenhagener rather than a Næstveder. Therefore, the Lb9 values are not included when the LOCAL Næstved voices are added (in stead, the values for the other LOCAL boy, Lb3, are included twice in the computation of the accent value, because we need to include four voice values for the computed L-value to be comparable with the computed M- and C-values). Table 8 shows the results for self-assured–uncertain at the level of accents. Now, Table 9 shows what this kind of pattern looks like on all eight personality traits. The patterns fall in two categories – arranged as the right and left halves of the table – and thus seem to reflect two underlying evaluative dimensions. LOCAL is downgraded relatively to CONSERVATIVE and MODERN on Downloaded By: [DEFF] At: 13:28 13 December 2009 186

Table 7. Evaluations of VOICES on the self-assured–uncertain scale. COPENHAGEN Mb11 Mg2 Cb7 Cg10 Mb5 Mg8 Cg4 Cb1 2,52 2,65 2,68 3,02 3,15 3,37 3,76 3,86 NÆSTVED Lb9 Mg2 Mb5 Cb7 Mb11 Mg8 Cg4 Lg6 Cg10 Cb1 Lb3 Lg12

2,13 2,55 2,56 2,65 2,91 2.93 3,46 3,55 3,61 3,69 4,00 4,02 Kristiansen T. VISSENBJERG Mb11 Mg2 Mb5 Cb7 Mg8 Cg4 Lb9 Cg10 Lb3 Cb1 Lg6 Lg12 2,54 2,88 2,94 3,02 3,15 3,25 3,29 3,38 3,58 3,69 4,35 4,46 ODDER Mb5 Mb11 Mg2 Mg8 Cb7 Cg10 Cg4 Cb1 Lb9 Lg6 Lb3 Lg12 2,27 2,49 2,56 2,84 2,87 2,93 3,01 3,18 3,40 3,41 4,12 4,51 VINDERUP Mb11 Mg2 Mb5 Cb7 Cg10 Mg8 Cg4 Lg12 Lg6 Lb9 Cb1 Lb3 2,35 2,72 2,81 2,86 2,89 2,99 3,19 3,53 3,65 3,93 3,94 4,42

The voices are ranked according to means on 7-point scales. C CONSERVATIVE (white), M MODERN (boxed), L LOCAL (shaded) b boy, g girl; numbers are the voices’ order of appearance on the stimulus tape¼ Friedman Test: Copenhagen p¼, 0,001 (n 127, chi¼2 114, df 7); Næstved¼ p ,¼0,001 (n 170, chi2 380, df 11); Vissenbjerg p , 0,001 (n 52, chi2 77, df 11); Odder p , 0,001 (n ¼162, chi2 ¼358, df ¼11); Vinderup p , 0,001 (n ¼ 81, chi2 ¼157, df ¼11). ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ Acta Linguistica Hafniensia 187

Table 8. Evaluations of ACCENTS on the self-assured–uncertain scale.

mean rank N chi2 sign

Copenhagen M *** C 1,29 1,71 127 26 *** Næstved M *** C *** L 1,40 2,04 2,56 170 120 *** Vissenbjerg M *** C *** L 1,47 2,01 2,52 52 31 *** Odder M *** C *** L 1,37 1,87 2,76 162 169 *** Vinderup M *** C *** L 1,46 1,96 2,57 81 56 ***

C CONSERVATIVE,M MODERN,L LOCAL; ranking based on mean rank values from Friedman¼ Test; differences¼ between tested¼ with Wilcoxon Signed Pair Test *** p , 0,001 ¼ all traits. The two dimensions result from a different treatment of the two Copenhagen accents: MODERN clearly beats CONSERVATIVE on what we have called dynamism values (‘self-assured’, ‘fascinating’, ‘cool’, ‘nice’), whereas CONSERVATIVE does as well or even better on what we have called superiority values (‘intelligent’, ‘conscientious’, ‘goal-directed’, ‘trustworthy’). These results accord perfectly with our previous findings (as described in 2.2.2).

7. Conclusion and discussion A main research interest of the LANCHART centre is with the role of social meaning making in processes of language variation and change. Such social meaning making will always relate, somehow, to ideological structures that dominate at the macro-level of society and organize different ways of speaking into hierarchies of ‘good and bad’ language. It is LANCHART policy to study language ideology empirically – by collecting and analysing language-use independent data and not just infer social meanings from patterns of use. We are Downloaded By: [DEFF] At: 13:28 13 December 2009 going to conclude and discuss the main results of the macro-level studies in terms of the three hypotheses we deduced from previous research on language ideology in Denmark (see section 3).

7.1 There are two value systems at two levels of consciousness The LANCHART results firmly establish that young Danes operate with two systems for valuation of language differences. The one system is put to use in contexts where the valuation is a conscious act; the principle behind the valuation seems to be a strong ‘local patriotism’ that opposes ‘Copenhagen’ and yields the hierarchy local dialect name . rigsdansk . københavnsk. Modifications to this hierarchy apply for Copenhagen and Næstved (see Table 2), but these do in fact follow from the principle of ‘local patriotism’. The other system applies in contexts where the valuation is subconsciously offered; it yields the hierarchy MODERN . CONSERVATIVE . LOCAL when the evaluative perspective is what we have called ‘dynamism’, and it yields the hierarchy CONSERVATIVE (.) Downloaded By: [DEFF] At: 13:28 13 December 2009 188

Table 9. Evaluations of ACCENTS on eight scales reflecting two dimensions.

SUPERIORITY DYNAMISM Intelligent Self-assured Copenhagen C *** M Copenhagen M *** C Næstved C *** M *** L Næstved M *** C *** L Vissenbjerg C * M *** L Vissenbjerg M *** C *** L Odder C ** M *** L Odder M *** C *** L Vinderup C ** M *** L Vinderup M *** C *** L Conscientious Fascinating Copenhagen C M Copenhagen M C *** *** Kristiansen T. Næstved C / M *** L Næstved M *** C ** L Vissenbjerg C / M *** L Vissenbjerg M *** C / L Odder C * M *** L Odder M *** C *** L Vinderup C ** M *** L Vinderup M *** C ** L Goal-directed Cool Copenhagen C / M Copenhagen M *** C Næstved M / C *** L Næstved M *** C ** L Vissenbjerg M / C *** L Vissenbjerg M *** C / L Odder C / M *** L Odder M *** C *** L Vinderup C / M *** L Vinderup M *** C ** L Trustworthy Nice Copenhagen C / M Copenhagen M * C Næstved M / C *** L Næstved M * C *** L Vissenbjerg M / C *** L Vissenbjerg M / C * L Odder C / M *** L Odder M *** C *** L Vinderup C / M ** L Vinderup M / C / L

C CONSERVATIVE,M MODERN,L LOCAL Wilcoxon Signed Pair Test: *** p , 0,001 **p , 0,01 * p , 0,05 / n.s. ¼ ¼ ¼ ¼ ¼ ¼ Acta Linguistica Hafniensia 189

MODERN . LOCAL when the evaluative perspective is what we have called ‘superiority’. It is really justified to talk of systems here, in the sense that these same evaluative hierarchizations reappear again and again across space – and also across time when we compare with results from previous research. Having established the stability and consistency of the two hierarchizations, the question we want to ask, of course, is how these systems are produced and reproduced. The answer must lie in how the accents are treated in the public sphere. This is hardly a bold assertion as far as the consciously offered hierarchization is concerned, as it seems to simply reproduce the ideology of elite discourse and thus testify to the efficiency of the institutions which propagate this ideology, not least the educational system (see 2.2.1). As to the ‘superiority vs. dynamism’ distinction which has been found to generally underlie the subconsciously offered evaluations of late-modern Danish accents, we have argued elsewhere that it appears as a new public domain, the spoken media, is added to the established domains of school and business and develops, starting in the 1960s, into an increasingly important element of late modern every-day life. Young Danes are said to operate with two ‘standards’ for language use: One for the school, where excellence is perceived in terms of ‘superiority’; and one for the media, where excellence is perceived in terms of ‘dynamism’ (Kristiansen 2001). Recalling that sociolinguists in general have been reluctant to accept any role for the spoken media in language change (see e.g. Chambers 1998), this last assertion may be more controversial. However, it seems quite obvious to us that the spread of a subconsciously held value system to adolescent communities all over Denmark – a very rapid spread which turns the consciously reproduced system of official ideology upside down – can only be understood as a media effect. Downloaded By: [DEFF] At: 13:28 13 December 2009

7.2 Language change is governed by subconscious values To claim a media effect on language ideology may not be that controversial after all; it is rather when we now continue by saying that language change is governed by subconsciously held values – spread by the media – that our claim no doubt will appear more than audacious to some. What we have established in the case of Denmark, is that developments in language use accord very badly with consciously offered attitudes, but very well with subconsciously offered attitudes. Certainly, our unveiling of a value system which accords well with changes in use does not in itself settle the issue of precedence – did ideology change before use, or did the change in ideology follow from change in use? – but we see our findings as strongly suggestive of a determining role for ‘covert’ language ideology in language change, and we are eager to see whether this conclusion will be supported as forthcoming studies apply other approaches to the issue (the design of such a study is presented in Kristiansen and Jørgensen 2005). 190 T. Kristiansen 7.3 Copenhagen is the only linguistic norm centre in Denmark In order to falsify our picture of Denmark as a one-way homogenizing speech community, we have searched for possible linguistic norm centres other than Copenhagen. Parting from the assumption that attributions like ‘best language’ or ‘excellence in language’ are defining for a linguistic norm centre, and that such attributions will have to be offered subconsciously in order to be of interest, we designed a speaker evaluation experiment that allows us to compare subconscious evaluative reactions to differently C/M/L-accented speakers. The LOCAL-accented speakers to be judged were from bigger local towns with an assumed competitive potential in relation to Copenhagen speech. Although previous studies in our potential norm centres themselves had shown relative downgrading of the LOCAL voices by adolescents (for Næstved see Kristiansen 1991, 1999; for Odense see Clausen 2003; for A˚ rhus see Maegaard 2005, for Holstebro see Svenstrup 2008), we reasoned that the attitudinal situation might be different in a nearby smaller community. To take the most obvious case, it seems reasonable to think that young speech from the nearby second largest Danish city of A˚ rhus might have a model-to-be-followed status among youngsters in Odder. The decisive criterion for us to claim the existence of alternative norm centres is that local speakers (speaking LOCAL) are subconsciously upgraded (or at least not downgraded) relative to Copenhagen speakers (speaking CONSERVATIVE or MODERN) – all while being ‘recognized’ as locals. This result did not obtain for any of our local stimulus voices. At the level of consciousness that counts, we have not found the slightest trace of ‘local patriotism’.

Notes on contributor Tore Kristiansen, Associate Professor, lic. phil., The Danish National Research Foundation LANCHART Centre, University of Copenhagen, Department of Scandinavian Downloaded By: [DEFF] At: 13:28 13 December 2009 Research.

References Andersen, Lena Wienecke. 2004. Hvor sætter man krydset na˚r det er ’bondsk’? Ærøsk? En undersøgelse af unges sprogholdninger pa˚ Ærø. MA Examination paper. Copenhagen University. Brink, Lars and Jørn Lund. 1975. Dansk Rigsma˚l 1–2. Lydudviklingen siden 1840 med særligt henblik pa˚ sociolekterne i København. Copenhagen: Gyldendal. Chambers, J.K. 1998. “TV makes people sound the same”. In Laurie Bauer and Peter Trudgill (eds.). Language Myths. London: Penguin books, 123–131. Clausen, Tina Bruun. 2003. Sprogholdninger blandt unge i Odense. En undersøgelse af 154 niendeklasseelevers holdninger til sproglig variation. MA Thesis. Copenhagen University. Gregersen, Frans and Inge Lise Pedersen (eds.). 1991. The Copenhagen Study in Urban Sociolinguistics 1–2. Copenhagen: C. A. Reitzel. Jørgensen, Jens Normann. 1980. “Det flade a vil sejre. En undersøgelse pa˚ socioling- vistisk grundlag af visse københavnske sprogforhold”. SAML 7. Institut for anvendt og matematisk lingvistik, Copenhagen University, 67–124. Acta Linguistica Hafniensia 191

Jørgensen, Jens Normann, and Kjeld Kristensen. 1994. Moderne sjællandsk. Copenhagen: C. A. Reitzel. Kjeld Kristensen. 2003. “Standard Danish, Copenhagen sociolects, and regional varieties in the 1900s”. In Kristiansen, Tore and Jens Normann Jørgensen (eds.). 2003, 29–43. Kristiansen, Tore. 1990. Udtalenormering i skolen. Skitse af en ideologisk bastion. Copenhagen: Gyldendal. Kristiansen, Tore. 1991. Sproglige normidealer pa˚ Næstvedegnen. Kvantitative sprogholdningsundersøgelser. PhD dissertation. Copenhagen University. Kristiansen, Tore. 1997. “Language Attitudes in a Danish Cinema”. In Nikolas Coupland and Adam Jaworski (eds.). Sociolinguistics. A Reader and Coursebook. London: Macmillan Press LTD, 291–305. Kristiansen, Tore. 1998. “The Role of Standard Ideology in the Disappearance of the Traditional Danish Dialects”. Folia Linguistica XXXII: 1–2 (thematic issue: Dialect levelling and the standard varieties in Europe, ed. by Peter Auer), 115–129. Kristiansen, Tore. 1999. “Unge sprogholdninger i Næstved 89 og 98”. Danske Folkema˚l 41, 139–162. Kristiansen, Tore (ed.). 2001. Changing Representations of Standardness in Late Modernity: The Case of Denmark (special issue of Language Awareness 10/1). Kristiansen, Tore. 2001. “Two Standards: One for the Media and One for the School”. In Kristiansen, Tore (ed.). 2001: 9–24. Kristiansen, Tore. 2003a. “Language attitudes and language politics in Denmark”. In Kristiansen Tore and Jens Normann Jørgensen (eds.). 2003: 57–71. Kristiansen, Tore. 2003b. “The youth and the gatekeepers: reproduction and change in language norm and variation”. In Jannis K. Androutsopoulos and Alexandra Georgakopoulou (eds.). Discourse Constructions of Youth Identities. Amsterdam/ Philadelphia: Benjamins, 279–302. Kristiansen, Tore. 2003c. “Sproglig regionalisering i Danmark?” In Gunnstein Akselberg, Anne Marit Bødal, and Helge Sandøy (eds.). Nordisk dialektologi. Oslo: Novus, 115–149. Kristiansen, Tore. 2004. “Social meaning and norm-ideals for speech in a Danish community”. In Adam Jaworski, Nikolas Coupland, and Dariusz Galasin´ski (eds.). Metalanguage. Social and Ideological Perspectives. Berlin/New York: Mouton de Gruyter, 167–192. Downloaded By: [DEFF] At: 13:28 13 December 2009 Kristiansen, Tore. 2007. The LANCHART language attitudes studies: Results. [Report to IC II meeting Copenhagen June 4–6, 2007, part I]. http://dgcss.hum.dk/ aarsberetninger/rapporter/ Kristiansen, Tore and Howard Giles. 1992. “Compliance-gaining as a function of accent: public requests in varieties of Danish”. International Journal of Applied Linguistics 2(1): 17–35. Kristiansen, Tore and Jens Normann Jørgensen (eds.). 2003. The Sociolinguistics of Danish (thematic issue of International Journal of the Sociology of Language 159). Kristiansen, Tore and Jens Normann Jørgensen. 2005. “Subjective factors in dialect convergence and divergence”. In Peter Auer, Frans Hinskens, and Paul Kerswill (eds.). Dialect Change. Convergence and Divergence in European Languages. Cambridge: Cambridge University Press, 287–302. Kristiansen, Tore and Malene Monka. 2006. Language ideology in Danish adolescents: two value systems at two levels of consciousness [Report to IC I meeting Copenhagen May 29–31, 2006]. http://dgcss.hum.dk/aarsberetninger/rapporter/ Kristiansen, Tore, Tina Bruun Clausen, and Merete Havgaard. 2002. “Sprogholdninger hos unge i Nakskov”. Danske Talesprog 3: 17–70. Labov, William. 1972. Sociolinguistic Patterns. Philadelphia: Pennsylvania University Press. 192 T. Kristiansen

Ladegaard, Hans J. 1992. “Sprogholdninger i Danmark. En socialpsykologisk analyse af stereotype holdninger til sprog og sprogbrugere i fire danske lokaliteter”. Nordisk Psykologi 44(3): 173–189. Ladegaard, Hans J. 2001. “Popular Perceptions of Standard Language: Attitudes to ‘Regional Standards’ in Denmark”. In Kristiansen, Tore (ed.). 2001: 25–40. Maegaard, Marie. 2001. “‘Jeg er da stolt af at jeg er sønderjyde – altsa˚ sa˚dan forholdsvis!’ Om sprogbrug og sprogholdninger hos sønderjyske unge”. Danske Talesprog 2: 77–166. Maegaard, Marie. 2005. “Language attitudes, norm and gender. A presentation of the method and results from a language attitude study”. Acta Linguistica Hafniensia 37 (thematic issue: Subjective Processes in Language Variation and Change, ed. by Tore Kristiansen, Peter Garrett, and Nikolas Coupland), 55–80. Milroy, Lesley and James Milroy. 1985. Authority in Language. London: Routledge. Nielsen, Bent Jul and Magda Nyberg. 1993. “Talesprogsvariationen i Odder kommune II. Yngre og ældre rigsma˚lsformer i sociolingvistisk belysning”. Danske Folkema˚l 35. Copenhagen: C. A. Reitzel, 249–348. Pedersen, Inge Lise. 1986. “Bondsk og ligefrem eller moderne og byagtig. En undersøgelse af holdninger til dialekt og dialekttalende pa˚ Fyn”. Danske Folkema˚l 28: 11–40. Pedersen, Inge Lise. 2003. “Traditional dialects of Danish and the de-dialectalization 1900–2000”. In Kristiansen, Tore, and Jens Normann Jørgensen (eds.). 2003: 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. Quist, Pia. 2005. Stilistiske praksisser i storbyens heterogene skole – en etnografisk og sociolingvistisk undersøgelse af sproglig variation. PhD dissertation. Copenhagen University. Svenstrup, Christoph Hare. 2008. Hvorfor er københavneren en smartass og jyden en bonderøv? – en undersøgelse af sprogholdninger i 8./9. kl.og 1.g. i Holstebro. MA Thesis. Copenhagen University. Downloaded By: [DEFF] At: 13:28 13 December 2009