Acquiring attitudes towards varieties of Dutch: a quantitative perspective

52
Acquiring attitudes towards varieties of Dutch: a quantitative perspective (short title: Acquiring attitudes towards varieties of Dutch) Author 1: Gunther De Vogelaer (corresponding author) University of Münster Institute of Dutch Philology Alter Steinweg 6/7 48143 Münster Germany [email protected] Author 2: Jolien Toye Ghent University Department of Linguistics, Dutch section Blandijnberg 2 9000 Gent Belgium

Transcript of Acquiring attitudes towards varieties of Dutch: a quantitative perspective

Acquiring attitudes towards varieties of Dutch: a quantitative perspective

(short title: Acquiring attitudes towards varieties of Dutch)

Author 1:

Gunther De Vogelaer (corresponding author)

University of Münster

Institute of Dutch Philology

Alter Steinweg 6/7

48143 Münster

Germany

[email protected]

Author 2:

Jolien Toye

Ghent University

Department of Linguistics, Dutch section

Blandijnberg 2

9000 Gent

Belgium

Acquiring attitudes towards varieties of Dutch: a quantitative perspective

Gunther De Vogelaer & Jolien Toye

Abstract

Even though adolescence is well-known to be a key period for the acquisition of vernacular

varieties, there seems to be little research on how attitudes change during adolescence. In

addition, most sociolinguistic studies on adolescent language hardly discuss developmental

factors. This study tries to amend for these gaps in our knowledge, by investigating how

attitudes towards a number of varieties of Dutch change in Flemish children between 8 and 18

years old. Adolescence is shown to be a period in which attitudes further emerge and change

considerably. The youngest children in our sample do seem to recognize Standard Dutch as a

model for their own speech, and are thus competent to distinguish between different varieties

of Dutch, but they hardly attribute any non-linguistic significance to language variation. As

children grow older, they realize that there is a correlation between language variation and

societal prestige. In addition, they become more sensitive to the ‘covert prestige’ of,

especially, the local variety, which is increasingly evaluated as indexing integrity and as a

means towards social and/or in-group success. Significant parallels are revealed between

sociolinguistic and psychosocial development, including 11-12-year-olds’ tendency to think

in terms of ‘perceived popularity’ (Cillessen & Rose 2005), and the peak around the age of 16

in conventional and social-clique dominated reasoning about friendship (Turiel 1983, Horn

2003).

1. Introduction: acquiring attitudes as a part of sociolinguistic competence

If sociolinguistic variation, in Eckert’s (2012) terms, expresses “the full range of social

concerns in a given community”, one obvious question is how and when children and

adolescents acquire the social meaning of variants to their full complexity, since this indeed

requires them to understand the complex social relationships within the communities in which

they live. In more than five decades of sociolinguistic research, a tendency is observed that

ascribes ever more sophisticated sociolinguistic skills to younger children. For instance,

Labov’s (1964) study, which is often quoted as the first one dealing with the acquisition of

sociolinguistic competence, proposes a six-stage model, in which the acquisition of “basic

grammar” precedes the acquisition of the vernacular, stylistic variation is only mastered in

adolescence, and the “acquisition of the full range of spoken English” (Stage 6) is only

attained by “college educated persons with special interest in speech” (1964: 191-192).

Consequent studies, however, have demonstrated fine sociolinguistic competences even in

young children (for a summary, see Chambers 2003). For instance, Smith, Durham & Fortune

(2007) show that the 2-3-year-old children in their corpus of Scottish English are able to

replicate usage frequencies of sociolinguistic variants in a number of situations observed in

their caregivers’ speech. Chambers (2003: 174) even goes as far as claiming that “there are no

studies indicating a time gap between the acquisition of grammatical competence and the

development of sociolinguistic competence”. Nevertheless, throughout childhood and

adolescence important shifts may be observed in the degree to which standard vis-à-vis

vernacular varieties are used. For African American English, Van Hofwegen & Wolfram

(2010: 428-429) discuss an alleged vernacular-reduction trajectory in childhood and pre-

adolescence, which is believed to be pushed forward by the exposure to prescriptive school

norms (see also Stewart 1965, Dillard 1972, Craig & Washington 2006). Other authors, such

as Wolfram (1969) and Fasold (1972), see a correlation between adolescence and greater

vernacularity, in line with a stronger orientation towards the peer group. Trajectories need not

be uniform, though: Van Hofwegen & Wolfram’s (2010) own longitudinal data show

considerable variation across their test subjects as to how vernacularity evolves with age,

which is likely to correlate with children’s general sense of identity within their speech

communities (cf. Eckert’s 1989 investigations on the speech of ‘jocks’ vis-à-vis ‘burnouts’).

The developing linguistic usage seems to be mirrored by emerging attitudes towards the

varieties at play, which lie at the focus of this article. According to Rosenthal (1974), the 3-5-

year-olds in her study on a semi-rural area in the US can already tell apart Standard English

and Black English and identify Standard English as the prestige variety. Day (1980) claims

the same with regard to Hawaiian 5-7-year-olds’ attitudes towards Standard English and

Hawaiian Creole English. A number of studies report the importance of age in the

development of attitudes: Cremona & Bates (1977) document how Italian children grow a

preference for the prestige variety between the ages of 6 and 10. Giles, Harrison, Creber,

Smith & Freeman (1983) observe a shifting preference in British children: whereas they rate

Welsh English more positively than Received Pronunciation (RP) at the age of 7, at the age of

10 they have identified RP as the prestige variety and evaluate Welsh English as funny.

During childhood, children are also getting better in detecting unfamiliar accents of their

mother tongue (Floccia, Butler, Girard & Goslin 2009 on British English), and in recognizing

varieties that are not spoken in their town of residence, especially if these varieties are

socially salient (Kristiansen 2010 on Spanish). Kinzler and DeJesus (2013), who compared

American monolingual English speaking 5-6-year-olds with 9-10-year-olds’ perceptions of

northern and southern accents, attest some asymmetries in the preference of certain varieties

of the mother tongue: younger children from the north (Illinois) preferred the northern group

of accents whereas southern children (Tennessee) did not manifest any particular preference.

They also claim that the older children’s perceptions reflect the stereotypical attitudes

observed in adults: the North was indeed perceived as “intelligent” and the south as “nicer”.

The (admittedly, rather scarce) volume of attitudinal research available, apart from

being predominantly concerned with the Anglo-Saxon world, can be characterized as 1)

focusing on (young) children rather than adolescents; and 2) emphasizing social or socio-

psychological rather than developmental factors in explaining changing attitudes. With regard

to the focus on young children, it seems indeed as if there is little research that has tried to

reveal how attitudes change in adolescence, even though we have some ethnographic studies

documenting the social meaning of certain types of language usage for adolescents (e.g.,

Eckert 1989, Jaspers 2005). With respect to the second aspect, pro-standard attitudes are

typically explained as the effect of formal schooling, which is supposedly fostering an

awareness towards prestige varieties and might even trigger negative attitudes towards

vernaculars (e.g., Cremona & Bates 1977, Day 1982; cf. also Van Hofwegen & Wolfram

2010 and references cited there on language usage). Explanations for the increasing

vernacularity of adolescent speech typically highlight peer group influence.

While explanations in terms of schooling and peer group influence may of course be

valid, such explanations do willingly or unwillingly downplay the role of children’s social

development, i.e. changes in the mind of the child itself. In this respect research on attitudes

constitutes no exception to the general state-of-affairs in sociolinguistics. Indeed Blum-Kulka

(2004: 197) observes that, in addition to being “concerned with toddlers and preschoolers

rather than with school-age children”, “most sociolinguistic studies lack a developmental

agenda.”

This study tries to amend for these gaps in our sociolinguistic knowledge, by

investigating the sociolinguistic competence of children and adolescents between 8 and 18

years of age. It is concerned in particular with how attitudes towards language variation are

acquired in Flanders, the Dutch-speaking part of Belgium. In order to do so, it employs a

quantitative method to detect how four age groups of East-Flemish children and adolescents

evaluate the Belgian variety of Standard Dutch, their hometown variety, and a number of

varieties of colloquial Belgian Dutch (or Flemish), including Brabantic, a variety claimed to

function as a prestige dialect and/or ‘informal standard’ in Flanders. Since the research

context in Flanders is quite different from the one in most previous investigations on the

topic, section 2 addresses the language situation in Flanders. After a discussion of the method

that has been used in the investigation (section 3), section 4 provides a statistical analysis of

the data obtained, mainly by means of Principal Components Analyses (PCAs). Section 5

contains some additional discussion on the question as to what extent the results reflect the

ongoing social development of the children under investigation, and links the findings to more

general insights on linguistic and social development. Finally, in section 6 a number of

conclusions are drawn with regard to the acquisition of sociolinguistic variation and the

methodological practice to extrapolate sociolinguistic hypotheses on the basis of research in

particular age groups, and some issues are formulated that require further research.

2. Attitudes in Flanders: context

Being the Dutch-speaking part of Belgium, Flanders is typically described as a region with a

‘delayed’ endoglossic standardisation,1 as the region was long dominated by French-speaking

1 Or standaardiseringsachterstand, see Van Hoof & Jaspers (2012) for discussion. Grondelaers & van Hout

(2011), however, seem to imply that the northern Dutch standard is exoglossic to Flanders as well. In line with Auer (2005), whose terminological apparatus to describe linguistic repertoires we adopt, we do not follow Grondelaers & van Hout and consider Standard Dutch to be an endoglossic standard in Flanders, since it is genetically closely related to the vernacular varieties, and therefore there is and always has been a potential for intermediate varieties to emerge (see Van Hoof & Vandekerckhove 2013 on this latter point). Grondelaers

minorities. Even though recent research somewhat problematizes the picture of Flanders’

infelicitous language history (Vosters, Rutten & van der Wal 2010), the use of French as a

prestige language in the past has long gone at the expense of the development of an

endoglossic standard, especially in contrast to the Netherlands, where the standardisation of

Dutch had been progressing continuously since the 17th

century. In an attempt to replace

French by an equally prestigious endoglossic standard, the Flemish adopted northern Standard

Dutch as their standard language, thereby striving to establish linguistic unity with the

Netherlands. The ‘integrationist’ agenda was decided upon in the 19th

century, but was

actually only properly carried out in the second half of the 20th

century, to a considerable

extent by means of massive propagation of the northern Dutch standard in the media and

education. Since the 1980s, an increased tolerance towards Belgian traits in the standard

language is observed, however, which is generally interpreted as a sign that a proper Belgian

variety of Standard Dutch has developed, which mainly differs from northern Dutch in its

phonology and vocabulary. Despite the increased penetration of Standard Dutch in Flanders,

heterogeneity in Belgian Dutch remains fairly high, especially in informal spoken contexts. In

fact informal spoken language can typically be described as a mixture of standard and

regional variants, which most Flemings would label Flemish rather than Dutch.2 In terms of

Auer’s (2005) typology of speech repertoires, most regions in Flanders therefore qualify as

examples of diaglossia, with varying degrees of dialect levelling and loss (Auer 2005: 25,

& van Hout’s (2011) argument that the prestige variety was taken over from an external speech community, is, in our opinion, better captured by describing northern Standard Dutch as a (historically) exocentric variety. 2 Flemish goes under a variety of other names, such as Verkavelingsvlaams ‘allotment Flemish’, tussentaal

‘interlanguage’ or Soapvlaams ‘Soap Flemish’. Most of these names appear to be given in an attempt to discredit this way of speaking, typically by people explicitly subscribing to the integrationist agenda. The terminological diversity in speaking about colloquial Flemish seems to be becoming another episode in the centennial debate on how to label linguistic practices in Flanders, which reflects Flanders’ complex linguistic (and political) history (see Deprez 1999 and Absilis, Jaspers & Van Hoof 2012 for discussion). The main reason why we adopt the term Flemish (or Vlaams) is that it is commonly used by laymen to describe their language of colloquial speech, even if many of them would describe the formal standard in Flanders as Dutch (Nederlands).

Plevoets 2008: 81, Grondelaers & van Hout 2011: 207). A notable exception is West-

Flanders, where many speakers maintain a diglossic repertoire and dialects are preserved well.

From an acquisitional point of view, an important fact is that many parents raise their

children in standard-near varieties of Dutch rather than in local dialects, even when both

parents would use dialect with each other (see De Houwer 2003 and Van de Mieroop, Zenner

& Marzo, forthcoming for an illustration). Depending on the region, however, many children

acquire dialectal features later in life, especially during adolescence, and some effectively

become proficient dialect speakers (Rys 2007). In such regions, dialects appear to function as

in-group varieties among adolescents (see Rys 2007: 332-333, 355 for East Flemish; Jaspers

& Vandekerchove 2009: 4 and Vandekerckhove & Nobels for West-Flemish; Kuppens & De

Houwer 2006 for Antwerp).

Quite independently from the introduction of northern Standard Dutch, Flanders’

leading economic region, Brabant, has been influencing the rest of Flanders for centuries. The

status of Brabant within Flanders has been the source of much controversy: it has been

claimed that some Brabantic features are actually gaining the status of ‘informal standard’,

turning Brabant into a region where an autonomous Flemish (informal) standard would be

developing (Goossens 1970, 2000; Cajot 1998; Willemyns 2005; Taeldeman 2008). However,

while the gradual spread of some Brabantic features over Flanders cannot be denied,

empirically thorough analyses by Plevoets (2008) and Grondelaers & van Hout (2011) show

that the main tendency to be observed in the present-day linguistic landscape is an increased

use of and tolerance towards all regional accents in more formal contexts, and not just

Brabantic, leading Grondelaers, van Hout & Speelman (2011) to conclude that Flanders is

going through a process of destandardisation.

In traditional descriptions of the Flemish language situation, attitudes are captured in

basically two dimensions, viz. prestige and solidarity. In this conception, which is in line with

many classic examples of diglossia (Ryan 1979, cf. H- and L-variety in Ferguson’s 1959

terms), Standard Dutch is considered the prestige variety, and is as such associated with traits

like richness or intellect, whereas dialects are found to express solidarity among peers

(speakers are evaluated as, for instance, sympathetic or friendly). Impe & Speelman (2007),

however, who describe the attitudes of 15-20-year-olds but take their data to represent

attitudes among Flemish adults, detect more sophisticated attitudes which are described in

three dimensions, viz. status, social attractiveness and personal integrity, as did Lambert,

Franckel & Tucker (1966) in a study on Canadian French. While Standard Dutch seems to

function as a prestige variety, regional varieties of Flemish trigger differential evaluations: the

peripheral West-Flemish dialect, for instance, scores high on personal integrity (e.g., traits

like trustworthiness or friendliness) whereas Brabantic is evaluated as socially attractive (and

associated with popularity, a sense of humour,...). An even more complex picture is obtained

by Grondelaers, van Hout & Speelman (2011), who distinguish no less than 4 dimensions

(competence, dynamism, solidarity and integrity), and signal extensive regional variation and

in-group and out-group biases in attitudes. This regional variation likely affects not just the

evaluation of the varieties at hand, but the basic structure of attitudes as well: Ghyselen

(2009) and Ghyselen & De Vogelaer (2013), who zoom in on a town in the western periphery

of Flanders, find no evidence for 3 or 4 evaluative dimensions and report a two-dimensional

landscape with a prestige and a solidarity variety (Standard Dutch vs. local dialect), in which

non-local accents such as Brabantic are evaluated negatively in all respects.

3. Measuring attitudes: method

3.1. General considerations

Attitudes are traditionally defined as “disposition[s] to react favourably or unfavourably to a

class of objects” (Sarnoff 1970 as quoted by Garrett 2007: 116). Even though the relation

between attitudes and behaviour is far from straightforward, attitudes are essential in

explaining language users’ preferences to use certain varieties and variants rather than other

ones. There is a vast literature on attitudes as a theoretical concept, both in socio-

psychological science in general and in sociolinguistics in particular (see Kristiansen 2011 for

a recent overview). Contrary to earlier conceptions of attitudes as simple and stable, most

present-day research assumes attitudes to be multidimensional in nature (Garrett 2010: 225-

226), and dynamic, i.e. largely dependent on a specific discourse context (Garrett, Coupland

& Williams 1999: 322-324; Giles & Billings 2004: 200-201).

Regarding the multi-dimensionality of language attitudes, experimental studies on

language attitudes have a long tradition in implementing models distinguishing several

evaluative dimensions, including the models already described above, viz. the

prestige/solidarity-model (based on Brown & Gilman 1960) and the competence/personal

integrity/social attractiveness-model (Lambert, Frankel & Tucker 1966). Since Zahn &

Hopper (1985), it has become good practice to use factor analyses to group a large number of

differential scales into small numbers of dimensions, which reduces the risk of a priori

projecting evaluative dimensions on the population’s sociolinguistic reasoning. More

qualitative methods, such as ethnography and discourse analysis, have equally shown that

language variation yields complex attitudes. For instance, Jaspers’ (2011) analysis of

stylisation patterns in schools in the Belgian city of Antwerp shows that, in the eyes of

youngsters of non-Belgian descent, the local dialect is associated with values as diverse as

racism and assertiveness.

In dynamic conceptions common in present-day research, attitudes to a large extent

depend on discourse context. With respect to the acquisition of attitudes towards

sociolinguistic variation, studies on language production have pointed out divergent attitudes

in different socialization environments such as the family and the peer group, and have tried

to reveal the possible influences of both environments (e.g., Payne 1980, Kerswill & Williams

2000). In European contexts of dialect loss, but also elsewhere, families typically favour the

use of prestige varieties in the home (e.g., De Houwer 2003 and Van de Mieroop, Zenner &

Marzo, forthcoming on Belgian Dutch; Smith-Christmas 2014 in a Gaelic – English context

on the Isle of Skye (Scotland)), whereas peer groups are usually associated with non-standard

language usage (see above). Yet prestige, or other connotations associated with particular

varieties, may vary along with a particular individual or group identity that emerges in the

course of interaction. Ghimenton & Chevrot (2006) described how parent usage of Veneto’s

regional language increased as children in the family got older, indicating that the children’s

age may be a factor influencing the social norms set within the family (Van de Mieroop,

Zenner & Marzo, forthcoming detect a similar tendency in Belgian Dutch). Ghimenton (2013)

also found that the interactional setting played an important role in the choice of Veneto’s

regional language when speaking to the younger members of the family. For instance, in

dyadic contexts parents gave preference to Italian whereas in multiparty interactions, the child

benefitted from extensive exposure to the regional language.

To the analyst, attitudes are best considered mental constructs, which cannot be

accessed directly. Garrett (2007) distinguishes between three approaches in sociolinguistic

research to tap into attitudes, viz. the societal treatment approach (e.g., ethnographic studies),

the direct approach (e.g., interviews on aspects of language) and the indirect approach (e.g.,

speaker evaluation experiments). While adolescents’ language attitudes can and have been

investigated through an array of methods, this investigation employs the speaker evaluation

method (Lambert, Hodgson, Gardner & Fillenbaum 1960), by confronting informants with

stimulus fragments and asking them a number of questions. The speaker evaluation method is

not uncriticised, as it seems to assume a static view on attitudes and tends to rely on artificial

stimulus material presented to the informants in a decontextualized, experimental setting

(Garrett, Coupland & Williams 1999: 322-323; Giles & Billings 2004: 200-201). Soukup

(2013), however, argues that the multi-dimensionality of language attitudes distinguishes

them from attitudes as used in most other research in social psychology and in fact allows to

reconceptualize quantitative speaker evaluation studies as ‘discursive events’ suitable to

detect the ‘social meaning’ of linguistic entities. Of course, proper scrutiny is in order as to

the situational parameters that may influence the results, which can impossibly all be

controlled for in an experiment’s design. A speaker evaluation experiment also allows for a

large informant sample and, especially, a rigid comparison (Garrett 2007: 117-119), for

instance across languages (as exemplified by the papers in Kristiansen & Grondelaers 2013),

or across age groups (as done below). In addition, quantitative survey-based research has

revealed interesting variation in the structure of adult attitudes in Flanders, which provides us

with a good baseline to compare our results to. Below, the investigation’s results will be

compared to findings from studies employing other methods too, in line with Garrett’s (2007:

120) plea for methodological triangulation, which is believed to deepen our knowledge on

attitudes in comparison to studies in which just one method is used (cf. also Niedzielski &

Preston 2003).

3.2. A quantitative study in Kluisbergen

The investigation was carried out in Kluisbergen, a small town in the Belgian province of

East-Flanders.3 Dialectologically speaking, Kluisbergen lies in the transitional zone between

East- and West-Flemish dialects of Dutch, as does Maldegem, the location where Rys (2007)

carried out an extensive fieldwork study revealing a strong increase in dialect proficiency

between the ages of 9 and 15. Since changing usage patterns are likely to be paralleled with

changing attitudes, significant attitude changes can be expected in adolescents from this

region.

The evaluated fragments were five renditions of the first sentences of the Dutch

translation of The dragon with red eyes, a fairy tale by the Swedish writer Astrid Lindgren

(see the appendix). These fragments, all some 90 seconds long, were read-out in Standard

Dutch and four non-standard varieties from, respectively, Kluisbergen (the local variety),

West-Flanders (a neighbouring area), Ghent (the province capital) and Brabant (an

economically dominant area with an allegedly prestigious dialect). Fragments were translated

and read by native speakers of the relevant varieties, with the instruction to use the relevant

varieties as spoken by young people today, so no attempt was made to approach traditional

dialects in the sample. Nevertheless, all fragments contain sufficient features to be

recognisable as typical for the relevant location or region, both through the particular

constellation with which regional features are found, and through the use of more local

features. Below, one sentence of each fragment is given, in which a number of non-standard

features are labelled. Since it is not attempted to provide a full overview of non-standard

features in these sentences, there is insufficient information to localize all fragments

unambiguously, but the transcriptions do give an idea of how the fragments relate to each

other.

3 The dialects spoken in the provinces East- and West-Flanders are referred to as Flemish in dialectology; the

area coincides more or less with the medieval county of Flanders. Quite confusingly, this area only makes up the western half of what is nowadays called Flanders, which comprises the entire Dutch-speaking Belgium and includes the Belgian parts of medieval Brabant and the area nowadays referred to as Limburg.

Examples of non-standard features:

(1) Standard Dutch: daar lag het grote moedersvarken met tien kleine biggetjes

dicht tegen haar aan in het stro

there lied the big mother.pig with ten little piglets

close to her on in the straw

‘There was the big mother pig lying in the straw with ten little piglets

close to her’

(2) Brabantic: daar lag t grote moedervarke met tien kleine biggetjes dicht tegen haar

aan in t stro

> features: mildly deviating vowels, e.g. monophtong ei in klein, closed i

in biggetjes, dicht, in

(3) Kluisbergen: der lag da groot verken met tien kleine bigskies tegen aar in t stro

> features: deletion of final t in da(t), e-vowel in varken, ie-diminutive in

bigskies, procope of h in haar

(4) Ghent: daar lag t groot moedervarke met tien kleine biggetjes dicht tegen haar

aan in t stro

> features: consistent use of uvular R

(5) West-Flemish: doa lag t grote moedersvarken met tien kleine biggetjes dichte tegen eur

oan in t stroë

> features: final-e in dicht, diphtong in stro, procope of h in heur/haar

While the fragments from Ghent and Brabant are somewhat more standard-like than the

Kluisbergen and West-Flemish fragments, they both display a number of highly frequent,

salient features setting them apart from Standard Dutch (e.g., uvular R in Ghent, a

consistently closed realization of all front vowels in Brabant). The fragments from

Kluisbergen and West-Flanders show a number of very local traits, such as diminutives on -ie

in Kluisbergen or a consistent realization of g as h in West-Flanders, of which numerous

instances appear in the fragments. Since it proved virtually impossible to find speakers

competent in several non-standard varieties, all non-standard fragments were read by different

speakers; the Standard Dutch fragment was read by the same speaker as the West-Flemish

fragment. Speakers were all male and 20-25 years old: in that way it was attempted to target

attitudes towards contemporary varieties rather than language usage associated with older

generations, but at the same time speakers were avoided that belonged to the informants’ age

groups.

The investigation targeted four age groups, and was carried out in two primary and one

secondary school, in both cases in what is known as grade 3 and 6, yielding 4 age groups: 8-

10-year-olds (n=37), 11-12-year-olds (n=38), 13-14-year-olds (n=21) and 17-18-year-olds

(n=20).4 Group sizes are relatively small and not controlled for additional social factors,

partly for convenience sake (e.g., children are not always able to answer questions re. their

family’s social status), but also because the samples more or less amount to the total number

of potential informants in the relevant location for the particular age groups. While sitting in

class, the informants were asked to listen to the five recorded fragments and individually fill-

out a questionnaire, consisting of 14 questions per fragment, and, so they were told, meant to

evaluate whether the persons reading-out the fragments were good storytellers. Informants

were asked to rate the fragments on 13 five-point Likert scales, relating to

prestige/competence, solidarity and integrity of the speaker, and to judge where the speaker

4 The informant sample contained more or less equal proportions of male and female informants in all age

groups, viz. 22 males vs. 16 females in 8-10-year-olds, 20 males vs. 17 females in 11-12-year-olds, 9 males vs. 12 females in 13-14-year-olds, and 7 males vs. 13 females in 17-18-year-olds. The analysis revealed no gender effects.

would come from (see the appendix for a questionnaire). Questions were based on Impe &

Speelman (2007), but partly reformulated to make them suitable for the youngest informants

in the study, and supplemented with questions as to whether informants judged the speaker as

sounding beautiful, as a suitable TV-presenter, as intelligible, and as to whether the

informants would like to speak like the person in the fragment. The questionnaire results were

entered in an SPSS-database and processed statistically, mainly by means of principal

component analyses (PCAs).

4. Attitudes in East Flemish children and adolescents: results

4.1. General picture

A principal components analysis (PCA) on the overall data provides a rather unstructured

picture of the attitudes of our informants: results pattern along no less than 5 dimensions

yielding eigenvalues higher than 1, with a total explained variance of 49,568%. Table 1 shows

the component matrix of these 5 dimensions as obtained through Varimax-rotation. In line

with most similar research in social sciences, factor loadings higher than .06 are considered

high and are bold-faced in the table. Factor loadings between .04 and .06 can be considered

intermediate, factor loadings lower than .04 are low, and hence suppressed from the table. To

link the attributes in the table to concrete questions asked to the informants, see the appendix.

Table 1 contains only a restricted number of high factor loadings, and hence many attributes

loading high on none of the five dimensions, which testifies to the rather unstructured nature

of the data.5

5 The fact that some features in a table do not load high on either dimension, and others on several

dimensions, is often taken as a reason for eliminating them from the data set. This is not done here, both on

Table 1. Rotated component matrix of PCA on overall attitudes

Component

1 2 3 4 5

SMART ,903

HELPFUL ,717

UNDERSTANDABLE ,694

LEADER ,441

SPEAK.LIKE ,722

TRUSTWORTHY ,636

FUNNY ,503

FRIENDLY ,703

TV-PRESENTER ,664

POPULAR

WELL-PAID ,813

FRIEND ,409

BEAUTIFUL ,470

Labelling these 5 dimensions is far from a straightforward issue: since we are dealing with

child, teenage and adolescent informants, it cannot be assumed that they think in similar

categories as found in research in adults, and the structure of attitudes is likely to change with

age. Therefore, rather than using labels common in older research, dimensions will be given

proper names reflecting their composition. Dimension 1 can be termed ‘linguistic prestige’, as

it seems to capture the extent to which speakers are perceived as linguistic role models, but

correlates only to a limited extent with non-linguistic attributes (cf. 3 factor loadings higher

than ,600, viz. UNDERSTANDABLE, SPEAK.LIKE and TV-PRESENTER, and only intermediate

loadings for LEADER and FRIEND). Factor 2 only displays a high loading for the feature

FRIENDLY, hence it is termed ‘friendliness’ here, even though the intermediate factor loading

for FUNNY (and the marginally lower but suppressed loadings for POPULAR and FRIEND, with

,396 and ,398, respectively) suggest a similarity to dimensions such as Impe & Speelman’s

(2007) ‘social attractiveness’. Factor 3 resembles prestige-dimensions in other investigations

statistical and analytical grounds. Indeed communalities for all variables are neither exceptionally high nor low, so little is to be gained from discarding them from the analysis. In addition, keeping variables out of consideration would lead to discrepancies in the attributes included in the analyses for the different age groups, which renders comparison and explanation more difficult.

in its high loading for the feature WELL-PAID, but only a moderate correlation is observed with

accent euphony (BEAUTIFUL), and there seems to be no correlation with attributes relating to

linguistic behaviour. Therefore this factor can best be labelled ‘economic prestige’. Since

factor 4 only loads high for the factor SMART, it can be labelled ‘intelligence’. Factor 5, with

high loadings for HELPFUL and TRUSTWORTHY, very much resembles Impe & Speelman’s

(2007) ‘personal integrity’-dimension.

All scores were transformed into dimension scores by means of regression factor

analyses. ANOVAs show that not all dimensions yield significant differences in the way the

varieties are perceived: only for factor 1 (‘linguistic prestige’; F=21.428, p < .001), 2

(‘friendliness’; F=3.320, p < .05) and 4 (‘intelligence’; F=3.009, p < .05), such differences are

found. Mean factor scores for the dimensions yielding significant differences are shown in

Figure 1.

Figure 1. Mean REGR Factor score for factor 1 (‘linguistic prestige’; fig.1a), factor 2

(‘friendliness’; fig.1b) and factor 4 (‘intelligence’; fig. 1c), broken down per variety (Brab =

Brabantic, ST = Standard Dutch; Kluis = Kluisbergen accent; Ghent = Gent accent; WVL =

West Flemish accent)

Post-hoc Scheffé analyses show that all differences in linguistic prestige are significant (with

p < .01), apart from the ones between Standard Dutch and the Ghent accent, and between the

Kluisbergen and West-Flemish accents. The main effect contributing to the significant

differences for factor 2 is the difference between the Kluisbergen and Brabantic accents (p <

.05). For factor 4, post-hoc Scheffé analyses yield no significant differences, despite the

overall significant effect of the parameter ‘variety’ on this factor.

As said, the general picture of our informants’ attitudes appears somewhat unstructured:

more dimensions are needed than in similar investigations in adults, in which, for instance,

‘linguistic prestige’, ‘economic prestige’ and ‘intelligence’ tend to appear as one dimension,

and in which, in general, fewer attributes show low loadings across the board. This

unstructured picture is due to two reasons: 1. the presence of young informants in the sample,

whose linguistic attitudes indeed appear less-structured than attitudes of older informants; 2.

the fact that the overall results collapse figures for several age groups, which tend to be

heterogeneous. Indeed as our informants grow older, their attitudes can be described along

ever fewer dimensions (four in 11-12-year-olds; three in 14-15 and 17-18-year-olds), with

total variance explained slightly increasing. The total proportion of variance explained in all

age groups is slightly higher than for the overall results (Table 2).

Table 2. PCA-results per age group: number of dimensions and total variance explained.

In interpreting Table 2, however, it must be borne in mind that differences in the answers of

the age groups are ambiguous, in that they need not necessarily reflect sociolinguistic

development: in principle, a part of the unstructured nature of the youngest age group’s

attitudes may be due to the fact that they do not have the same cognitive abilities as older

informants or are simple less accustomed to filling out questionnaires. So rather than relying

on the mere listing of dimensions and the variance they explain, proper analysis is needed on

their composition, which does indeed show robust age-related differences that can only be

understood as consequences of social development (e.g., when a dimension loses certain

attributes, or the composition of a dimension shifts along lines predictable from literature on

psychosocial development). This is further illustrated below, by means of separate PCAs per

age group.

Apart from age, a second social variable in our informants is gender. Neither the overall

results, nor the results for the age groups show gender differences, however. Hence this

variable is left out of consideration in the remainder of this article.

4.2. Age group 8-10

The results of the youngest age group mirror the overall results quite closely, in that 5

dimensions can be distinguished with eigenvalues higher than 1, yielding an explained

variance of 50,09%. The 5 dimensions capture more or less the same features as in section

4.1, the main contrast with the overall results being that the fifth dimension only includes

TRUSTWORTHY, and not HELPFUL. So the dimensions could be labelled ‘linguistic prestige’,

‘friendliness’, ‘intelligence’, ‘economic prestige’, and ‘trustworthiness’, respectively.

Regarding the first dimension, note that above, the possibility was hinted at that age-related

differences in attitudes are to some extent due to different abilities in our age-groups to handle

questionnaires, leading to reliability issues especially in the results for the youngest

informants included in this study. Now, the mere fact that a dimension is revealed that shows

three high factor loadings and two intermediate ones, evidences that the speaker evaluation

paradigm in general, and this experiment in particular, have the potential to detect structure in

the attitudes of 8-10-year-olds. It will be shown below that, as children move on in

adolescence, the structure of their sociolinguistic attitudes does not simply become tighter, as

could be expected if differences are explained through the fact that they develop better

abilities to handle questionnaires, but alters in more profound ways as well, and in line with

findings from research on psychosocial development.

The attributes in the ‘linguistic prestige’-dimension indicate that media language seems

to function as a model for children’s own linguistic behaviour. Assuming media influence on

how people speak is controversial (see Sayers 2014 and reactions to his article for discussion),

but the fact that eligibility as a TV-presenter is one of the highest loading factors on prestige

dimensions in all age groups, indicates that there is a strong consensus on the prestigious

nature of media language and, in all likelihood, also on how media language is composed.

Regardless of the question whether mass media actually contribute to the formation of norms,

which cannot be answered with our data, this testifies to the potential of TV and other media

in propagating linguistic norms. A second crucial feature is intelligibility. The role of

intelligibility ties in with research on standard language ideologies and standardisation in

Europe (e.g., Davies 2008, Jaspers & Van Hoof 2013), in which it is seen as the main factor

contributing to a variety’s prestige (however, there may be a reversed causal relation between

prestige and intelligibility as well; see Davies 2012): clearly, children have internalized this

preference for intelligible speech already at an early age.

Table 3. Rotated component matrix of PCA on attitudes of age group 8-10

Component

1 2 3 4 5

SMART ,714

HELPFUL -,465 ,477

UNDERSTANDABLE ,693

LEADER

SPEAK.LIKE ,685

TRUSTWORTHY ,866

FUNNY

FRIENDLY ,756

TV-PRESENTER ,668

POPULAR ,436

WELL-PAID ,809

FRIEND ,515

BEAUTIFUL ,445

According to an ANOVA on the regression factor scores, only the first dimension, ‘linguistic

prestige’, provokes different attitudes for the varieties under study (ANOVA: F=6,418, p <

.001), in that Brabantic is perceived as significantly less linguistically prestigious than

Standard Dutch and the Ghent accent (post-hoc Scheffé test, with p < .01), and the West

Flemish accent (with p < .05). The Kluisbergen accent, the children’s hometown variety,

behaves neutral with respect to all other varieties. Figure 2 displays the mean regression factor

score for each variety.

Figure 2. Mean REGR Factor score for Age group 8-10, factor 1 (‘linguistic prestige’),

broken down per variety (Brab = Brabantic, ST = Standard Dutch; Kluis = Kluisbergen

accent; Gent = Ghent accent; WVL = West Flemish accent)

At first sight, the 8-10-year-olds in our sample show an emergent sensitivity towards prestige

accents, since Standard Dutch and the very standard-near fragment from Ghent are evaluated

more positively, albeit only significantly so in comparison to a Brabantic accent. It is

tempting to believe that the higher scores for the Standard Dutch and the Ghent fragments are

triggered by their standard-like nature, but it cannot be ruled out that the differences in Figure

2 are due to low scores for the Brabantic accent. These could be the result of the fact that the

children may have been exposed to Brabantic to a lesser degree. While Brabantic is indeed

perceived to be the dominant regional accent in Flanders, it is nowhere as prominently present

as Standard Dutch, neither in the audio-visual media nor as a supraregional variety in informal

contexts in everyday life in East Flanders. In addition the Brabantic region is situated further

away from the hometown of our informants (Kluisbergen) than are Ghent and West-Flanders,

making real-life contacts with speakers of Brabantic somewhat more unlikely. While the

present data are inconclusive with respect to the question whether our test persons indeed

recognize prestige varieties of Dutch as such, any prestige they attribute to these varieties is a

purely linguistic one, as it does not correlate with typical non-linguistic prestige attributes

such as one’s income or intelligence,6 nor are these fragments evaluated as particularly

beautiful.

4.3. Age group 11-12

The 11-12-year-olds in our sample show slightly more structured attitudes, which can be

captured in 4 instead of 5 dimensions, and in which fewer attributes are observed that fit in

neither dimension. As for the first dimension, the features UNDERSTANDABLE, SPEAK.LIKE and

TV-PRESENTER now pattern with leadership qualities (LEADER) and accent euphony

(BEAUTIFUL). This cluster not only contains all attributes directly relating to speaking (i.e.

‘linguistic prestige’), but also the ability to lead a group. While leadership skills may in the

eyes of our informants relate to one’s verbal skills, this nevertheless provides a sign that our

informants perceive a correlation between the use of certain varieties and societal prestige and

success, which is richly documented in similar investigations, both in adults and in children.

The second dimension includes TRUSTWORTHY and FRIENDLY, and thus resembles ‘personal

integrity’-dimensions from other investigations. ‘Intelligence’ (SMART) remains a dimension

of its own, whereas the fourth dimension pairs popularity (POPULAR) to a negative score for

helpfulness (HELPFUL). While such a combination of attributes appears to be rare or even

absent in sociolinguistic research, the mixture of prosocial and antisocial traits is well-known

to occur frequently in children’s evaluation of their peers, leading psychologists to adopt the

label ‘perceived popularity’ for children who are evaluated as such (see Cillessen & Rose

2005 for discussion). The negative loading for helpfulness is reminiscent of correlations found

6 At least any such attitudes are not shared. It cannot be excluded on the basis of pooled results that a small

minority of the children in the study attribute more and/or stronger social evaluations to the accents under investigation. Given that children with a multivarietal background are, at least at the age of 7, better in discriminating between their own vis-à-vis other varieties of a language (Floccia, Butler, Girard & Goslin 2009), one would, for instance, expect such children to also acquire other sociolinguistic competences earlier.

between perceived popularity and, for instance, aggression and bullying, which, interestingly,

also vary with age: Cillessen & Rose (2005: 104) observe this correlation in 12-15-year-old

adolescents, but not in 9-11-year-olds. Hence the results in Table 4 show a relatively early

negative association between popularity and lack of helpfulness, which is, in all likelihood, a

trait that can be ascribed to psychosocial development.

Table 4. Rotated component matrix of PCA on attitudes of age group 11-12

Component

1 2 3 4

SMART ,813

HELPFUL -,803

UNDERSTANDABLE ,711

LEADER ,620 ,450

SPEAK.LIKE ,640 ,505

TRUSTWORTHY ,783

FUNNY ,465

FRIENDLY ,733

TV-PRESENTER ,752

POPULAR ,695

WELL-PAID ,453

FRIEND ,414 ,435

BEAUTIFUL ,756

Regression factor loadings depend on the variety evaluated only for two dimensions:

‘linguistic prestige/leadership’ (ANOVA: F=6.958, p < .001) and ‘intelligence’ (ANOVA:

F=2.539, p < .05). Mean scores for these dimensions are rendered in Figure 3. Post-hoc

Scheffé tests show that with regard to ‘linguistic prestige/leadership’ the Standard Dutch and

Ghent fragments are evaluated more positively than Brabantic (p < .01), and the Kluisbergen

and West-Flemish fragments are evaluated neutrally (Figure 3a). Similarly as for the results in

Figure 2, it cannot be determined whether these differences are explained by a more positive

evaluation of the standard-like fragments, a more negative evaluation of Brabantic, or both

factors simultaneously. No significant between-variety effects are observed for ‘intelligence’

(Figure 3b).

Figure 3. Mean REGR Factor score for factor 1 (‘linguistic prestige/leadership’; fig.3a) and

factor 3 (‘intelligence’; fig. 3b), broken down per variety (Brab = Brabantic, ST = Standard

Dutch; Kluis = Kluisbergen accent; Gent = Ghent accent; WVL = West Flemish accent)

4.4. Age group 13-14

In the informants of the age group 13-14, a further reduction of dimensions with eigenvalues

higher than 1 is observed: three dimensions explain 53,223% of the total variance. The first

dimension contains three ‘linguistic prestige’-attributes (UNDERSTANDABLE, TV-PRESENTER,

BEAUTIFUL) and shows a moderate factor loading for intelligence (SMART) (and to some extent

also for WELL-PAID, with a suppressed score of ,399). This dimension resembles the prestige-

dimension from other investigations. Remarkably, the loading for SPEAK.LIKE is only

moderately high, and in fact lower than for the second dimension, which groups the attributes

LEADER, FUNNY, POPULAR and FRIEND, making this dimension resemble Impe & Speelman’s

(2007) ‘social attractiveness’. This weakening association between prestige and the

SPEAK.LIKE attribute on the one hand, and the higher loading on social attractiveness suggests

an emergent susceptibility for covert prestige in our informants. The third dimension, with

high loadings for TRUSTWORTHY and FRIENDLY and a moderate one for HELPFUL, could be

called ‘personal integrity’, also in line with Impe & Speelman’s (2007) earlier investigation.

The attribute WELL-PAID gets a moderate loading on this dimension.

Table 5. Rotated component matrix of PCA on attitudes of age group 13-14

Component

1 2 3

SMART ,592

HELPFUL ,497

UNDERSTANDABLE ,664

LEADER ,679

SPEAK.LIKE ,453 ,480

TRUSTWORTHY ,651

FUNNY ,784

FRIENDLY ,728

TV-PRESENTER ,740

POPULAR ,720

WELL-PAID ,489

FRIEND ,757

BEAUTIFUL ,788

Scores for all dimensions correlate with the fragment that is evaluated (ANOVA: 1. ‘linguistic

prestige’: F=13.088, p < .001; 2. ‘social attractiveness’: F=14.586, p < .001; 3. ‘personal

integrity’: F=3.947, p < .01). Mean regression factor scores are shown in Figure 4. Post-hoc

Scheffé analyses show that the Standard Dutch and Ghent fragments are evaluated

significantly more positively than the other three fragments (Brabantic, Kluisbergen and

West-Flemish) when it comes to linguistic prestige (with p < .001 for all differences with

Standard Dutch, and p < .05 for Ghent vs. Brabantic, < .01 for Ghent vs. West-Flemish and <

.001 for Ghent vs. Kluisbergen) (see Figure 4a). For ‘social attractiveness’ (Figure 4b), the

picture is more complex. The Kluisbergen fragment receives the most positive evaluation,

significantly better than the Brabantic and Ghent fragments (p < .001) and marginally

significantly so than Standard Dutch (p < .10). The variety which is evaluated second best,

West-Flemish, only shows significant differences with Brabantic (p < .001). The difference

between Standard Dutch and Brabantic is significant as well (p < .01). Finally, ‘personal

integrity’-scores for West-Flemish are significantly lower than for the Ghent fragment (with p

< .05); the difference with Kluisbergen is marginally significant (p < .10) (Figure 4c).

Figure 4. Mean REGR Factor score for factor 1 (‘linguistic prestige’; fig.4a), factor 2

(‘social attractiveness’; fig.4b) and factor 3 (‘personal integrity’; fig. 4c), broken down per

variety (Brab = Brabantic, ST = Standard Dutch; Kluis = Kluisbergen accent; Gent = Ghent

accent; WVL = West Flemish accent)

These results show a positive evaluation of the standard-like fragments when it comes to

linguistic prestige and of the local variety (Kluisbergen) for social attractiveness. The latter

aspect was absent in younger age groups, and can be linked to Rys’s (2007) observation that

dialect proficiency in Flanders improves dramatically between the ages of 9, 12 and 15, and to

the more general observation that local varieties are often used as in-group varieties by

Flemish adolescents.

4.5. Age group 17-18

The oldest age group in our sample shows similar attitudes as the 13-14-year-olds, in that

there are three dimensions with eigenvalues higher than 1 explaining somewhat more than

50% of the total variance (56,610%). The composition of the dimensions has shifted,

however. The first dimension, which has all characteristics of a ‘prestige’-dimension, remains

a cluster of attributes relating to characteristics of speech, viz. intelligibility

(UNDERSTANDABLE), sounding like a TV-PRESENTER, and accent euphony (BEAUTIFUL), which

is associated with economic success (WELL-PAID), and, in contrast to 13-14-year olds, (again)

functions as a model for the informants’ own speech (SPEAK.LIKE). The second dimension is

now ‘personal integrity’, with the same combination of HELPFUL, TRUSTWORTHY and

FRIENDLY found in age group 13-14, and a somewhat surprising negative loading for

intelligence (SMART). The third dimension resembles the ‘social attractiveness’ of the

previous age group, with the attributes LEADER and POPULAR, but the at best moderate

loadings for FUNNY and FRIEND contrast with the structure of a typical ‘social attractiveness’-

dimension and suggest that (perceived) popularity no longer serves an equally important role

in friendship choices as it did in younger informants. Hence the label ‘in-group popularity’

seems to be more accurate. This finding again relates to adolescents’ psychosocial

development, in particular to the observation that, in contrast to the 13-14-year-olds, 17-18-

year-olds have passed the peak when it comes to conventional and social-clique dominated

reasoning about friendships (Turiel 1983, Horn 2003; see also below).

Table 6. Rotated component matrix of PCA on attitudes of age group 17-18

Component

1 2 3

SMART -,613

HELPFUL ,734

UNDERSTANDABLE ,747

LEADER ,638

SPEAK.LIKE ,706

TRUSTWORTHY ,718

FUNNY ,466 ,560

FRIENDLY ,662 ,405

TV-PRESENTER ,732

POPULAR ,703

WELL-PAID ,656

FRIEND ,407 ,485

BEAUTIFUL ,817

As shown in Figure 5, evaluations for all dimensions vary significantly for the varieties under

study (ANOVA: 1. ‘prestige’: F=11.208, p < .001; 2. ‘personal integrity’: F=4.347, p < .01; 3.

‘in-group popularity’: F=4.549, p < .01). As for the ‘prestige’-dimension (Figure 5a), both

Standard Dutch and the standard-near Ghent fragment are evaluated more positively than the

other three varieties (Post-hoc Scheffé: p < .01 re. Brabantic and Kluisbergen and p < .001 re.

West-Flemish). As for ‘personal integrity’ (Figure 5b), there is only a significant difference

between Kluisbergen and Standard Dutch (Post-hoc Scheffé: p < .01) and a marginally

significant one between Kluisbergen and West-Flemish (Post-hoc Scheffé: p < .10). For ‘in-

group popularity’ (Figure 5c) both the Kluisbergen and Standard Dutch fragment get a

significantly higher score than the Brabantic fragment (Post-hoc Scheffé: p < .05).

Figure 5. Mean REGR Factor score for factor 1 (‘prestige’; fig.5a), factor 2 (‘personal

integrity’; fig.5b) and factor 3 (‘in-group popularity’; fig. 5c), broken down per variety (Brab

= Brabantic, ST = Standard Dutch; Kluis = Kluisbergen accent; Gent = Ghent accent; WVL

= West Flemish accent)

The results in Figure 5 are for the larger part in line with the observation that Standard Dutch

functions as a prestige variety in Belgium. Standard Dutch sounds socially attractive to our

informants, but is associated with a lack of personal integrity. The standard-near Ghent

fragment is awarded similar scores for most dimensions. The use of the Kluisbergen variety of

Dutch, on the other hand, is perceived as not prestigious but both integer and popular, as

could be expected from a variety that is believed to function as an in-group variety for our

informants.

Less expected results are found for Brabantic, which is perceived as low in ‘in-group

popularity’. While this matches the results for social attractiveness in 13-14-year-olds, it

contrasts rather sharply with Impe & Speelman’s (2007) findings. Impe & Speelman (2007)

discuss pooled data from informants with a West-Flemish and Limburgian background, who

perceive of Brabantic as particularly socially attractive. In contrast, Ghyselen (2009), who

investigates attitudes in a West-Flemish location, reports negative attitudes towards Brabantic

regarding prestige, and attitudes comparable to those for Standard Dutch regarding her

‘solidarity’-dimension, which comprises attributes falling apart in ‘personal integrity’ and ‘in-

group popularity’ in the present investigation. The attitudes towards Brabantic we observe in

East Flanders are more consistent with Ghyselen’s (2009) than with Impe & Speelman’s

(2007), but share with the latter a structure along 3 dimensions rather than 2. In principle,

such differences should be treated with caution: the investigations under discussion proceed

along similar lines as ours, and typically hypothesize on the basis of experiments including

only one segment per variety. But while the role of particularities of a certain fragment or

speaker cannot be excluded, it is observed that such systematic attitudinal differences

regarding any other variety than Brabantic are rare, which corroborates the reliability of the

method, and suggests significant variation in the evaluation of Brabantic, probably depending

on the region where an investigation is carried out.

4.6. Critical assessment

As pointed out in the methodological discussion in section 3, speaker evaluation experiments

have been criticised for their underlying assumption that it is possible to decontextualise

speech samples and isolate the role of linguistic variation in their evaluation. As Soukup

(2013) argues, this criticism does not by definition render such experiments invalid, but it

necessitates proper reflection as to the experiment’s set-up and how this may influence the

results. Before moving on to a discussion of the main findings, therefore, a critical assessment

is in order of our data. A comparison of our data to similar investigations carried out within

the speaker evaluation paradigm seems to confirm the reliability of the method. Indeed the

most robust findings, which concern the prestige of the standard language, and the integrity,

social attractiveness and popularity of the local variety (the fragment from Kluisbergen), quite

closely match results from investigations like Impe & Speelman (2007), Ghyselen (2009),

Grondelaers, van Hout & Speelman (2011) and Ghyselen & De Vogelaer (2013). All these

investigations employ a similar, experimental method, and generalise on the basis of a small

number of fragments (typically one per variety). That, with respect to the evaluation of

Standard Dutch and local non-standard varieties, our findings clearly echo those from

previous studies, strongly suggests that the present investigation yields reliable data, and

indeed succeeded in foregrounding the role of geographical variation in the evaluation of the

speech samples that were used.

In addition to being corroborated by similar, experimental studies, the results show clear

similarities with data obtained through other types of research, including Van Hoof’s (2014)

analysis of language variation in TV fiction (a ‘societal treatment’ study, in Garrett’s 2007

terms), and Lybaert’s (2014) in-depth interviews with laymen on Flanders’ language situation

(exemplifying Garrett’s ‘direct approach’). According to Van Hoof (2014: chapter 6), TV-

fiction employs Standard Dutch for more formal interactions and in ‘modern’ situations.

Characters speaking Standard Dutch typically come from higher social classes, but may also

be unworldly and pretentious. Dialect indexes informality, rurality and tradition, and

profanity, but it is also used by funny and sympathetic characters. Lybaert’s (2014: 113,124)

informants in in-depth interviews characterise dialects in general as rather unintelligible and

coarse, but also easy-going and warm, and Standard Dutch as more intelligible, suitable for

the media, formal, but also somewhat cold. Specifically for adolescent language, the status of

local varieties as in-group language that emerged in our oldest age groups is documented for

many regions in Flanders (e.g., Rys 2007: 332-333, 355 for East Flemish; Jaspers &

Vandekerchove 2009: 4 and Vandekerckhove & Nobels for West-Flemish; Kuppens & De

Houwer 2006 for Antwerp).

Parallels with research such as Van Hoof’s (2014) and Lybaert’s (2014) show our

experimental method not only to be reliable, but, in general, also valid. Nevertheless, there are

a few findings that are without clear parallels. These mainly concern the evaluation of the

non-standard varieties in the speech samples that are external to the town were the

investigation was carried out. Grondelaers, van Hout & Speelman (2011) observe extensive

geographical variation in the evaluation of non-local, non-standard varieties, so the lack of

parallels has no negative implications as to the reliability of the data. The fragment from

Ghent in general triggered attitudes comparable to Standard Dutch, which was explained as a

result of its standard-near character. To show such an explanation to hold, however, requires

attitudinal data on more dialect-near Ghent speech, which are lacking. The results for the

fragments from Brabant and West-Flanders proved more problematic, since these varieties

have triggered divergent results in previous investigations by Ghyselen (2009), Ghyselen &

De Vogelaer (2013), and Impe & Speelman (2007). Our results show more similarities with

Ghyselen’s (2009) and Ghyselen & De Vogelaer’s (2013) results, and differ from Impe &

Speelman’s (2007). Awaiting further investigation, these divergences resist interpretation,

even more so because our findings cannot straightforwardly be extrapolated to adults’

attitudes. Even if the oldest of our informants are approaching adulthood, their real-life

exposure to varieties spoken outside the region may be low, as the geographical range of

high-school students is typically more restricted than of students and adults, who tend to

attend colleges or work in locations further away from their residence than most high

schools.7 It does not seem unlikely that exposure to more varieties in adulthood will change

attitudes, especially the evaluation of non-standard varieties from other regions.

5. Discussion: triggers for attitude change

More central to the goal of the present study than the evaluation of the varieties under study,

are attitudinal shifts observed in adolescence. It was shown that the youngest informants in

our study have rather rudimentary attitudes towards language variation, whereas the older age

groups have developed more adult-like attitudes as they went through adolescence. In general,

forms of linguistic prestige (e.g., traits like intelligibility or speaking like a TV-presenter) get

increasingly associated with societal prestige (through traits as intelligence or economic

success). Also, non-standard varieties get increasingly associated with forms of covert

prestige. During this process, several smaller but significant changes are observed, which are

further discussed below.

The question can be addressed what the cause is behind these changes in attitude

structure. Often-mentioned triggers for changing sociolinguistic attitudes are altering life

conditions. One change in adolescents’ linguistic environment is that parents, as their children

grow older, tend to shift towards non-standard language usage (De Houwer 2003; Ghimenton

& Chevrot 2006; Van de Mieroop, Zenner & Marzo, forthcoming). Although data appear to

be lacking, the same may hold for the audiovisual media, which are strongly associated with

the prestige dimension in all age groups in our study. In the Flemish context, however, at least

adults in certain circumstances favour non-standard varieties (for a case study in which,

7 In fact of the three recent surveys on attitudes in Flanders, Ghyselen (2009) is the only one discussing the age

factor, thereby concluding that age plays a relatively small role. Impe & Speelman (2007) explicitly extrapolate on the basis of data gathered in high schools; Grondelaers, van Hout & Speelman (2011) mention universities and university colleges as research sites without specifying whether the investigation was carried out in students or staff.

however, the role of the target audience’s age is not investigated, see Van Hoof 2014). The

best described factor influencing children and adolescents’ language usage is probably

schooling, which is found to impose prestige norms on children and foster an awareness

towards prestige varieties, and often correspondingly negative attitudes towards vernaculars

as well (Labov 1964, Cremona & Bates 1977, Day 1982).

Another determinant of language attitudes and behaviour is peer group influence and the

construction of a social identity (Labov 1972). Obvious effects of peer group influence are

observed in kids who move to other towns and are being included in new peer groups, which

is known to lead to processes of second dialect acquisition (e.g., Chambers 1992, Kerswill

1994). But peer group influence ranges beyond situations in which geographical mobility is

involved. For instance, Eckert (1989) shows that adolescents’ affiliation with certain groups

in high-school significantly steers variant linguistic usage. Thus, self-chosen peer

relationships are bound to have linguistic impact. Apart from language variation, peer group

influence is also deemed responsible for aspects of language development as diverse as

narrative competences (e.g., storytelling) or argumentation skills (Hoyle & Adger 1998: 12-

15).

The differences observed between the age groups in this study can hardly be due to

radically changing life conditions, however. Children in the target population do change

school around the age of 12 when they shift from primary to secondary school. But while this

involves increasing contact with children from other locations, these contacts still qualify as

very regional. As a result, the differences between age groups 2 and 3, i.e. before and after the

school change, are of a comparable magnitude as those between groups 1 and 2, and 3 and 4,

respectively, who attend the same schools. Hence the attitudinal changes seem to be triggered

by the ongoing psychosocial development on the children and adolescents’ side rather than by

external stimuli, which may or may not be mediated by altering compositions of social

networks. Indeed adolescence is associated not just with significant changes in the way

children interact with their peers but also with important shifts in the motivation and

evaluation of their and others’ social behaviour, which are arguably more fundamental than

behavioural aspects like network structure and help explain them (Killen, Rutland & Jampol

2009: 252). While this investigation is not tailored to mapping the psychosocial development

of our test persons in great detail, a comparison of subsequent age groups shows attitudes

shifting along lines well-known from research on psychosocial development. In comparison

to the youngest group, for instance, the 11-12-year-olds relate pro- and antisocial traits to each

other (e.g. being popular to not being helpful) in a way typical for their age (Cillessen & Rose

2005: 104). In 13-14-year-olds, the weakening association between prestige and the

SPEAK.LIKE attribute on the one hand, and the higher loading on social attractiveness suggests

an emergent susceptibility for covert prestige, that can be linked to an increased thinking in

function of the peer group, which is typical for adolescents. In particular, developmental

psychologists have observed adolescents’ conventionalized thinking about peer relations,

which is assumed to ‘peak’ around the age of 16 (Turiel 1983, Horn 2003). The present data

show that this age indeed seems to represent a (minor) turning point in the development of

sociolinguistic attitudes.

The inclusion of a question about the eligibility of a fragment’s speaker as a friend on

the questionnaire (the FRIEND-attribute) allows us to gain some insight in our test persons’

conception of peer relations. Figure 6 depicts all attributes showing a significant correlation

with the FRIEND-attribute for at least three age groups; non-significant correlations are

displayed as ‘zero’ in the graph.8

8 For completeness’ sake it needs to be added that UNDERSTANDABLE and BEAUTIFUL are not included, as they

do not straightforwardly relate to children’s thinking about their peer group structure as discussed in this paragraph. Intelligibility (UNDERSTANDABLE) appears to correlate with the eligibility as a friend in all age groups apart from the 13-14-year olds (to a stable strength, with .236 < r < .264), accent euphony in all age groups apart from the 8-10-year-olds (with r = .357, .248 and .369, respectively).

Figure 6. Correlation coefficients to attributes correlating with the FRIEND-attribute

The most spectacular development is found for popularity, which is the single feature

correlating with the eligibility as a friend in the youngest age group, gains weight in the

second and third age groups, but loses significance in the oldest informants. A similar rise and

(less outspoken) drop are observed for the LEADER and FUNNY attributes. The only attribute

which correlates more strongly with the eligibility as a friend in the oldest age group in

comparison to the 13-14-year olds, is trustworthiness, a characteristic hardly relating to

behaviour in groups, but rather to interpersonal relations between individuals.

Against the backdrop of changing psychosocial thinking, language attitudes differ

subtly between the third and fourth age group, especially with regard to the local variety: the

Kluisbergen variety develops from an ‘across-the-board’ socially attractive variety towards a

variety indexing integrity, which yields popularity within the peer group only to the same

extent as Standard Dutch, and which no longer influences the eligibility as a friend for 17-18-

year olds as it did for 13-14-year-olds (cf. above). It seems not far-fetched to hypothesize that

attitudes in 17-18-year olds are somewhat less favourable (though not hostile) towards the use

of vernacular varieties. Such a hypothesis ties in pretty well with results obtained for Black

English in the US. Indeed Van Hofwegen & Wolfram’s (2010: 439) ‘roller-coaster trajectory’

describes lower degrees of vernacularity towards the end of the high school period for the

majority of their subjects.

6. Conclusions

From the results presented above, it can be concluded that adolescence is a period in which, at

least in the Flemish context, language attitudes further develop and/or change considerably.

The youngest children in our sample do seem to recognize Standard Dutch as a model for

their own speech, and are thus competent to distinguish between different varieties of Dutch,

but they hardly attribute any non-linguistic significance to language variation. As children

grow older, they realize that there is a correlation between language variation and societal

prestige. In addition, they become more sensitive to the ‘covert prestige’ of, especially, the

local variety, which is increasingly evaluated as indexing integrity and as a means towards

social and/or in-group success. The attitudes of the oldest age groups in the sample match

pretty well to findings of other investigations, most notably Impe & Speelman (2007), who

also detect attitudes structured along three, highly similar dimensions, and Grondelaers, van

Hout & Speelman (2011), with their four-dimensional landscape. Comparison with these

studies, however, is hampered through the fact that they only provide analyses on pooled data

from several regions, even though Grondelaers, van Hout & Speelman explicitly state that

important attitudinal differences are observed in different regions in Flanders (as also

illustrated by Ghyselen 2009). In the region in which this investigation is carried out, Rys

(2007) documents an increasing dialect proficiency in adolescence, which is even stronger in

subjects reporting positive attitudes towards the dialect. This correlation is also observed in

other linguistic areas (see Rys 2007: 117-118 for discussion), and taken as evidence for a

causal connection between attitudes and increasing proficiency, which may run in either

direction: positive attitudes may stimulate the acquisition of the dialect, but adolescents

gaining social success through the use of the dialect may also develop more positive attitudes

towards it.

It remains an open question whether or when these changes in sociolinguistic attitudes

come to an end. Given the widely held (and to some extent empirically proven) assumption in

linguistics that language usage remains fairly stable in adulthood (see Sankoff & Blondeau

2007: 582-584 for discussion), one would not expect major attitudinal changes beyond the age

of, say, 20, and would accordingly explain attitudinal differences between age groups as a

result of change-in-progress. In developmental psychology, however, it is not uncommon at

all to assume ongoing psychosocial development in adulthood. For instance, in Erikson’s

(1959) influential 8-stage-model of the life cycle, adulthood falls apart in 3 stages

characterized by, among other things, proper challenges in the domains of love, parenting, and

work. It remains to be investigated whether ongoing psychosocial development in adulthood

also affects linguistic attitudes. In any case, the present results cast some doubt over the

current methodological practice of extrapolating hypotheses on language attitudes from

investigations carried out in high schools and universities.

The development of sociolinguistic competence is just one aspect of social development

and therefore changes observed in the general psychosocial reasoning of adolescents are

expected to trickle through in the attitudes towards language variation. While the

questionnaire used to capture linguistic attitudes was not geared towards mapping children’s

general psychosocial development, a number of significant parallels are revealed between

sociolinguistic and psychosocial development: in 11-12-year-olds for instance, the correlation

between a prosocial attribute like popularity and an antisocial one like not being helpful is

very reminiscent of this age group’s thinking in terms of ‘perceived popularity’ (Cillessen &

Rose 2005), and the loosening of the correlation between the friendship and popularity

attributes after the age of 16 coincides with passing the peak in conventional and social-clique

dominated reasoning about friendships (Turiel 1983, Horn 2003). Systematic research into

such parallels will undoubtedly reveal more of them, and provide further insight in the

characteristics of youth language in general, and phenomena relating to its vernacularity in

particular (cf. the fluctuations observed by Van Hofwegen and Wolfram 2010). In order to

obtain such insights, however, we need to place developmental considerations much higher on

the sociolinguistic agenda.

References

Absillis, Kevin, Jaspers, Jürgen & Van Hoof, Sarah. 2012. De manke usurpator. Inleiding. In

De manke usurpator. Over Verkavelingsvlaams, Kevin Absillis, Jürgen Jaspers & Sarah

Van Hoof (eds), 3-35. Gent: Academia Press.

Auer, Peter. 2005. Europe's sociolinguistic unity, or: A typology of European dialect/standard

constellations. In Perspectives on variation: Sociolinguistic, historical, comparative.

Nicole Delbecque, Johan van der Auwera & Dirk Geeraerts (eds), 7-42. Berlin: Mouton

de Gruyter.

Blum-Kulka, Shoshana. 2004. The role of peer interaction in later pragmatic development: the

case of speech representation. In Language development across childhood and

adolescence, Ruth Berman (ed), 191–210. Amsterdam: John Benjamins.

Cajot, José. 1998. Een omgangstaal voor alledag. Vlaanderens eigen weg. Streven 65: 999-

1008.

Chambers, Jack. 1992. Dialect acquisition. Language 68: 673–705

Chambers, Jack. 2003. Sociolinguistic Theory. Oxford: Blackwell.

Cillessen, Antonius & Rose, Amanda. 2005. Understanding Popularity in the Peer System.

Current Directions in Psychological Science 14: 102-110.

Craig, Holly & Washington, Julie. 2006. Malik Goes to School: Examining the Language

Skills of African American Students from Preschool–5th Grade. Mahwah, New Jersey:

Lawrence Erlbaum Associates.

Cremona, Cristiana & Bates, Elizabeth. 1977. The development of attitudes toward dialect in

Italian children. Journal of Psycholinguistic Research 6: 223-232.

Davies, Winifred. 2008. 'Sprachkultur' in lay and academic discourse in modern Germany.

German Life and Letters 61: 435-450.

Davies, Winifred. 2012. Myths we live and speak by: ways of imagining and managing

language and languages. In Standard languages and multilingualism in European

history, Matthias Hüning, Ulrike Vogl & Oliver Moliner (eds), 45-69. Amsterdam: John

Benjamins.

Day, Richard. 1980. The development of linguistic attitudes and preferences. TESOL

Quarterly 14: 27-37.

Day, Richard. 1982. Children’s attitudes toward language. In Attitudes towards language

variation. Social and Applied contexts, Ellen B. Ryan & Howard Giles (eds), 116-131.

London: Edward Arnold.

Deprez, Kas. 1999. De taal van de Vlamingen. In Nationalisme in België. Identiteiten in

beweging (1780-2000), Kas Deprez & Louis Vos (eds), 106-116. Antwerpen: Houtekiet.

[reprinted in Absilis, Jaspers & Van Hoof 2012]

De Houwer, Annick. 2003. Language variation and local elements in family discourse.

Language variation and change 15: 329-349.

Dillard, Joey Lee. 1972. Black English: Its History and Usage in the United States. New

York: Random House.

Eckert, Penelope 1989. Jocks and Burnouts: Social Identity in the High School. New York:

Teachers College Press.

Eckert, Penelope. 2012. Three waves of variation study: The emergence of meaning in the

study of sociolinguistic variation. Annual Review of Anthropology 41: 87–100.

Erikson, Erik. 1959. Identity and the life cycle. New York: International Universities Press.

Fasold, Ralph. 1972. Tense Marking in Black English: A Linguistic and Social Analysis.

Arlington, Virginia: Center for Applied Linguistics.

Ferguson, Charles. 1959. Diglossia. Word 15: 325–340.

Floccia, Caroline, Butler, Joseph, Girard, Frédérique & Goslin, Jeremy. 2009. Categorization

of regional and foreign accent in 5- to 7-year-old British children. International Journal

of Behavioral Development 33: 366-375.

Garrett, Peter. 2007. Language attitudes. In The Routledge companion to sociolinguistics,

Carmen Llamas, Louise Mullany, Peter Stockwell (eds), 116-121. London: Routledge.

Garrett, Peter. 2010. Attitudes to Language. Cambridge: Cambridge University Press.

Garrett, Peter, Coupland, Nikolas & Williams, Angie. 1999. Evaluating dialect in discourse:

Teachers' and teenagers' responses to young English speakers in Wales. Language in

Society 28: 321-354.

Ghimenton, Anna. 2013. Reading between the code choices: Discrepancies between

expressions of language attitudes and usage in a contact situation. International Journal

of Bilingualism. <Online before print: doi: 10.1177/1367006913509900>

Ghimenton, Anna & Chevrot, Jean-Pierre. 2006. Language acquisition in a multilingual

society: a case study in Veneto, Italy. In Language variation – European perspectives,

Frans Hinskens (ed), 71-81. Amsterdam: John Benjamins.

Ghyselen, Anne-Sophie. 2009. Ne zelfzekere leraar of gewoon nen enthousiaste mens? Een

matched-guise onderzoek naar de attitude tegenover tussentaal bij West-Vlamingen.

Taal en Tongval 61: 83-113.

Ghyselen, Anne-Sophie & De Vogelaer, Gunther. 2013. Endoglossic standardisation in

Flanders: an attitudinal perspective. In Language (De)standardisation in Late Modern

Europe: Experimental Studies, Tore Kristiansen & Stef Grondelaers (eds), 153-170.

Oslo: Novus.

Giles, Howard & Billings, Andrew. 2004. Language attitudes. In The Handbook of applied

Linguistics, Alan Davies & Catherine Elder (Eds.), 187-209. Oxford, UK: Blackwell.

Giles, Howard, Harrison, Chris, Creber, Clare, Smith, Philip & Freeman, Norman (1983).

Developmental and contextual aspects of children's language attitudes. Language &

Communication 3: 141-146.

Goossens, Jan. 1970. ‘Belgisch beschaafd Nederlands’ en Brabantse expansie. De Nieuwe

Taalgids 63: 54-70.

Goossens, Jan. 2000. De toekomst van het Nederlands in Vlaanderen. Ons Erfdeel 43: 3-13.

Grondelaers, Stef & van Hout, Roeland. 2011. The standard language situation in the Low

Countries: Top-down and bottom-up variations on a diaglossic theme. Journal of

Germanic Linguistics 23: 199-243.

Grondelaers, Stef, van Hout, Roeland & Speelman, Dirk. 2011. A perceptual typology of

standard language situations in the Low Countries. In Standard Languages and

Language Standards in a Changing Europe, Tore Kristiansen & Nikolas Coupland

(eds), 199-222. Oslo: Novus Press.

Kristiansen, Tore & Grondelaers, Stefan. 2013. Language (de)standardisation in Late Modern

Europe: Experimental Studies (eds.). Oslo: Novus Press.

Horn, Stacey. 2003. Adolescents’ reasoning about exclusion from social groups.

Developmental Psychology 39: 11-84.

Hoyle, Susan M. & Adger, Carolyn Temple. 1998. Introduction. In Kids Talk: Strategic

Language Use in later Childhood, Susan M. Hoyle & Carolyn Temple Adger (eds), 3-

22. Oxford: Oxford University Press.

Impe, Leen & Speelman, Dirk. 2007. Vlamingen en hun (tussen)taal. Een attitudineel mixed

guiseonderzoek. Handelingen van de Koninklijke Maatschappij voor Taal- en

Letterkunde en Geschiedenis 61: 109-128.

Jaspers, Jürgen. 2011. Strange bedfellows. Appropriations of a tainted urban dialect. Journal

of Sociolinguistics 15: 493-524.

Jaspers, Jürgen & Van Hoof, Sarah. 2013. Hyperstandardisation in Flanders: extreme

enregisterment and its aftermath. Pragmatics 23: 331-359.

Jaspers, Jürgen & Vandekerckhove, Reinhild. 2009. Jong Nederlands. Nederlandse Taalkunde

14: 2-7.

Jaspers, Jürgen. 2005. Tegenwerken, belachelijk doen. Talige sabotage van Marokkaanse

jongens op een Antwerpse middelbare school. Een sociolinguïstische etnografie.

Brussel: VUB press.

Kerswill, Paul. 1994. Dialects converging: Rural Speech in urban Norway. Oxford: Oxford

University Press.

Kerswill, Paul & Williams, Ann. 2000. Creating a new town koine: children and language

change in Milton Keynes. Language in Society 29: 65-115.

Killen, Melanie, Rutland, Adam & Jampol, Noah Simon. 2009. Social exclusion in childhood

and adolescence. In Handbook of peer interactions, relationships, and groups, Kenneth

Rubin, William Bukowski & Brett Laursen (eds), 249-266. New York: The Guildford

Press.

Kinzler, Katherine & Jasmine DeJesus. 2013. Northern = smart and southern = nice: The

development of accent attitudes in the United States. The Quarterly Journal of

Experimental Psychology 66: 1146-1158.

Kristiansen, Gitte. 2010. Lectal acquisition and linguistic stereotype formation: an empirical

study. In Advances in cognitive Sociolinguistics, Dirk Geeraerts, Gitte Kristiansen &

Yves Peirsman (eds), 225-264. Berlin: Mouton de Gruyter.

Kristiansen, Tore. 2011. Attitudes, ideology and awareness. In The SAGE Handbook of

Sociolinguistics, Ruth Wodak, Barbara Johnstone & Paul Kerswill (eds.), 265-278.

London: SAGE Publications.

Kuppens, An & De Houwer, Annick. 2006. Van Alijs tot Zapt’em: De zelfgerapporteerde taal

van Antwerpse jongeren. In Artikelen van de vijfde sociolinguïstische conferentie, Tom

Koole, Jacomine Nortier & Bert Tahitu (eds), 319-329. Delft: Eburon.

Labov, William. 1965. Stages in the acquisition of Standard English. In Social Dialects and

Language Learning, Roger W. Shuy (ed), 77–103. Champaign, Illinois: National

Council of Teachers of English.

Lambert, Wallace, Hodgson, Richard, Gardner, Robert & Fillenbaum, Samuel. 1960.

Evaluational reactions to spoken languages. Journal of Abnormal and Social

Psychology 60: 44-51.

Lambert, Wallace, Franckel, Hannah & Tucker, Richard. 1966. Judging personality through

speech: A French-Canadian example. Journal of Communication 16: 305-321.

Lybaert, Chloé. 2014. Het gesproken Nederlands in Vlaanderen. Percepties en attitudes van

een spraakmakende generatie. Ph.D-dissertation Ghent University.

Niedzielski, Nancy & Dennis Preston. 2003. Folk linguistics. Berlin: Mouton de Gruyter.

Payne, Arvilla. 1980. Factors controlling the acquisition of the Philadelphia dialect by out-of-

state children. In Locating Language in Time and Space, William Labov (ed), 143-178.

New York: Academic Press.

Plevoets, Koen. 2008. Tussen spreek- en standaardtaal. Een corpusgebaseerd onderzoek naar

de situationele, regionale en sociale verspreiding van enkele morfosyntactische

verschijnselen uit het gesproken Belgisch-Nederlands. Ph.D-dissertation KULeuven.

Rosenthal, M.S. 1974. The magic boxes: pre-school children’s attitudes toward black and

standard English. The Florida FL Reporter (Spring/Fall 1974): 55-93.

Ryan, Ellen B. 1979. Why do low-prestige varieties persist? In Language and social

Psychology, Howard Giles & Robert St. Clair (eds), 145-157. Oxford: Basil Blackwell.

Rys, Kathy. 2007. Dialect as a Second Language: Linguistic and Non-Linguistic Factors in

Secondary Dialect Acquisition by Children and Adolescents. Ph.D-Dissertation Ghent

University.

Sankoff, Gillian & Blondeau, Hélène. 2007. Longitudinal Change across the lifespan: r in

Montreal French. Language 83, 560-588.

Sarnoff, Irving. 1970. Social Attitudes and the Resolution of Motivational Conflict.

Harmondsworth: Penguin.

Sayers, Dave. 2014. The mediated innovation model: A framework for researching media

influence in language change. Journal of Sociolinguistics 18: 185–212.

Smith, Jennifer, Durham, Mercedes & Fortune, Liane. 2007. “Mam, my trousers is fa’in

doon!”: Community, caregiver, and child in the acquisition of variation in a Scottish

dialect. Language Variation and Change 19: 63-99.

Smith-Christmas, Cassie. 2014. Being socialised into language shift: the impact of extended

family members on family language policy. Journal of Multilingual and Multicultural

Development. <online first, doi: 10.1080/01434632.2014.882930>

Soukup, Barbara. 2013. The measurement of ‘language attitudes’ - a reappraisal from a

constructionist perspective. In Language (De)standardisation in Late Modern Europe:

Experimental Studies, Tore Kristiansen & Stefan Grondelaers (eds), 251-266.

Copenhagen: Novus Press.

Stewart, William A. 1965. Urban Negro speech: Sociolinguistic factors affecting English

teaching. In Social Dialects and Language Learning, Roger W. Shuy (ed), 10–18.

Champaign, Illinois: National Council of Teachers of English.

Taeldeman, Johan. 2008. Zich stabiliserende grammaticale kenmerken in Vlaamse tussentaal.

Taal en Tongval 60: 26-50.

Turiel, Elliot. 1983. The Development of social Knowledge: Morality and Convention.

Cambridge: Cambridge University Press.

Van de Mieroop, Dorien, Zenner, Eline & Marzo, Stefania (forthcoming). Standard and

Colloquial Belgian Dutch pronouns of address: A variationist-interactional study of

child-directed speech in dinner table interactions. Ms. KULeuven.

Van Hofwegen, Janneke & Wolfram, Walt. 2010. Coming of age in African American

English: A longitudinal study. Journal of Sociolinguistics 14: 427–455.

Van Hoof, Sarah. 2014. Feiten en fictie. Een sociolinguïstische analyse van het taalgebruik in

fictiereeksen op de Vlaamse openbare omroep (1977-2012). Gent: Academia Press.

Van Hoof, Sarah & Jaspers, Jürgen. 2012. Hyperstandaardisering. Tijdschrift voor

Nederlandse Taal- en Letterkunde 108: 97-125.

Van Hoof, Sarah & Vandekerckhove, Bram. 2013. Feiten en fictie: taalvariatie in Vlaamse

televisiereeksen vroeger en nu. Nederlandse taalkunde 18: 35-64.

Vandekerckhove, Reinhild & Nobels, Judith. 2010. Code eclecticism: linguistic variation and

code alternation in the chat language of Flemish teenagers. Journal of sociolinguistics

14: 657-677.

Vosters, Rik, Rutten, Gijsbert & van der Wal, Marijke. 2010. Mythes op de pijnbank. Naar

een herwaardering van de taalsituatie in de Nederlanden in de achttiende en negentiende

eeuw. Verslagen en Mededelingen van de Koninklijke Academie voor Nederlandse

Taal- en Letterkunde 120: 93-112.

Willemyns, Roland. 2005. Verkavelingsbrabants. Werkt het integratiemodel ook voor

tussentalen? Neerlandica extra muros 43: 27-40.

Wolfram, Walt. 1969. A Sociolinguistic Description of Detroit Negro Speech. Washington,

D.C.: Center for Applied Linguistics.

Zahn, Christopher & Hopper, Robert. 1985. Measuring language attitudes: The speech

evaluation instrument. Journal of Language and Social Psychology 4: 113-123

Appendix: questionnaire

Fragment

We hebben thuis een draakje gehad. Ik zal nooit die ochtend in april vergeten dat ik hem voor

het eerst zag. Mijn broer en ik gingen het varkenshok binnen om naar de biggetjes te kijken

die ‘s nachts geboren waren. Daar lag het grote moedersvarken met tien kleine biggetjes dicht

tegen haar aan in het stro maar helemaal alleen in een hoekje stond een klein groen draakje

met boze oogjes. “Wie zou dat zijn?”, vroeg mijn broertje stomverbaasd. “Ik denk dat het een

draak is,” zei ik, “het moedervarken heeft tien biggetjes gekregen en een draakje.” Het is echt

waar maar hoe het precies gebeurd is zullen we nooit te weten komen. Ik geloof dat het

varken even verbaasd was als wij. Ze moest in het begin niet veel van haar drakekind hebben

maar na een tijdje raakte ze er wel aan gewend. Ze kon er alleen niet aan wennen dat ze

tijdens het drinken telkens door hem werd gebeten. Op het laatst kreeg ze daar zelfs zo genoeg

van dat ze hem niet meer wilde voeden. Mijn broer en ik moesten toen elke dag naar het

varkenshok met drakevoer, dat bestond uit stukjes kaars touwtjes en nog meer van dat soort

lekkernijen, en het draakje zou vast van de honger zijn omgekomen als mijn broer en ik niet

trouw met ons mandje naar het varkenshok waren gegaan. Zodra we de deur openmaakten

begonnen alle biggetjes te knorren van de honger maar het draakje stond alleen maar stil naar

ons te staren met z’n kleine rode ogen. Er kwam geen geluid uit z’n keel. Alleen als hij klaar

was met eten liet hij een paar flinke boeren en kwispelde tevreden met zijn staart.

I remember our dragon. I shall never forget that morning in April when I saw him for the first

time. My brother and I went to the pigpen to watch the piglets born that night. There was the

big mother pig lying in the straw with ten little pigs close to her. All alone in a corner stood a

little green dragon with angry eyes. "Who would that be?", asked my brother perplexed. "I

think it's a dragon," I said, "the mother pig got ten piglets and a dragon." This is really true,

but we will never know how this could ever happen. I believe that the pig was as surprised as

we were. At first she wasn’t too fond of her dragon child but after a while she got used to it.

All she could not get used to was that she was always bitten by him when he was drinking. In

the end she even got so fed up that she did not want to feed him anymore. My brother and I

then had to go to the pigpen with dragon food every day, which consisted of pieces of candle

wicks and other such delights, and the dragon would surely had starved to death if my brother

and I would not have gone to the pigpen every day with our basket. As soon as we opened the

door all the pigs started grunting from hunger, but the little dragon just stood there staring at

us with his little red eyes. No sound came from his throat. Only when he had finished eating,

he burped a few times and happily wagged his tail.

[translation is mine and may differ from the commercial translation]

Questions per fragment

1) Volgens mij is de spreker van dit fragment een slimme man // SMART

According to me, the speaker of this fragment is smart.

2) De spreker van dit fragment is volgens mij iemand die graag andere mensen helpt //

HELPFUL

According to me the speaker of this fragment likes to help other people.

3) De spreker van het fragment was moeilijk te verstaan // UNDERSTANDABLE (reversed)

The speaker was hard to understand.

4) De spreker van dit fragment zou goed leiding kunnen geven over een groep // LEADER

The speaker would be able to lead a group well.

5) Ik zou zelf graag zoals deze persoon spreken // SPEAK.LIKE

I would like to speak like this person.

6) De spreker van dit fragment is iemand die je kan vertrouwen // TRUSTWORTHY

The speaker is someone you can trust.

7) Volgens mij is deze spreker een grappige man en kan hij goed mensen aan het lachen

brengen // FUNNY

I think this speaker is a funny man and he can easily make people laugh.

8) De spreker lijkt mij geen vriendelijke man // FRIENDLY (reversed)

The speaker does not appear to be a friendly man.

9) Deze spreker zou een goede presentator op televisie zijn // TV-PRESENTER

The speaker would be a good TV-presenter.

10) Deze man is eenzaam // POPULAR (reversed)

This man is lonely.

11) De spreker van dit fragment heeft een goedbetaald beroep // WELL-PAID

The speaker has a well-paid job.

12) Deze spreker zou ik als vriend willen hebben // FRIEND

I would like to have this speaker as a friend.

13) Deze spreker sprak mooi // BEAUTIFUL

This speaker spoke beautifully.

14) Uit welke streek denk je dat de spreker die je hebt gehoord afkomstig is?

Which region do you think this speaker comes from?