Language Attitudes of Adults Living in Friesland
Towards the Frisian Language
Femke Swarte
Research Master Linguistics
Faculty of Arts
University of Groningen
Supervisors:
Dr. N.H. Hilton (University of Groningen)
Dr. E. Klinkenberg (Fryske Akademy)
ii SUMMARY
This study investigated the language attitudes of the inhabitants of Friesland towards Frisian after a period in which the Province of Friesland was concerned with the encouragement of using the Frisian language more frequently and in which the number of trilingual schools (where Frisian plays an important role as a language of education) was expanding rapidly. Among other instruments, the Matched Guise Technique was used to measure the attitudes of adults living in Friesland towards Frisian. This has not been done before on such a large scale as in the present study. The results show that the attitudes towards Frisian partially changed over the years. Frisian has gained more status among the inhabitants of Friesland. Nevertheless, the attitudes towards learning Frisian are very negative among people who do not speak Frisian.
Secondly, this study investigated if the attitudes of parents of children attending trilingual schools differ positively from those of parents of children that attend monolingual schools. This was not the case.
iii GEARFETTING
Yn dit ûndersyk is ûndersocht hokker taalattitudes de ynwenners fan Fryslân tsjinoer it Frysk ha, nei in perioade weryn de Provinsje Fryslân besocht hat it gebrûk van de Fryske taal oan te moedigjen en it oantal trijetalige skoallen (werby it Frysk als instruksjetaal in grutte rol spilet) hurdt groeid is. Njonken oare metoades is the Matched Guise Technique brûkt om de taalattitudes fan folwoeksenen út Fryslân mjitte te kinnen. Dat is noch nea op sa’n grutte skaal dien as yn dit ûndersyk. De resultaten litte sjen dat de attitudes tsjinoer it Frysk troch de jierren hinne in lyts bytsje feroare binne. It Frysk hat mear status krigen ûnder de ynwenners fan Fryslân. De attitudes fan mensken die gjin Frysk prate tsjinoer it learen fan de Fryske taal binne lykwols noch tige negatyf.
As twadde hold dit ûndersyk him dwaande mei de fraach oft de attitudes fan âlders fan learlingen fan trijetalige skoallen positiver binne dan dy fan âlders fan bern fan inkeltalige skoallen. Dat wie net it gefal.
iv SAMENVATTING
In dit onderzoek werd onderzocht welke taalattitudes tegenover het Fries de inwoners van Friesland hebben, na een periode waarin de Provincie Friesland geprobeerd heeft het gebruik van het Fries aan te sporen en waarin het aantal drietalige scholen (waarin het Fries een grote rol speelt als instructietaal) snel gegroeid is. Naast andere methodes, is er gebruik gemaakt van de Matched Guise Technique, om de taalattitudes van volwassenen uit Friesland te meten. Voorheen is dit nog nooit op zo’n grote schaal gedaan als in dit onderzoek. De resultaten tonen aan dat de attitudes tegenover het Fries door de jaren heen gedeeltelijk veranderd zijn. Het Fries heeft onder de inwoners van Friesland meer status verworven. De attitudes van mensen die geen Fries spreken tegenover het leren van Fries zijn echter zeer negatief.
Ten tweede hield dit onderzoek zich bezig met de vraag of de attitudes van ouders van kinderen van drietalige basisscholen tegenover het Fries positiever zijn dan die van ouders van kinderen van enkeltalige scholen. Dit was niet het geval.
v ACKNOWLEDGMENTS
I would like to give special thanks to my supervisors, Nanna Haug Hilton and Edwin Klinkenberg, for their help and supervision. I would also like to thank the Fryske Akademy for giving me the opportunity to do the interesting internship that resulted in this thesis. I want to thank my colleagues, especially Kobe, Adrie and Saskia, for making every working day at the Fryske Akademy very enjoyable for me.
I thank all the people that participated in my project and the people that spread my survey among their friends and family, especially Jelle Bangma (CEDIN), Idske and Tjallien. Special thanks goes out to the three speakers that recorded the fragments for my Matched Guise investigation.
vi TABLE OF CONTENTS SUMMARY ii GEARFETTING iii SAMENVATTING iv ACKNOWLEDGMENTS v TABLE OF CONTENTS vi LIST OF FIGURES ix LIST OF TABLES xi 1. Introduction 1 2. Literature Review 4
2.1 Theorising Language Attitudes 4
2.1.1 Towards a Definition of ‘Language Attitudes’ 4 2.1.2 Main Approaches towards Studying Language Attitudes 7
2.1.2.1 Direct Approaches 7
2.1.2.2 Indirect Approaches 9
2.2 Previous Studies into Language Attitudes 13
2.2.1 Studies into Attitudes towards Frisian 13
2.2.1.1 Direct Methods 13
2.2.1.2 Indirect Methods 15
2.2.1.3 Language Surveys 19
2.2.2 Research into Attitudes towards Other Minority Languages 21
2.2.2.1 Welsh 22
2.2.2.2 Catalan 24
2.3 The Present Study 27
3. Methodology 31
3.1 Instruments 31
3.1.1 The Matched Guise Test 31
3.1.2 Direct Questionnaire 35
3.2 Background Variables 38
3.3 Setting 38
vii
3.5 Sample 40
3.6 Processing the Data 41
4. Results 43
4.1 The Matched Guise Test 43
4.1.1 Overall Ratings 43 4.1.2 Dimensions 44 4.1.2.1 Attractiveness Ratings 44 4.1.2.2 Status Ratings 47 4.1.2.3 Integrity Ratings 48 4.1.3 Personality Traits 48 4.1.3.1 Overall Ratings 48 4.1.3.2 Friendliness Ratings 51
4.1.3.3 Perceived Wealth Ratings 51
4.1.3.4 Honesty Ratings 53 4.1.3.5 Attractiveness Ratings 53 4.1.3.6 Intelligence Ratings 56 4.1.3.7 Helpfulness Ratings 57 4.1.3.8 Summary 57 4.2 Statements 58 4.2.1 Personality Traits 58 4.2.2 Learning Frisian 59 4.3 Job Positions 62 4.4 Situations of Conversation 64
4.4.1 Answering a Dutch Question in Frisian 64
4.4.2 Giving Someone a Ticket to a Football Game 67
4.4.3 A Teacher who speaks Frisian in a Dutch School 70
4.4.4 Going to the Hairdresser 72
4.4.5 Summary 75
5. Discussion 77
5.1 Language Attitudes in Friesland – Overall Results Discussed 77 5.2 The Difference Between Parents with Children Attending Trilingual Schools
and Parents Attending Monolingual Schools 78
5.3 Differences Between Overt and Covert Attitudes 80
viii
6. Conclusion 86
BIBLIOGRAPHY 88
APPENDIX A 93
LIST OF FIGURES
1 The three components attitudes are based on according to the multiple
component definition 5
2 Semantic Differential scale, form I 10
3 Semantic Differential scale, form II 10
4 The home language of the participants in 1980 and 1987 (Woolard and Gahng) 25 5 The correlation between the level of Frisian and the scores of the Frisian
fragments 44
6 The correlation between age and the attractiveness scores of the Frisian
fragments 45
7 The correlation between level of Frisian and the attractiveness scores of
the Frisian fragments 46
8 The correlation between level of Frisian and the attractiveness scores of the
Dutch fragments 46
9 The correlation between level of Frisian and the status scores of the Frisian
fragments 48
10 The ratings of the Dutch fragments on the personality traits 49 11 The ratings of the Frisian fragments on the personality traits 50 12 The correlation between age and the friendliness scores of the Dutch fragments 51 13 The correlation between level of Frisian and the perceived wealth scores of the
Frisian fragments 52
14 The correlation between age and the attractiveness scores of the Frisian
fragments 54
15 The correlation between age and the attractiveness scores of the Dutch
fragments 54
16 The correlation between level of Frisian and the attractiveness scores of the
Frisian fragments 55
17 The correlation between level of Frisian and the attractiveness scores of the
Dutch fragments 55
18 The correlation between level of Frisian and the intelligence scores of the
Frisian fragments 57
19 Direct and indirect ratings of Frisian 59/81
x 21 The correlation between level of Frisian and the attitudes towards learning
Frisian 61
LIST OF TABLES
1 Source information about responses (Ajzen 1988) 6 2 The speakers and the fragments they recorded (Jonkman 1982) 16 3 The outcomes of earlier research projects into language attitudes towards Frisian 27
4 The duration of the fragments in seconds 34
5 Background of the participants of the pilot study 40
6 Background of the participants 41
7 Labels of the background variables 42
8 The significant background variables for the overall scores, the dimensions and
the personality traits, ordered by language 58
9 Internal reliability coefficients of the statements about learning Frisian 59 10 The association between L1 and reactions to someone who answers a Dutch
question in Frisian 65
11 The association between home languages and reactions to someone who
answers a Dutch question in Frisian 66
12 The association between place of birth and reactions to someone who
answers a Dutch question in Frisian 66
13 The association between level of Frisian and reactions to someone who
answers a Dutch question in Frisian 67
14 The association between L1 and the choice to give a ticket to a football game
to a Frisian or Dutch person 68
15 The association between home languages and the choice to give a ticket to a
football game to a Frisian or Dutch person 69
16 The association between place of birth and the choice to give a ticket to a
football game to a Frisian or Dutch person 69
17 The association between level of Frisian and the choice to give a ticket to a
football game to a Frisian or Dutch person 70
18 The association between home languages and reactions to a Frisian speaking
teacher in a Dutch school 71
19 The association between level of Frisian and reactions to a Frisian speaking
teacher in a Dutch school 71
xii 21 The association between home languages and going to a Frisian or Dutch
hairdresser 73
22 The association between level of Frisian and going to a Frisian or Dutch
hairdresser 74
23 The association between place of birth and going to a Frisian or Dutch
hairdresser 74
24 The association between the place where the participants were raised and
going to a Frisian or Dutch hairdresser 75
25 Summary of the significant background variables of the situations of
1 1. Introduction
Frisian is a minority language in the Netherlands, where mainly Dutch is spoken. Since 1985, policy documents have been written in order to determine the position of the Frisian language in Friesland and the Netherlands. At first, policy makers were mostly concerned with achieving equal rights for Frisian and establishing the status of the Frisian language. In 1970, Frisian gained the status of second official language of the Netherlands (cf. Van Ruijven and Ytsma 2008: 11). Gaining equal rights for Frisian and establishing the status of Frisian is still an important part of the Frisian language policy. For example, the Dutch government plans to include a paragraph about the Frisian language in the Dutch constitution at the moment (cf. Rijksoverheid 2011).
Besides focusing on the rights and status of Frisian, the past few years the focus has rather been on promoting the Frisian language. The title of the most recent policy report Beleidsplan Friese taal 2008-2010. Van recht naar praktijk. Fries in Fryslân: taal tussen mensen (‘Language Policy 2008-2010. From Law to Practice. Frisian in Friesland: Language among People’) illustrates this shift in policy making. Since using Frisian in several domains has been legally permitted, the Province of Friesland is now concerned with getting the language to actually be used more by its inhabitants. To achieve this, the Province formulated a number of goals in 2008, which described how Frisian should be used in several domains of the Frisian society. Education and parenting play the most important role in promoting the Frisian language. For example, one of the goals was that in 2010, half of the bilingual couples living in Friesland were informed about the advantages of bilingualism, so that they could make a deliberate choice to raise their children mono- or bilingually. Secondly, Frisian should play a bigger role in education (cf. Province of Friesland 2011a: 35ff). A prominent example of this is the project of the Trijetalige Skoalle (‘Trilingual School’). This project already started in 1997 at a number of elementary schools in Friesland. The idea of the project was to teach pupils systematically in Dutch, English and Frisian; 50% Dutch and 50% Frisian in grades 1 to 6 (4 to 10 year-olds), 40% Dutch, 40% Frisian and 20% English in grades 7 and 8 (10 to 12 year-olds), to improve the Frisian pupils’ capability of reaching the attainment goals for those languages (cf. Van Ruijven and Ytsma 2008: 8).
2 schools that participated in the project. An important conclusion is that focussing more on Frisian does not lead to worse results in Dutch. Interestingly, the evaluation also showed that the pupils who attend trilingual schools do not get better results in English, but are more confident when speaking English than the pupils attending the control schools (cf. Van Ruijven and Ytsma 2008: 52f.).
In the school year of 2005-2006 eight schools were certified as trilingual schools. Today, 41 schools participate in the network of trilingual schools (CEDIN 2011). Another 20% of the primary schools in Friesland are bilingual and use Frisian next to Dutch as a medium of instruction, however not as structurally as the trilingual schools. At the remaining schools, Frisian is only taught as a subject. In most of these cases, 30-45 minutes per week are reserved for Frisian lessons, which add up to 240 lessons in six years (cf. Gorter 2005: 62).
At this moment (2011) a new language policy is expected and most of the goals of the policy 2008-2010 should be attained. The use of the Frisian language has been encouraged and the number of trilingual schools is expanding. Has this paid off? The most recent quick scan of the Frisian language (Province of Friesland 2011b) revealed that about 95% of the people living in Friesland can speak Frisian, 74% are able to speak it, about 76% can read it and only 30% can write Frisian. In comparison to the quick scan of the state of the Frisian language in 2007, the year before the language policy 2008-2010 started, the situation of the Frisian language is nearly stable. Only a small increase in people who can write Frisian can be observed. In 2007, about 27% could write Frisian, in 2011, this is 30% (cf. Province of Friesland 2011b: 7). Following from that, we can conclude that the use of Frisian has hardly changed since 2007. Therefore, the main research question of the present study is what kinds of attitudes the inhabitants of Friesland hold towards the Frisian language and what they think about learning Frisian, in general and at school, after a period in which the use of Frisian was strongly stimulated by the Province of Friesland. Since more and more schools are becoming part of the project of the Trijetalige Skoalle, the second research question is whether the attitudes of parents of children attending a trilingual school are more positive than the attitudes of parents of children that attend monolingual schools. The final research question is whether different measurements lead to different results. People might have certain prejudices about a minority language such as Frisian which they probably do not like to show. It is interesting to examine if indirect measurements can uncover these prejudices.
4 2. Literature Review
2.1 Theorising Language Attitudes
This section covers the theoretical background concerning language attitudes. In section 2.1.1 a definition as well as a few important features of language attitudes are discussed. Section 2.1.2 is concerned with a number of approaches towards the study of language attitudes.
2.1.1 Towards a Definition of ‘Language Attitudes’
The field of study on attitudes originally comes from the field of social psychology (cf. Ebertowski 1978: 38). According to Fishbein and Ajzen (1975), attitudes consist of three ‘basic features: the notion that attitude is learned, that it predisposes action, and that such actions are consistently favourable or unfavourable toward the object’ (cf. Fishbein and Ajzen 1975: 6). Ebertowski (1978) concludes from this definition, that a ‘language attitude’ can be defined as a ‘learned predisposition from the language user to react consistently positively or negatively to a language (-variety, -form)’ (Author’s translation from Ebertwoski 1978: 39).1
Attitudes are learned in a way that they are based on a person’s personal experiences and social environment. Because of the fact that attitudes are based on past experiences they are ‘not momentary but relatively ‘enduring’’ (Agheyisi and Fishman 1970: 139) and relatively hard to change. Two processes play an important role in learning attitudes. The first one is ‘observational learning, which involves noticing the behaviour of other people and the consequences of that behaviour’ (Garrett 2010: 22). The second one is the process of ‘instrumental learning’ (Garrett 2010: 22), in which an individual gets familiar with the consequences that come with certain attitudes.
Attitudes can have two different functions. On the one hand, a favourable attitude towards a certain language can be input (in the form of motivation) to learn a specific language. One could for example think that a language sounds very beautiful and start learning the language for this reason. ‘In this sense, attitude is a predisposing factor, affecting the outcomes of education’ (Baker 1992: 12). On the other hand, bad education can negatively affect one’s attitude about a language. For example, one could have a horrible French teacher and hate French because of that. In that way, the attitude is the output (Baker 1992: 12f.).
1
5
Fig. 1: The three components attitudes are based on according to the multiple component definition. Source: Baker (1992): 13.
According to the ‘multiple component definition’, attitudes are not 'directly observable or measurable. Instead they are inferred from the way we react to particular stimuli’ (Rosenberg and Hovland 1960: 1). These reactions can be divided into three components: A cognitive, an affective and a conative one (cf. figure 1). The cognitive component concerns the beliefs an individual has about a particular group. These beliefs are based on the knowledge that one has about the object of the attitude. The affective side of language attitudes involves emotions that an individual has about people who speak a certain language. The conative aspect encloses the behavioural intentions of an individual towards the object of the attitude (cf. Agheyisi and Fishman 1970: 139; Ajzen 1988: 5ff.; Baker 1992: 12f.; Garrett 2010; Rosenberg and Hovland 1960: 1f.).
It seems logical to think that when one changes a person’s attitude, that person’s behaviour against the object of the attitude changes automatically and vice versa. However, this is not always the case. A prominent example of this is the study of LaPiere (2010)2 carried out in 1934. This study showed a discrepancy between peoples’ behavioural intentions and their actual behaviour. LaPiere visited 66 hotels and 184 restaurants together with Chinese people. In almost all establishments they were served without problems (except from one). Half a year later, LaPiere let representatives of the places they visited fill out a questionnaire, in which he asked if Chinese people were welcome in their restaurant or hotel. 91% of the hotels and 92% of the dining places gave a negative answer. (cf. Garrett 2010: 25f.; LaPiere 2010: 8 ff.). Consequently, the conative side of a attitude does not say anything about the actual behaviour, but only about how one wishes or thinks to behave (cf. Jonkman
2
6 1982: 12). Studies such as these indicate that people are not always aware of their attitudes. Some attitudes are held consciously (overt attitudes) and some are held subconsciously (covert attitudes). The difference between the reported behaviour and the actual behaviour in LaPiere’s study is a clear example of this. People can be heavily influenced by their environment and therefore act differently in certain situations from what they think they would do. Following from that, someone’s consciously held attitudes can differ from his subconsciously held attitudes. A solution to this problem could be to let someone else report on a person’s behavioural attitude. Attitudes can be inferred from three authorities. Firstly, the observer can investigate the reactions from the judging subject. Secondly, the subject can uncover his or her own judgements. Finally, information about the reactions can be obtained through reports made by friends and family of the subject (cf. table 1). Another solution could be to use different measurements to uncover overt (conscious) and covert (subconscious) attitudes. This will be further discussed in the next section.
Table 1: Source information about responses. Source: Ajzen 1988: 3.
7 2.1.2 Main Approaches towards Studying Language Attitudes
Language attitude research is important for a number of reasons. First of all, attitudes towards a certain language give an indication of ‘the status, value and importance’ (Baker 1992: 10) of that language. It can be said that they ‘provide a measure of the health of the language’ (Baker 1992: 9). This is especially the case for minority languages like Frisian, which are always under threat from a more dominant or larger language. In that sense, attitude measurements can function as ‘social indicators of changing beliefs and the chances of success in policy implementation’ (Baker 1992: 9). As we have seen in the previous chapter, it is important to make a distinction between overtly and covertly held attitudes. People do not always behave in the way they say or think they do. They can subconsciously hold different attitudes and therefore act in another way in real situations as what they say they do when they are reporting on their behaviour. We should keep this in mind, if we understand language attitudes as predictors of the success of implementing policies. Secondly, ‘attitude studies can also tell us about within-community and cross-community variation and cultural differences’ (Garrett, Coupland and Williams 2003: 14). Attitudes are after all learned behaviour and as a consequence they are partially ‘related to social-group membership’ (Garrett, Coupland and Williams 2003: 14).
The study of language attitudes is usually divided into two main approaches; direct and indirect methods. Direct approaches are meant to measure the overtly held attitudes, the ones that people are conscious about. The indirect methods work slightly more undisclosed and are supposed to measure the covertly held attitudes. People are not directly conscious of having these attitudes (cf. Garrett 2003: 24ff.; Agheyisi and Fishman 1970: 145).
In the present study, the division of the main approaches towards the study of language attitudes in direct and indirect measurements is followed. In section 2.1.2.1 a few direct measurements are discussed. Section 2.1.2.2 covers some indirect methods.
2.1.2.1 Direct Approaches
8 First of all, in questionnaires overtly held attitudes can be measured through open and closed questions. The advantage of open questions is that the participants are able to give their own opinions and do not have to choose an answer out of multiple given answers. However, one of the disadvantages can be that people do not make the effort to write the answer down. Open questioning can form a problem for the researcher as well, because the answers are hard to score (cf. Agheyisi and Fishman 1970: 147).
Consequently, the big advantage of closed questions for the researcher is that the answers are usually a lot easier to score compared to the answers of open questions. The informants can also benefit from closed questioning, since these questions are easier to fill out. A disadvantage of closed questions however is that if they are too easily built up, the participants can answer the questions automatically, without considering if the answer they choose really is the correct answer for them (cf. Agheyisi and Fishman 1970: 148).
Likert-type scaling is a frequently used scaling technique in closed questioning.3 With this method, informants have to express to which extent they agree or disagree with a number of statements through a scale. Ideally, the number of negative statements is the same as the number of positive statements (cf. Garrett, Coupland and Williams 2003: 40). Mostly, a Likert-scale includes five points: strongly agree, slightly agree, undecided, slightly disagree, strongly disagree (Sommer 2001: 155). The midpoint of this scale is often seen as problematic. Usually, the Likert-scale is seen as reliable to measure the ‘intensity of attitudes’ (Garrett, Coupland and Williams 2003: 41). The problem with the midpoint of the scale is that the option ‘undecided’ does not really express an intense commitment (cf. Garrett, Coupland and Williams 2003: 41 and Oppenheim 1992: 200). Other problems of the Likert-scale concern the scoring of the items. Although the Likert-scale is an ordinal scale, it is often falsely treated as an interval scale (cf. Edmondson 2005: 127).
Interviews are another way of asking people direct questions about their attitudes. They have a great advantage, which is the immediate contact with the interviewer. Because of that, participants usually take the questions more seriously. Interviews enable the researcher to prevent boredom and irritation of the participant. However, this can form a problem as well, if the researcher sends out the wrong signals. For example, the risk of biasing exists, i.e. the informants can give socially desired answers (‘social desirability bias’) or answer every question positively to favour the researcher (‘acquiescence bias’) (cf. Agheyisi and Fisman 1970: 149; Garrett, Coupland and Williams 2003: 28). The presence of the researcher is a
3
9 general problem of direct measurements, also for questionnaires. Certain characteristics of the researcher can provoke reactions of the participants. Labov (2006) refers to this as the ‘Observer’s Paradox’. The ‘Observer’s Paradox’ represents a situation in which participants can be influenced by the presence of a researcher (the observer) and answer in a different way as compared to when the researcher is not in the same room observing them (cf. Garrett, Coupland and Williams 2003: 29f.; Labov 2006: 86)
Finally, it is important to point out that most of the mentioned problems are common difficulties of carrying out research. Nevertheless, the influence of some of these problems can be decreased by using indirect methods to measure language attitudes. The next section focuses on (a selection of) these methods.
2.1.2.2 Indirect Approaches
The indirect method to measure language attitudes that is most often used is the Matched Guise Technique (MGT). It was developed by Lambert et al. (1972). He investigated the language attitudes towards French and English held by French- and English-speaking people living in Montreal. Lambert et al. developed the technique because he was ‘interested in reactions that are attributable primarily to the language itself’ (Lambert, Hodgson, Gardner and Fillenbaum 1972: 81). Therefore, Lambert et al. let their participants hear the same text twice, once recorded in French and once recorded in English. Both fragments were recorded by the same bilingual speaker. The subjects were asked to evaluate the speakers with regard to several personality traits. They did not know that they were hearing some of the voices twice. That way, the covertly held attitudes towards both languages could be measured (cf. Lambert, Hodgson, Gardner and Fillenbaum 1972: 80f.).
In total Lambert used ten fragments. Eight of those were recorded by four bilingual speakers. The two remaining voices were recorded by one English and one French speaking person. These fragments did not form a MGT-pair, but were used as ‘filler voices’ at the beginning of the test, to let the participants get used to the form of questioning (cf. Lambert and Taylor 1972: 80). Lambert’s study led to interesting results. Both English- and speaking participants evaluated the English guises higher. It is striking that the French-Canadian subjects rated the French guises even lower than the English-speaking subjects did (cf. Lambert, Hodgson, Gardner and Fillenbaum 1972: 86ff.).
10 on synaesthesia (perception through a mixture of senses). Osgood (1957) adapted the scale to measure meaning. The idea of the semantic differential is to let subjects judge something on a scale. This scale consists of a bipolar pair, for example ugly – pretty. The scale contains a number of points, which enable the informants to point out the intensity of their judgement (cf. Garrett 2010: 55).
Basically, three things are needed for a semantic differential scale: stimuli in the form of a concept (the thing that is judged), responses (the subject’s judgements) and scales (the bipolar pairs). It is important to take relevant scales that are semantically stable. Semantically stable means that the used bipolar pair should call forth as few connotations (emotional associations) as possible. The concept small – large can for example be used for concepts like ELEPHANT and INSECT without any problems, but if MY AUNT is large or small can be problematic, because that is sometimes more a question of personal associations (cf. Osgood 1957: 77ff.).
There are two forms to present a semantic differential scale. Both times, the participants mark the point which suits their judgements best with a cross. The outer points represent the most extreme judgements, the points between the middle and the extremes, represent the less extreme judgements. If a participant does not have an opinion, he or she can mark the point in the middle.
LADY rough ___:___:___:___:___:___:___ smooth ME fair ___:___:___:___:___:___:___ unfair, etc.
Fig. 2: Semantic differential scale, Form I. Source: Osgood et al. 1957: 81.
In the first form the scale is directly behind the concept. The advantage of this form is that the participants constantly have to judge another concept, so that they cannot answer automatically.
LADY
rough ___:___:___:___:___:___:___ smooth fair ___: ___:___:___:___:___:___ unfair
active ___:___:___:___:___:___:___ passive etc.
11 In the second form different scales appear under one concept. This form is easier to fill out for the informants. There is no evidence that the two different forms lead to different results (cf. Osgood 1957: 81f.).
Most semantic differential scales have seven points. Research of Osgood (1957) has shown that this amount of points is the most suitable for a semantic differential scale. On scales with more points, the most outer points are hardly used. When the scales have less than seven points, participants claimed to not have enough points to express their judgement (cf. Osgood 1957: 85).
In MGT-studies, the bipolar pairs are often grouped into different dimensions. Researchers often have different opinions about what these dimensions should look like. Zahn and Hopper (1985), for instance, used pairs from the dimensions ‘superiority’ (such as strong – weak), ‘attractiveness’ (such as nice – awful) and ‘dynamism’ (such as energetic – lazy) (cf. Zahn and Hopper 1985: 117f.). Lambert (1972) on the other hand uses the dimensions ‘competence’ (like intelligent – not intelligent), ‘personal integrity’ (like reliable – unreliable) and ‘social attractiveness’ (like friendly – unfriendly). Not only the choice for dimensions, but also the determination of which traits belong to which dimension, is sometimes debatable. As mentioned above, Lambert sees ‘friendliness’ as part of the attractiveness dimension, while in other projects (for example Ytsma 1990) it is seen as part of the solidarity dimension.
Something that should be taken into account when using semantic differential scales is that a high score on a personality trait does not automatically lead to a positive attitude. If someone rates a certain concept high on a scale, it does not always imply that his attitude towards the concept is positive. We can think someone is rich, but that does not automatically mean that we hold a positive attitude towards that person. For example, we could dislike rich people because we are poor ourselves (cf. Garrett 2010: 71). If a high rating on a certain scale can be seen as positive can depend on the judgements of concepts and associations with the personality traits used in the scale.
12 have different opinions about what bad grammar or standard language is (Garrett 2010: 58). Secondly, the recording of the fragments might cause problems. Mostly, the speakers are asked to read aloud a written text. This may sound slightly formal. Therefore it cannot be assumed ‘that more spontaneous speech will be evaluated in the same way’ (Garrett 2010: 58). The most frequently uttered criticism of the MGT is that it assumes that ‘each population or sub-population is characterized or identifiable by a single language variety’ (Agheyisi and Fishman 1970: 146). Because of the problems that are linked to the MGT, adaptations of the technique as well as other indirect methods are used in a number of language attitude studies. A selection of these investigations will be discussed here.
Instead of making recordings with bilingual speakers, in some studies different (monolingual) speakers are asked to record the speech-fragments. This is known as the ‘Verbal Guise Technique’. This technique is mostly used when a researcher is not able to find multilingual speakers (cf. Garret, Coupland and Williams 2003: 53).
Another example of an adapted version of the MGT is Kimple’s (1968) study into language varieties in conversations. He called his method the ‘Mirror Image Techniqe’ (MIT). In this method, speakers recorded two different conversations in multiple languages. The voices were divided over the following four versions of each of the two conversations (cf. Agheyisi and Fishman 1970: 147):
i. One, in which all speakers spoke the same language A;
ii. Another version, in which some of the speakers spoke language A in some roles, and B in other roles;
iii. Another conversation, in which all speakers spoke the same language B;
iv. And a last one, in which the languages in the particular roles of the second conversation were switched (mirror image).
13 2.2 Previous Studies into Language Attitudes
This section provides an overview of earlier investigations into language attitudes. On the one hand, a number of projects that focus on attitudes towards the Frisian language will be discussed. On the other hand a small insight in investigations into language attitudes towards other minority languages is given, to be able to examine if certain aspects of language attitudes towards Frisian are general for language attitudes towards minority languages.
2.2.1 Studies into Attitudes towards Frisian
The discussed projects about attitudes towards Frisian in the following sections are divided into three categories: Projects in which direct methods are used (section 2.2.1.1), projects, in which indirect methods are used (section 2.2.1.2) and language surveys (section 2.2.1.3). In Friesland, a number of language surveys have been held to examine what the state of the language is. In these language surveys aspects such as people’s competence in Frisian and the use of the language have been investigated. Language attitudes are always an important part of these surveys. Therefore, a separate section is reserved for discussing those surveys.
2.2.1.1 Direct Methods
Over the years, a few direct studies, in which language attitudes towards Frisian were directly measured, have been carried out. Five of those are discussed in this section.
Smith (1979) carried out one of the first studies into language attitudes towards Frisian. His study focussed on language behaviour and language attitudes in Terherne, a bilingual community near the town Sneek, in the south-western part of Friesland. Because of the fact that Terherne is located next to the Lake of Sneek, a large number of people own a holiday home in Terherne and live there every summer (Smith 1978: 152ff.; Smith 1979: 7ff.).
14 preservation of the Frisian language than their Dutch and summer fellow-villagers (cf. Smith 1978: 133ff.; Smith 1979: 34ff.).
Another project was carried out by Gorter in 1985. He investigated the language situation in the Frisian municipality It Hearrenfean. About 593 inhabitants of this region participated in the project. Gorter conducted his questionnaires together with students of administrative schools in Friesland. Gorter measured the attitudes of the inhabitants of It Hearrenfean while at the same time measuring the attitudes of the students towards Frisian. For his investigation, Gorter used 14 Likert-scales (Gorter 1985: 87ff.; Gorter en Ytsma 1988: 66).
The results showed that language background was an important variable: The Frisian speaking inhabitants of It Hearrenfean held a more positive attitude than the participants who did not speak Frisian. Furthermore, the students of the administrative schools had a more negative attitude than the population of It Hearrenfean. Gorter explains that by referring to the level of education and claims that people with a higher level of education have a more negative attitude towards Frisian than people who are less educated (cf. Gorter and Ytsma 1988: 67f.).
In 1990, Jonkman investigated the attitudes towards Dutch, Frisian and the Ljouwert Vernacular (Town Frisian). In contrast with his other language attitude investigations (cf. section 2.2.1.2). Jonkman did not use the Matched Guise Technique, but he used ‘conceptual guises’, i.e. he asked his participants to evaluate the concepts of certain languages (in this study Frisian, Dutch and the Ljouwert Vernacular) on a number of personality traits. 282 inhabitants of Leeuwarden (Ljouwert), the capital of Friesland, participated. 74 of them had Frisian as first language, 153 Dutch and 55 spoke the Ljouwert vernacular. It turned out that Dutch scored the highest on the status traits. The participants evaluated Frisian the highest on the solidarity traits. In total, the participants had the most positive attitude towards Dutch, followed by Frisian. The Ljouwert vernacular scored the lowest (cf. Jonkman 1990: 16ff.).
15 knowledge of Frisian, the more positive the attitudes towards Frisian are (cf. Spelberg and Postma 1995: 32ff.).
A last project worth mentioning in this context is the investigation done by Bangma (2009). Bangma examined the attitudes held by pupils in 7th and 8th grade (9- to 13-year-olds) of primary schools at the Frisian country side. In the scope of the project the Trijetalige Skoalle (‘Trilingual School’), that was carried out by the Fryske Akademy and the Province of Friesland (cf. section 1. Introduction), she compared the results of pupils attending a trilingual school with the results of pupils that attended a mono- or bilingual school. The project had 198 respondents. 30% of the participants attended a monolingual school, 34% a bilingual one and 36% a trilingual one (cf. Bangma 2009: 27ff.). The results of Bangma’s investigation showed that the pupils of multilingual schools have a more positive attitude towards the Frisian language. This also holds for children with Frisian as first language. The results also showed that the educational system does not affect the level of self-confidence of the children. These results are in contradiction with the results of Van Ruijven and Ytsma’s (2008) report on the effects of the trilingual school. They found that pupils attending trilingual schools are more confident when speaking English than the pupils attending mono- or bilingual schools.
2.2.1.2 Indirect Methods
Considering indirect methods, which are meant to measure people’s covertly held attitudes, only a few studies have been carried out investigating language attitudes towards Frisian using the Matched Guise Technique (MGT). Van der Plank (1980) was the first one to investigate language attitudes towards the Frisian language through an MGT-research. He tried to investigate how people in Friesland react to Dutch and Frisian (cf. van der Plank 1980: 23). He used a conversation between a social worker and a client that was recorded in Frisian and Dutch by the same speakers.
Thirty respondents participated in the project. 15 of them were students with a Frisian or a Dutch language background. The other half consisted of ‘normal’ people with Frisian as a first language. Considering the fact that the questionnaire consisted of two fragments only, the risk that people would notice that they were hearing the same speakers twice was very present. For that reason, Van der Plank decided to let half of the sample hear the Dutch fragment and the other half the fragment in Frisian (cf. van der Plank 1980: 26f.).
16 outgoing – shy) and ‘authority’ (for example self-confident – insecure). The positive and negative words appeared in turns on the left and right side of the scale to avoid automatic judging (cf. van der Plank 1980: 28).
The outcomes showed that the subjects with a language background in Frisian rated the Frisian speakers higher than the Dutch speakers. That was the same for both students and ‘normal’ participants. The Dutch informants showed no differences in rating the Frisian and Dutch speakers (cf. van der Plank 1980: 37).
Jonkman (1982) was the second person to carry out an investigation of language attitudes towards Frisian and a selection of Frisian dialects. This was his first MGT-project. He examined the language attitudes held by students born in Friesland, but living in Groningen at the time of the investigation. He investigated their attitudes towards standard-Frisian, standard-Dutch, ‘neutral’ Frisian and the two dialects Kleifries (a dialect spoken in the northern part of Friesland) and Zuidwesthoeks (a dialect spoken in the south-western part of Friesland) (cf. Jonkman 1982: 23ff.). Jonkman used route descriptions (Paadwizers) to a number of places in Leeuwarden, the capital of Friesland. In total, the project contained six fragments (all with a different text); standard Frisian was used in two fragments, the other four varieties were each used in one fragment.
Speaker: Fragment 1: Fragment 2:
Speaker 1 Bad Frisian Neutral Frisian Speaker 2 Kleifries (dialect) Standard Frisian Speaker 3 Standard Dutch Standard Frisian
Table 2: The speakers and the fragments they recorded. Source: Jonkman 1982: 30ff.
The fragments were recorded by three bilingual speakers. One speaker covered a fragment in ‘bad’ Frisian (not a real variety, but rather ungrammatical Frisian) and one in ‘neutral’ Frisian. The second speaker recorded one fragment in Kleifries and one in standard Frisian. The last speaker took care of the standard Dutch fragment and the other standard Frisian fragment (cf. Table 2). Jonkman recorded the fragments along the side of the road, to let the fragments seem as ‘real’ as possible (cf. Jonkman 1982: 28ff). This is a strange choice, considering the fact that each fragment has other background noises that way. This may have affected the judgements of the participants.
17 from the dimensions ‘competence’ (such as intelligent – stupid), ‘attractiveness’ (such as friendly – unfriendly) and ‘integrity’ (such as fair – unfair) on a seven-point-scale (cf. Jonkman 1982: 38ff.).
Jonkman’s results showed that the judgements of the Frisian subjects did not show a lot of variation. This was the case for the ratings of the Frisian dialect-fragments as well as the standard Dutch fragments. Although Jonkman did not expect to find many differences in the ratings of the Dutch participants, the results of those showed more differences than the outcomes of the Frisian subjects. The Dutch subjects did make a distinction between the various Frisian dialects. They judged the Frisian dialect-fragments lower than the recordings in standard speech (cf. Jonkman 1982: 44ff.).
In 1985 Jonkman carried out another investigation using the MGT. This time he investigated the status of the Frisian and Dutch language. In addition one of his research questions was what the consequences of certain language behaviour (for example answering in Frisian, when someone is asking a question in Dutch) are (cf. Jonkman 1985: 37f.). For his second Matched-Guise-project Jonkman used route descriptions again, this time in the form of three conversations, in which the first speaker gave the second speaker a route description to certain places. In the first fragment, in which a person wanted to go to the cinema, both speakers spoke Dutch. In the second fragment, in which a person wanted to go to the library, both speakers spoke Frisian. In the last fragment, a person asked where the city hall was in Dutch and the second speaker answered in Frisian. The three conversations were all recorded twice. The first time, the three different roles were recorded by three different speakers. The second time, the three different roles were recorded by one and the same bilingual speaker. Following from that, the last series of fragments were the actual MGT-fragments. Because the same person recorded different roles, the roles could not influence the evaluation of the language. That is why Jonkman called these recordings Skaadwizers (‘shadow guides’). The other series of fragments were recorded by other speakers and were put between the three Skaadwizer-fragments to reduce the risk that participants would notice that they were rating the same speaker (cf. Jonkman 1985: 40ff.).
18 The results of the Skaadwizer-investigation showed that the Frisian participants rated Frisian higher on the solidarity traits. The Dutch participants rated Frisian and Dutch the same on the solidarity traits. For the judgements on the status traits there were no differences found between the Dutch and Frisian fragments. With regard to language behaviour, both groups disapproved of the conversations, in which the second speaker answered in Frisian, while the first one was asking a question in Dutch (cf. Jonkman 1985: 61ff.).
Ytsma (1990) was the first one to carry out a MGT-project in primary schools. He investigated the language attitudes towards Frisian held by pupils of primary schools in the Frisian country side. He considered primary language, gender and grade (5th grade and 8th grade) as background variables. He used four fragments (two in Frisian, two in Dutch), recorded by two male bilingual speakers. The speakers read out aloud a text about the weather (cf. Ytsma 1990: 172f.). In total, 156 pupils (82 boys and 60 girls) participated in the investigation. 46 of them had Dutch as native language, 96 had a language background in Frisian. 69 pupils attended the 5th grade and 73 the 8th grade. The pupils had to rate the fragments on eleven personality traits. In addition, they had to answer the question if they would want to have the speaker as a teacher, father or neighbour and they had to point out, what kind of career they thought the speaker would have. Apart from the MGT, the children had to fill out 10 Likert-items on Frisian as a language at school, using Frisian as an oral language, the symbolic value of Frisian, Frisian on television and their own ethnicity. Ytsma included the Likert-scales, so that he could make a comparison between direct and indirect measurements (cf. Ytsma 1990: 173f.).
The results of the Likert-items showed that only the language background made a difference. The Dutch children had a negative attitude towards Frisian, the Frisian pupils held a neutral attitude towards their own language. A striking result was that the Frisian pupils had a more negative attitude towards Frisian in the 8th grade than they had in the 5th (cf. Ytsma 1990: 174f.). It seems that those Frisian speaking pupils that hold a neutral attitude towards their own language in the 5th grade have taken over the negative attitude that their Dutch speaking class mates hold towards Frisian by the time they attend 8th grade.
19 Ytsma found two correlations, which were both very weak. On the one hand he found that when someone rated Frisian lower on solidarity and Dutch higher on status in a MGT-part, he or she held a more negative overt attitude towards Frisian as well. Both dimensions of the MGT correlated as well; a low solidarity ranking for Frisian went together with a high status ranking for Dutch (cf. Ytsma 1990: 179).
2.2.1.3 Language Surveys
Aside from the investigations discussed in the previous sections, there have been a few language surveys carried out in Friesland. The study of language attitudes played a role in these. The methods and results of those surveys will be discussed here.
In 1969, Pietersen carried out the first language survey in Friesland. The part of the survey concerned with attitudes towards Frisian contained questions about language behaviour, in the form of reading Frisian books and listening to Frisian radio, and questions about what Pietersen called language ideology (the system of norms and expectations that one has about a language, on which one justifies the use of a language in certain situations and the feeling one has for that language) in addition, there were a number of questions about the attractiveness and status of Frisian (cf. Pietersen 1969: 91).
Pietersen’s questionnaire contained 76 closed questions, mostly containing Likert-scales (with five or three points). 800 people (12 years and older) living in Friesland participated in the project. Since people of eleven municipalities in Friesland took part in the investigation, this group formed a representative sample for the population of Friesland (the Frisian isles not included) (cf. Pietersen 1969: 10).
The results concerning the language behaviour showed that 53% of the people living in Friesland read a Frisian book sometimes. However, 73% of those still read Dutch books more often. About half of the people living in Friesland sometimes listen to Frisian radio-programs (cf. Pietersen 1969: 130). Considering the language ideology, the following conclusions were made:
i. Elderly people showed more affinity for Frisian than younger people;
20 iii. People who lived on the countryside had a greater feeling for Frisian than the people
who lived in the urban regions;
iv. Religious people (mostly Christian) seemed to have a greater connection with Frisian than people, who never visited a church;
v. Farmers had the greatest affinity with Frisian, members of the middle and higher class the weakest;
vi. Income did not affect the amount of affinity with Frisian;
vii. Education did; the higher the level of education, the weaker the connection with Frisian (for all conclusions, cf. Pietersen 1969: 133f.).
Furthermore, the survey contained a question in which the participants were asked to make a rank order of seven languages (Frisian, Dutch, English, German, French and two Dutch dialects: Limburgs and Gronings), with the most beautiful language on top. Frisian was ranked as the most beautiful language, followed by Dutch. This was related to the rank order of population groups: Frisians were ranked the highest, followed by the Dutch. Although the participants claimed Dutch to have more power than Frisian, the results did not confirm that Dutch had more status than Frisian (cf. Pietersen 1969: 138ff.).
Finally, the informants were asked in which language they address people in certain domains. The results showed that 93% of the Frisians claimed to speak Frisian if they opened the door after someone rang their doorbell. In the supermarket, 85% of the Frisians would speak Frisian. Only 58% of the Frisians would address a person in a higher function (doctor, director etc.) in Frisian (cf. Pietersen 1969: 59ff.).
Based on Pietersen’s project, about ten years later, Gorter and Jonkman (1984) carried out another large language survey in Friesland. 1126 people participated in this investigation (cf. Gorter and Jonkman 1984: 264). In a section about language ideology, the conclusions were the same as in Pietersen’s survey (cf. i-vii above). However, Gorter and Jonkman (1984) found that language background played the largest role in the forming of a language ideology. The results showed as well, that people (both Frisian and not Frisian) living in areas where Frisian is the main spoken language held more positive attitudes than people who lived in areas where most people spoke Dutch (cf. Gorter and Jonkman 1984: 425ff.).
21 occupations with a lower status. The results showed that people in lower occupations are more often addressed in Frisian than people in higher occupations. The outcomes also showed that participants with a higher social economic status were less positive towards Frisian than those with a lower social economic status (cf. Gorter and Jonkman 1984: 181ff.).
A final interesting result brought the question about Frisian as a language in school. 16% of the participants were against the law that makes the teaching of Frisian in Frisian primary schools obligatory. These were mostly people that moved from other Dutch provinces to Friesland, people who do not have Frisian as their first language, or elderly people (cf. Gorter and Jonkman 1984: 428f.).
In 1995, Gorter and Jonkman carried out another language survey, from which they compared the results to the results of the surveys from 1969 an 1984. This time, 1368 people participated (cf. Gorter and Jonkman 1995: 5). The results were approximately the same as in 1969 and 1984. The language attitudes towards Frisian did not change and language background was still the most important background variable (cf. Gorter and Jonkman 1995: 32ff.). The results for the language behaviour hardly changed as well. Only the use of Frisian in public places showed a small decrease (cf. Gorter and Jonkman 1995: 16ff.).
2.2.2 Research into Attitudes towards Other Minority Languages
In this section an overview of research projects into attitudes towards a number of other minority languages is made to investigate if certain tendencies that were noticeable through the results of the Frisian projects can be seen as general for minority languages.
First of all, a few investigations on attitudes towards Welsh are summarized. Welsh is a minority language spoken in the south-western part of the United Kingdom. Since 1901, the number of people that speak Welsh has been decreasing. Since 1991, however, this number is stabilizing, probably because more people are realizing what the consequences of the extinction of the Welsh language would be, and because of the growth of the Welsh media (cf. Laugharne 2007: 208f.).
22 1970). Due to this fact, a lot of changes in language policies have taken place in Friesland as well as Catalonia (cf. Woolard and Gahng 1990: 314ff.).4
2.2.2.1 Welsh
Wales is a region in the south-western part of the United Kingdom. About 5.2 million people live in Wales. Today, 20% of these people speak Welsh, a language that is part of the Celtic language family. The number of Welsh speakers has been decreasing for a long time. For example, in 1901 50% of the Welshmen were able to speak Welsh. Following from that, this number decreased with 30% in one century. On the other hand, this number has been stabilizing since 1991, probably because growing sense of the consequences of the extinction of the language and the increase of Welsh media. This situation offers a lot of opportunities for the study of language attitudes (cf. Laugharne 2007: 208ff.).
In 1973, Bourhis, Giles and Tajfel carried out a Matched Guise project to investigate how Welsh people evaluated Welshmen that spoke different languages. The 63 participants were divided into three groups:
i. ‘Welshmen who prefer to use their native tongue,
ii. Welshmen who cannot speak [Welsh] but who nevertheless have traces of a Welsh accent in their speech,
iii. Welshmen who cannot speak [Welsh] and whose English is characterized by an RP5 -style of speech’ (Bourhis, Giles and Tajfel 1973: 451)
Two male bilinguals recorded the same text three times; once in a South-Welsh dialect, once in English with a South-Welsh accent and once in Standard English. The speakers read a neutral text consisting of 290 words in 90 seconds. In addition, four filler voices were used (cf. Bourhis, Giles and Tajfel 1973: 452).
The results showed that the level of Welsh language background did not play a role in forming attitudes. All groups rated the South-Welsh dialect-fragments higher than the recordings in Standard English (cf. Bourhis, Giles and Tajfel 1973: 457).
4
Interestingly, Welsh and Catalan are often linked to Frisian, cf. for example Ytsma, Viladot and Giles 1994; Huguet 2007 and the PowerPoint presentation from Hilton, Gooskens, Schüppert, Van Bezooijen and Van Heuven at the Dei fan de Fryske Taalkunde from the Fryske Akademy in October 2010: Attitudes towards Frisian – in the Netherlands and Beyond.
5
23 In the same year, Sharp carried out a project about language attitudes of secondary school pupils towards Welsh and English. 4015 pupils participated in the project. They had to fill out a questionnaire and scale conceptual guises of English and Welsh on a semantic differential scale (cf. Sharp 1973: 158).
The main outcomes were as follows: The older the pupils were, the more positive they were towards English and the more negative attitudes they held towards Welsh. It was striking that English speaking pupils attending bilingual schools were more positive to Welsh than pupils that were actually Welsh speaking (cf. Sharp 1973: 155f.). The results from the semantic differential scales showed some interesting results as well. English was rated high on all scales, while Welsh only scored high on the integrity traits, such as friendly and warm (cf. Sharp 1973: 314ff.).
In 2007, Laugharne carried out another research project about language attitudes towards Welsh and English in the scope of the project ‘Multilingualism in European Bilingual Contexts’. 203 trainee teacher students were tested. The sample was slightly unequal. 46% of the students had English as their native language, 25% spoke Welsh as mother tongue, 26% considered themselves native speakers in both languages and 3% claimed to have another first language (cf. Laugharne 2007: 217).
In general, participants were more favourable towards Welsh than they were to English. The results of Laugharne’s project showed that the first language of the respondents played an important role in the forming of their attitudes. The participants with Welsh as a first language (bilinguals included) held a more positive attitude towards Welsh than the other informants. Town size was also an important background variable. Respondents that came from towns with less than 100.000 inhabitants held more positive attitudes towards Welsh than informants from larger cities (cf. Laugharne 2007: 219ff.).
24 outcomes of Laugharne’s (2007) project show that people who come from smaller towns hold more positive attitudes towards Welsh than people who live in larger cities. Similar results were found in Friesland. The language surveys of Pietersen (1969), and Gorter and Jonkman (1984, 1995) point out that people who live in the Frisian countryside hold more positive attitudes towards Frisian than the inhabitants of the Frisian cities.
2.2.2.2 Catalan
During the regime of the Spanish dictator Franco, the minority language Catalan, spoken in Catalonia, was severely suppressed. After Franco’s death in 1975, the language situation in Catalonia changed a lot. Catalonia became politically autonomous and policies to preserve Catalan were formed. The great amount of Spanish-speaking immigrants moving to Catalonia from South-Spain slowed down this process. In 1983, however, the official language status of Catalan was established. Since then it is allowed to use Catalan as a language in government and administration, a Catalan television channel has been set up and it is obligatory to teach at least one subject in Catalan at school (cf. Woolard and Gahng 1990: 314ff.).
Because of the changes in language policies, Catalonia forms an interesting region for language attitude research. Woolard and Gahng (1990) investigated changes in language attitudes towards Catalan and Spanish between the period five years after the end of Franco’s regime and 1987, when the moment that Catalan had gained the status of an official language (1983) was a few years ago. Their first measurement took place in Barcelona in 1980, a few years after the end of the regime of Franco, when Catalan just got the status of an official language. 240 secondary school students participated. As method, the Matched Guise Technique was used. Four young women read a one-minute-long text with an academic subject, once in Spanish and once in Catalan. Together they recorded eight fragments. The respondents had to rate the fragments on solidarity and status traits (cf. Woolard and Gahng 1990: 317f.).
25 speakers of Catalan were likely to be discouraged rather than recruited in informal interaction’ (Woolard and Gahng 1990: 315).
The second project was carried out in 1987. The same method and fragments as in 1980 were used. This time, 276 secondary school students participated. Their language background differed slightly from that one of the respondents in 1980 (cf. figure 4). The number of participants that had Spanish as first language was the same. In 1987 there were less respondents that spoke Catalan as mother language, but the amount of informants that indicated both languages as their home language was larger than in 1980 (cf. Woolard and Gahng 1990: 319).
Fig. 4: The home language of the participants in 1980 and 1987. Castilian is another name for Spanish. Source: Woolard and Gahng 1990: 319.
The results of the second measurement showed, that the difference between the solidarity and status ratings of the Catalan fragments had decreased. The judgements on the status of Catalan stayed the same. The solidarity ratings, on the other hand, were more positive than in 1980. This was the case for attitudes towards the use of Catalan by Spanish native speakers as well (cf. Woolard and Gahng 1990: 320ff.).
26 In total, 257 school students (13- and 14-year-olds) filled out a questionnaire: 94 from Catalonia and 163 from Aragon. The results showed that the respondents from Catalonia were more positive towards Catalan than the participants from Aragon. Participants that took classes in Catalan had a more favourable attitude towards that language than the respondents that did not take classes in Catalan. For the results of the Catalan respondents, the home language did not affect the attitudes towards Catalan, but did play a role for the attitudes towards Spanish; bilingual informants of Catalonia were more positive towards Spanish. For the results of the respondents living in Aragon, it was exactly the other way around. The home language did not affect the attitudes of the participants of Aragon towards Spanish, but Catalan speakers from Aragon were more positive towards Catalan than the subjects from Aragon that did not speak Catalan. Social professional status did not affect the outcomes (cf. Huguet and Llurda 2001: 318ff.).
In 2007, Huguet carried out another language attitude study in the scope of the project ‘Multilingualism in European Bilingual Contexts’.6 Through a direct questionnaire, he investigated the language attitudes of 309 students from the universities of Girona and Lleida towards Catalan, Spanish and English. The distribution of the participants was very spread out. 88% of the respondents were females and only 12% were males. 64% of the participants spoke Catalan as first language, 22,5% Spanish, 13% had both languages as mother tongues and 0,5 % had another native language (cf. Huguet 2007: 26f.).
In total, the participants were most positive towards Catalan. Spanish and English were judged the same. The first language of the informants affected the results insofar that the respondents with Catalan as native language held a more favourable attitude towards Catalan than those with other language backgrounds. The ones that had education in Catalan were more positive towards Catalan than the participants that did not. Another variable that affected the outcomes was the size of the city the participants lived in. Respondents from cities with more than 100.000 residents held more positive attitudes towards Spanish than the informants from smaller cities. Social economic status did not play a role for the results (cf. Huguet 2007: 27ff.).
Summarizing, just as for Welsh, there can be found similarities between the attitudes towards Catalan and the attitudes towards Frisian. For example, people who speak Catalan as a first language hold more positive attitudes towards Catalan than people who have Spanish as a native language. This also counts for Frisian; Frisian speaking people hold more positive
6
27 attitudes towards Frisian than Dutch speaking people (cf. cf. Smith 1979; Van der Plank 1980; Gorter 1985; Spelberg and Postma 1995 and Ytsma 1990). Although there are similarities between the attitudes towards Frisian and the attitudes towards Catalan, there are differences as well. For instance, in the investigation of Woolard and Gahng (1990) Catalan scores higher on status traits than Spanish. None of the investigations discussed in this literature review lead to a similar result for Frisian.
The comparisons with attitudes towards Catalan and Welsh can be helpful to understand the attitudes towards Frisian in a broader perspective. Some tendencies seem to be general for minority languages, like the fact that people who speak a minority language as a first language often hold more positive attitudes towards that minority language than people who speak the majority language of a region. Other tendencies are more specific for a certain minority language. The fact that some people attribute a higher status to Catalan than to Spanish, which probably has to do with the ‘superior economic position of Catalans’ (Woolard and Gahng 1990: 351), can be seen as an example of this.
2.3 The Present Study
This literature review discussed the theory of language attitudes and earlier investigations into attitudes towards Frisian. Based on the results of these earlier investigations, the research questions and hypotheses for the present study can be formulated in this section. First, the outcomes of the previous studies into language attitudes discussed in the literature review are summarized in the table below. The table provides a quick overview of previous investigations and the background variables that played an important role in relationship to the attitudes towards Frisian measured in the described investigations. The hypotheses for the present study are based on those outcomes.
Type of Investigation
Sample Results
Frisian – Direct methods
Smith (1979) Frisian and Dutch inhabitants and summer villagers of Terherne
1. All three groups hold positive attitudes towards Frisian 2. Frisians speaking people are more positive about the
preservation of the Frisian language than Dutch speaking people
Gorter (1985) Inhabitants of It Hearrenfean and students of administration schools
1. Frisian speaking people hold a more positive attitude towards Frisian than Dutch speaking people
