Loading...

The Social Status of Languages in Finland and Lithuania

A Plurimethodological Empirical Survey on Language Climate Change

by Stephan Kessler (Volume editor) Marko Pantermöller (Volume editor)
Edited Collection 282 Pages
Open Access

Table Of Content


←6 | 7→

Amei Koll-Stobbe

Series editor’s short introduction

This is the eleventh volume in my book series on Language Competence and Language Awareness in Europe. I am happy that this volume highlights the ideologically and sociologically loaded topic area of language policy. The Baltic Scholar Kessler and the Finno-Ugrian Philologist Pantermöller study language contacts and conflicts within and across the borders of sparsely populated national cultures in Scandinavia (Sweden and Finland), and the Baltics (Lithuania, Estonia). These European Union member states who, according to Eurostat (May 2019), are the member states at the bottom line in population density of all 27 states, went through complex political and social histories during the last hundred years. As a consequence the population is faced with diverse language options within and across their national boundaries: Languages serve different functions, and have a variant status depending on who chooses which communicative code with whom, and in which communicative context. This fringe part of Europe is also of particular interest to systemic contact linguists since languages of different typologies clash into each other in this part of Europe (Indo-European/Germanic and Finno-Ugric/Finnic languages). On top of that we can encounter clashes between less and more demographically powerful national cum regional languages (Swedish versus Finnish; Latvian/Estonian/Lithuanian versus Russian), and modern, or even global versus ancestral languages that characterize the language hub (e.g. English and e.g. Sami).

The volume stem from a project that empirically, and critically researched language policy in the poly-lingual language cultures of Finland and Lithuania. At the core of the quantitative and qualitative study lies the (semantically) fuzzy concept of language attitudes. Language attitudes, and language opinions, or sentimentalism in the European North/North-East are grounded in political changes and migration before, and after World War II, as well as in the social and economic changes following the break-down of the soviet regime almost thirty years ago

The publication as a collected volume developed from the editors’ final project report to the German Research Foundation in 2016/17. The report was supplemented by invited papers from specialists in Baltic, and Finnic language cultures that focus on language choice in varying domains and genres. It is the first empirical study of language attitudes and language policy in Finland and Lithuania in book form.

Greifswald, September 2019

←8 | 9→

Editors’ Preface

For both Finland and Lithuania, there is already a well-established tradition in sociolinguistics of dealing with the language situation and policies of the respective countries. There is no shortage of related social-empirical surveys and above all, the social de facto-status of languages—i.e. the attitudes of the population towards majority and minority languages (i.e. Finnish or Swedish in Finland, and Lithuanian, Polish or Russian in Lithuania)—has also been the subject of investigation. However, since we found that these attitudes have not been surveyed either by indirect methods only or by a combination of direct and indirect methods, we were able to obtain funding from the German Research Foundation (DFG) for the years 2014–2016 to conduct a survey to correct this diagnosed deficiency. We want to thank the German Research Foundation for the trust placed in our project.

Our survey investigated the social status de facto, which language speakers in Finland and Lithuania assess as most prestigious. Both countries have similarities in that their inhabitants are highly respectful of their languages and aim for politically correct behaviour according to language policy. Therefore, in addition to language status, statements made about the everyday life of the state’s language policies were also collected from the subjects of the survey. Judgments, from within each society, about the social effects of these linguistic/political measures are, to some extent, entirely different from official lines. The wide-ranging survey intends to give the opportunity to compare both countries using parallelism of the statistical data collected.

In Finland, often cited as a reference country concerning language policy, studies about questions of language loyalty and the status of Finnish and Swedish have already been conducted. However, the political tensions relating to language issues, which have recently manifested themselves, frequently constitute a contradiction in the results of different studies. From this, it would seem that previous research under a more extended period has not been able to illustrate the potential for social tensions resulting from language problems, the effects of which are undeniable and sometimes bear unexpected consequences constituting severe discussions between political decision-makers in Finland and Lithuania. Therefore, the methodology of our research project has been of vital importance. For the first time, a combination of an indirect and direct method—the experimental matched-guise technique and a more traditional questionnaire—has been applied in both countries. By using two research methods simultaneously, ←9 | 10→we hoped to achieve optimal results. Particular attention focused on the question of the extent of respondents’ knowledge about the subject of research, which would influence the results of the survey.

At an event in the Finnish Embassy in Brussels in October 2016, European language policy experts, as well as representatives of Finland and Lithuania, were informed about the first key results of our research project, which was used as a basis for subsequent discussion between scientists, political representatives, regional and language protection activists. The open discourse revealed a particular sensitisation gap between scientists and activists on the one hand and political officials on the other.

The research material has been obtained from several towns in Lithuania and Finland and has been differentiated according to relevant social factors. Our geographical variation of the research locations draws into account the varying conditions of the usage of the languages. The central part of this book contains the documentation of the project including its evaluations. Yvonne Bindrim’s study surveyed the relationship between the Finnish-speaking and Swedish-speaking population in Finland. The comparison of the results reveals that indirectly and directly elicited stances need to be distinguished from one another and that stereotypes are considerably more common than previously assumed. Anastasija Kostiučenko investigated the language situation in Lithuania and its sociolinguistic constellation. Her article presents detailed documentation of the study conducted by her in Lithuania. As a result, people in Lithuania were tolerant of each other and language issues only became ‘hot’ when raised to a political level.

The book is followed by three shorter contributions which look at the language situation in Finland and Lithuania from different perspectives and thus reveals more facets for the reader. Vava Lunabba’s analysis looks at language in Finland from a political-historical perspective which takes account of the legislative process. In the history of Finland’s national languages, there has been an era of language disputes. However, changes in the population structure have had their effects on the language conditions in Finland. The general language climate appears to have become harsher during recent years in Finland. Meilutė Ramonienė examines the linguistic behaviour of Lithuanian city-dwellers in the private sphere and new trends of urban multilingualism in Lithuania. She analyses the linguistic repertoire—the use of languages at home—in mental processes (such as thinking or counting) and when using the media. Her report is based on data from three large-scale surveys carried out from 2007–2012 in Lithuanian cities. Laima Kalėdienė reviews trends appearing in public usage of the Lithuanian language. She evaluates the changes that have emerged during ←10 | 11→the first fifteen years of the new millennium as well as how the problems of management of the language policy that resulted from the trends have succeeded in solving both state and society.

Finally, we would like to thank our staff, Dr Anastasija Kostiučenko and Dr Yvonne Bindrim, who have carried out the project in Lithuania and Finland with great enthusiasm and personal commitment. In terms of data analysis, both scholars have worked intensely on the necessary statistics. Thanks to them, the public now holds the interesting results of our project in their hands.

Greifswald, in Spring 2019

Marko Pantermöller and Stephan Kessler

←12 | 13→

Yvonne Bindrim

On the Relationship between Language
Attitudes and Linguistic Opinions in Finland

Abstract. This study aims to survey the relationship between the Finnish-speaking and the Swedish-speaking population in Finland using an indirect and a direct method in combination: the matched-guise test and a questionnaire. The comparison of the results reveals that indirectly and directly elicited stances need to be distinguished from one another, and that stereotypes are considerably more common than previously assumed.

Overview

1 Introduction

Finland is a country with two national languages: Finnish and Swedish (CF 731/1999). With about 5 % of the total population, the Swedish-speaking community makes up a rather small proportion, but both the Finnish Constitution and the Language Act guarantee the Swedish language a status equal to Finnish. Thanks to this, Finland is often cited as a model multilingual country with regard to the protection of linguistic minorities. However, the legal equality of languages and language users cannot govern the ways in which speakers of different languages treat each other. It does not control what stance a community’s linguistic majority, in particular, takes on speakers of a lesser used language; and it does not touch the opinions on the legal status of the lesser used language and on the linguistic rights that speakers of this lesser used language enjoy. Despite the exemplary legal situation, tensions exist between speakers of Finnish and ←13 | 14→Swedish in Finland’s everyday life, and the country’s bilingualism is a topic of recurrent debates.

In 1997, the Finnish organization Folktinget (the Swedish Assembly of Finland, Swed./Finn. Svenska Finlands Folkting) together with sociologist Erik Allardt with their survey Vårt land, vårt språk (Folktinget 1997) laid the foundation for a number of extensive studies on the status of the Swedish language in Finland. Recent studies, e.g. those by the Swedish-speaking Finnish think tank Magma (2008; 2013), and Samforsk, the Social Science Research Institute at Åbo Akademi (2014), build upon this early work. None of the three studies alone can bring to light the tensions between speakers of Swedish and Finnish in everyday life. By comparing the results of the above-mentioned studies, however, a negative trend in the relationship between the language groups becomes apparent (cf. e2 2017: 40). The conflicts are not new. What is new, though, is that with populism increasing in Finland like in other countries, it has become more acceptable to voice critical or negative opinions about minorities there, too. A very recent example is ethnonationalist comments made by the populist Finns Party youth wing (HS 12/01/2019).

The earlier studies have two things in common: firstly, they do not identify any negative or strongly negative stance in the Finnish-speaking participants towards the Swedish language or speakers of Swedish. Secondly, both studies take on a direct approach, i.e. the participants are aware of the topic of the study. Both Folktinget and Magma collected their data with the help of personal interviews as part of omnibus-surveys. In a direct survey, the salience of the object of study can be another factor influencing the participants’ responses, in addition to the well-known interviewer effect and halo effect. It activates not only the participants’ knowledge and their experiences with the topic, but also stereotypes. This effect may be either welcome or undesirable, depending on the study’s objective. In addition, the participants may be inclined to manipulate their answers because of social conventions like political correctness and taboos or in order to be provocative. Since each of the studies with a direct approach was not capable of diagnosing the conflicts between speakers of the two national languages, the question arises whether a combination of an indirect and a direct approach would be better suited to find out about the participants’ stance on the object of study.

In a study with an indirect approach, the real object of study is not revealed to the participants, which makes it possible to identify their privately held stance towards it. For the study at hand, it is necessary to distinguish between the participants’ attitudes towards languages or speakers of a specific language with regard to the means of elicitation: A stance that is elicited via a direct approach ←14 | 15→is labelled ‘opinion’ and the stance identified in an indirect way is referred to as ‘attitude’. The studies mentioned above refer to directly elicited stances on the Swedish language and Swedish speakers in Finland as ‘attitudes’; except for the researchers at Samforsk (2014), who use the term ‘opinion’. In these studies, a distinction between the types of stance identified within a study is not necessary since all studies used direct means of elicitation.

The term ‘attitude’ has been used with a very general sense in earlier studies, and even though the terminology has been refined over time, the term’s use still varies today. Ajzen gives the following definition of attitude: ‘An attitude is a disposition to respond favourably or unfavourably to an object, person, institution, or event’ and ‘a hypothetical construct that, being inaccessible to direct observation, must be inferred from measurable responses.’ These responses are made up of components that go as far back as to Plato: cognition, affect, and conation. (Ajzen 2005: 3–4.) Attitudes are generally assumed to be learned. Peter Garrett points out the influencing factors, which are relevant also to my study: ‘Two important sources of attitudes are our personal experiences and our social environment, including the media’ (Garrett 2010: 22).

For a long time, studies pointed out only a weak correlation between the attitudes that were elicited by surveys and actual responses (i.e. behaviour observed). This issue was addressed with the differentiation between implicit and explicit attitudes: ‘Implicit attitudes—being automatically activated—are assumed to guide behaviour by default unless they are overridden by controlled processes’ (Ajzen 2005: 36). In the study of language attitudes, Lambert and colleagues found out that participants’ responses were influenced by their knowing what the true object of study was (e.g. when using a questionnaire, one possible direct method). An indirect method, such as the matched-guise test, was found to evoke ‘more private emotional and conceptual reactions’ (Lambert et al. 1965: 90). Other terms for implicit and explicit attitude include subconscious and conscious attitudes (Kristiansen 2009) as well as (implicit) attitudes and (explicitly reported) views (Mattfolk 2011) or attitudes and opinions (Östman/Mattfolk 2011). Lambert’s hypothesis is confirmed, among others, by Kristiansen’s (2009) study of Danish dialects. In his study, Kristiansen showed that language change can be explained with the help of subconscious attitudes, but that the conscious attitudes stand in contrast to the language use as it is observed. Thus, attitudes and opinions are not necessarily congruent.

The terminological distinction in the present study is in line with Östman/Mattfolk (2011: 80), distinguishing between ‘attitudes’ and ‘opinions’: a stance that is expressed implicitly/(un-)/subconsciously is referred to as ‘attitude’ and an explicitly/consciously expressed stance is labelled ‘opinion.’ The first is elicited by ←15 | 16→indirect methods and the latter via direct methods. Throughout this article, these two terms are used exclusively in the respective sense presented right above.

The objective of this study is to establish whether the relationship between speakers of Finnish and Swedish is as strained as can be concluded from the public discourse or whether there is no urgent need for worries. The latter is what interpretations of earlier extensive empirical data on the language situation tend to hint at. In the study, these questions are addressed by a survey consisting of two parts and combining a direct and an indirect method applied to the same sample of participants. A comparison between the two methodologically different parts is facilitated by points of comparison integrated into each part. The parallel survey allows for a comparative contextualisation of the previous studies that were conducted primarily via direct methods.

The results presented here are based on the data of a larger survey that was conducted from 2014 to 2016 in five towns in Finland. For this article, I examined the attitudes of different groups towards the Finnish and Swedish language in Finland: Finnish-speaking participants on the one hand and Swedish-speaking as well as both Finnish- and Swedish-speaking participants on the other hand. In the second part, the participants were questioned for their opinion on issues of language policy, the Swedish language, Swedish speakers, as well as their everyday experience as language users.

In addition to identifying the participants’ stance, the results provide insights into the methods used. By comparing the results of individual questions in the two parts of the survey, answers can be found to methodological questions. One main question is whether the combination of an indirect and a direct method can bring to light the existing conflicts or areas of tension between the language groups.

The participants’ first language1 is assumed to be one of the factors that influence their attitudes and opinions. As a second potential factor of influence, the presence of the other language is included in the study. The latter is operationalised through the choice of towns with different linguistic composition. For this article, I analysed data collected in the towns of Joensuu, Helsinki and Vaasa. These three towns represent three different linguistic majority ratios. The target group of the survey were participants fulfilling all three criteria: having ←16 | 17→been socialised in Finland, speaking exclusively Finnish and/or Swedish as their first language(s) and having one of the three towns as a place of residence.

The next chapter presents the study design, the conduct of the survey and the participants. It is followed by the analysis of responses, which is done separately for the two methods, after which the methods are compared. In chapter 2.5.1, the results of the experimental indirect method, the matched-guise test, are presented. The test addresses the question of whether and to which degree the language that a person uses (here: Finnish and Swedish) influences how another person feels about the speaker. The differences in perception are analysed quantitatively (degree of influence) as well as qualitatively (type of influence). Chapter 2.5.2 presents the questionnaire responses to fundamental issues of language policy. Chapter 2.5.3 offers a comparison between the responses from the matched-guise test and the questionnaire. The comparison brings to light the relation between the participants’ attitudes towards and opinions on the speakers of Swedish (with attitudes and opinions being identified for the language groups as well as for individuals).

As mentioned above, when referring to my study, the term ‘attitude’ is used for an indirectly elicited stance on languages and their speakers and ‘opinion’ for a directly elicited stance. ‘Stance’ is used as a general umbrella term. Another ambiguous term is ‘significance.’ Its use here is limited to the context of statistical testing methods and therefore refers exclusively to significance in its statistical sense. Whenever ‘language group’ is mentioned in this article, this term refers to the two groups of participants with different first languages that are being compared here: one is the Finnish-speaking participants, and the second group consists of Swedish-speaking together with both Finnish- and Swedish-speaking participants.2

Frequently used terms for language groups are abbreviated, e.g. Pssw+ for ‘Swedish-speaking and both Finnish- and Swedish-speaking participants.’ The abbreviation aims at ensuring a smoother reading of the text. The names of the ←17 | 18→surveyed towns are abbreviated as well: JNS for Joensuu, HKI for Helsinki and VAA for Vaasa.

2 The study

The study consists of two methodologically different parts. The first is an experimental part, in which an indirect method is employed: the matched-guise technique (abbreviated MGT in the following).3 It was developed in the 1950s by Wallace Lambert and colleagues in Canada as a method of identifying the attitudes of English and French speakers, respectively, towards speakers of the two languages. The researchers assume that

listener’s attitude toward members of a particular group should generalize to the language they use. From this viewpoint, evaluational reactions to a spoken language should be similar to those prompted by interaction with individuals who are perceived as members of the group that uses it, but because the use of the language is one aspect of behaviour common to a variety of individuals, hearing the language is likely to arouse mainly generalized or stereotyped characteristics of the group. (Lambert et al. 1960: 44)

Since its introduction, the method has been used mainly for research on people’s attitudes towards speakers of regional or social linguistic varieties. In line with the application by Lambert et al., the method is used here in a multilingual context (Finland) for the identification of attitudes towards speakers of different languages (Finnish and Swedish).

In the second part of the study, participants were questioned about their personal attitudes towards issues of language policy and about their own experience as a language user in various contexts of everyday life.

The participants’ answers were collected with the help of LimeSurvey, an online survey tool. The computers or other devices they used to participate in the survey (desktop computer, laptop, tablet, smartphone) were either the participants’ own or belonged to the institution that they were associated with and as a member of which they took part in the survey. Before the participants started the test, they were assured that the survey was conducted anonymously. They were further made aware of the fact that there were no right or wrong answers to the questions, but that instead, their personal stance was of interest. The test was conducted on site for all participant groups with me, the writer of this article, as the conductor of the study. I analysed the resulting data with the help of the statistics software SPSS and additionally with the spreadsheet programme Microsoft Excel for the questionnaires.

2.1 Matched-guise test

The participants (abbreviated Ps) listened to eight different (digital) recordings of one and the same text. The recordings in Finnish and Swedish and by male and female voices were played alternately (Fig. 1). The participants’ task was to imagine the speakers and evaluate them while listening. The evaluation of each recording was to be marked on a semantic differential comprising 12 scales (Fig. 2). The semantic differentials were titled ‘speaker 1’ to ‘speaker 8.’

Fig. 1. Set-up of the matched-guise test

←18 | 19→

The Ps were not informed that the eight recordings they heard were not made by eight different people, but only by six. Two persons, the guise speakers (abbreviated G in general or Gm and Gf, respectively, for the male and female speaker), read the text twice: once in Swedish and once in Finnish. These recordings are referred to as Swedish guise and Finnish guise of the Gm and the Gf. All other speakers, called fillers (F1–3), mainly served to distract the Ps from the fact that they heard and assessed two speakers twice. For the assessment of speakers 1–8 a set of relevant bipolar scales was used (Fig. 2).

Fig. 2. The semantic differential

An odd-point scale is most effective in preventing frustration among the Ps because they are not forced to pick one or the other side of the scale (they were obliged to tick one point in each scale). In the pre-tests, however, this led to a large proportion of Ps ticking the neutral, central point of the scale.4 In reaction, a compromise was chosen for the main test. It aimed at counteracting the Ps’ strategy of choosing the central point of the scale as their ←19 | 20→ ←20 | 21→default answer and only picking a side for cases of a relatively strong stance. At the same time, the risk of frustrating the Ps with the obligation to pick a side should not be increased. The compromise was a 7-point scale with the central point marked as ‘no answer’ (n/a in Fig. 2). In the oral instruction prior to the beginning of the survey, the conductor of the study pointed this out explicitly. The ‘n/a’-answers were not included in the analysis. To process the data, the scales were coded with 6 points (as depicted in Fig. 2), 6 marking the highest degree of a trait’s presence (e.g. ‘friendly’) and 1 marking the highest degree of negation of the respective trait (e.g. ‘unfriendly’). Only the extreme ends of the scales were marked with adjectives; the individual points on the scales were not labelled.

2.2 The questionnaire

Following the matched-guise test, the participants (Ps) were immediately directed to the questionnaire, where they were asked to fill in their basic personal details (first language, age etc.). After this step, the digital survey was paused for a conversation between the Ps and me as the conductor of the study about the first part of the survey. The Ps learned about the functioning of the matched-guise technique and were prepared for the topic of the questionnaire, which is Finland’s bilingualism and language policy.

The questionnaire consisted of one part that was identical for all Ps and that comprised questions on current topics of language policy. It was followed by the second part with questions specific for the individual Ps: students were asked about their language use at school and employees were asked about their language use at work, for instance. Chapter 2.5.1.8 will present the answers to a ←21 | 22→selection of these questions. In part, the wording of the questions was tailored to the individuals, e.g. to the Ps’ respective first language so that the Finnish-speaking Ps were evaluating an issue from the perspective of the linguistic out-group and the Swedish-speaking as well as the both Finnish- and Swedish-speaking Ps took the perspective of the linguistic in-group. The questionnaire was dynamic not only with regard to the wording but also in that it adapted in the course of the survey based on previous details and answers.

2.3 Data collection

In debates on language policy in Finland, the division line between the camps is mostly congruent with the division between the Finnish-speaking population and the Swedish-speaking population. However, the stance on language policy also varies regionally. This suggests that in addition to a person’s first language, the intensity of a person’s contact to the Swedish-speaking population and culture also has an impact on their stance on issues of language policy. For this study, the proportion of Swedish-speaking population in the towns where the survey was conducted was used as an indicator for the intensity of the contact. Therefore, I included the answers from three towns of different linguistic composition in the analyses.

The first town is Joensuu (JNS), situated in monolingual Finnish-speaking Eastern Finland, more precisely in North Karelia (Finn. Pohjois-Karjala, Swed. Norra Karelen), at a long distance from Swedish-speaking areas. With 0,1 % of the municipality’s population being Swedish-speaking (Fig. 3; stat.fi 1), JNS is officially monolingual Finnish-speaking (see Lunabba, 2019: 8, for the criteria for determining the linguistic status of a municipality, and see Fig. 1 of Lunabba’s contribution in this book for a map of bilingual municipalities in Finland). In JNS, the Swedish-speaking culture is not visible; it is thus likely that to the people living there, Swedish feels more like a foreign language than like the other domestic language.

The second place where the survey was conducted, is Helsinki (Swed. Helsingfors; HKI), Finland’s capital situated on the southern coast. The city is officially bilingual with a Finnish-speaking majority and a proportion of Swedish-speaking population of 5.7 % (stat.fi 1) at the time of the survey. Many of the nearby municipalities in the Helsinki-Uusimaa Region (Finn. Uusimaa, Swed. Nyland) are bilingual as well, with Finnish as the majority language. In HKI, the Swedish language is present to a certain extent, primarily and mostly limited to its presence on public signs (in contrast to commercial and private signs; see also Syrjälä 2012: 78). In everyday life, Swedish is of only limited use, ←22 | 23→except for local and state authorities. The infrastructure for the language is good, however, and cultural flagships like the Swedish Theatre (Swed. Svenska Teatern), the Swedish Adult Education Centre of Helsinki (Swed. Svenska arbetarinstitutet, short ‘Arbis (i Helsingfors)’) are familiar institutions for the entire population. The absolute number of speakers of Swedish is high enough (36,004 speakers; stat.fi 2) that some events are held separately from the Finnish-speaking population. This is the case for the traditional picnic on May, 1st (Finland Swed. vappen) and the party on its eve as well as for Finnish Swedish Heritage Day (Swed. Svenska dagen) and Saint Lucy’s Day. The events are announced in the country’s Swedish-speaking media.

←23 | 24→

The third place surveyed is Vaasa (Swed. Vasa; VAA), situated at the west coast in the Ostrobothnia region (Finn. Pohjanmaa, Swed. Österbötten). The city is officially bilingual with a Finnish-speaking majority and a proportion of Swedish-speaking population of 22.6 % (stat.fi 1). The surrounding areas are strongly bilingual as well, including municipalities with a vast Swedish-speaking majority and even almost monolingual Swedish-speaking ones.6 In VAA, both the Swedish and Finnish language can be heard and seen throughout the town, and good bilingual infrastructure is provided for both language groups. Vaasa is the only relatively large bilingual town in Finland with a comparatively high percentage of Swedish-speaking population and an infrastructure of national and local authorities that is similar to the capital.

While in JNS, the percentage of Swedish-speaking population is negligible and only Finnish-speaking persons took part in the survey, a large number of Ps from both language groups could be recruited for the study in the bilingual towns of HKI and VAA.

As mentioned at the beginning of this chapter, the underlying assumption for the data collection was that the Ps’ first language, as well as the intensity of their contact to Swedish-speaking people and culture, were two decisive factors of influence on their stance on issues of language policy. Accordingly, for the analysis that is to follow the Ps are grouped by first language(s) and by place of survey. We then have five subgroups of Ps:

Finnish-speaking Ps in Joensuu (JNSfi),

Finnish-speaking Ps in Helsinki (HKIfi),

Swedish-speaking Ps together with both Finnish- and Swedish-speaking Ps in HKI (HKIsw+),

Finnish-speaking Ps in Vaasa (VAAfi), and

Swedish-speaking Ps together with both Finnish- and Swedish-speaking Ps in VAA (VAAsw+).

←24 | 25→

In order to be comparable, the subgroups need to be similar with regard to other factors that potentially determine their stance on issues of language policy. To achieve this, similar subpopulations were chosen for data collection in each town and within the two language groups.7 The subpopulations include employees and students in educational institutions ranging from comprehensive school to vocational school and university. The survey was conducted in groups of usually 5–20 people with both employees and students. In order to ensure similarity between the subpopulations, employees working for similar employers were chosen for the study. The employers chosen are local and state authorities that can be found in all surveyed towns. In addition, a number of teachers of the above mentioned educational institutions took part in the survey. The second important reason why local and state authorities were chosen for the study in addition to their presence in all relevant towns is that they have special linguistic obligations towards their clients, which directly or indirectly affect the employees. In the educational institutions, one language is (usually) fixed as the language of instruction (Finn. opetuskieli, Swed. undervisningsspråk).8

Due to the fact that the choice of towns to be surveyed provided an infrastructure for data collection that was both appropriate and comparable among the places, it was possible for the most part to collect data from Ps from similar subpopulations. They were complemented by subpopulations which are typical for each town: students of Finnish-Russian schools in JNS on the one hand, and of language immersion classes (Swedish-Finnish) in the bilingual municipalities on the other hand.9 Including these groups allowed for taking into account the special characteristics of the different places present in the study. (Cf. Bindrim 2019: 158, for a detailed account of the sample.)

←25 | 26→

2.4 Participants

A total of 415 people took part in the survey in the three towns presented above. A categorization based on the two assumed determining factors for the stance on issues of language policy (first language and place of the survey) yields five subgroups, as mentioned in the previous chapter. These five subgroups will be presented here in more detail with reference to a number of traditional background variables of the participants (Ps).

Fig. 4. Participants grouped by place of the survey and first language

Legend: JNSfi Finnish-speaking Ps in Joensuu. HKIfi Finnish-speaking Ps in Helsinki. HKIsw+Swedish-speaking Ps and Finnish- and Swedish-speaking Ps in HKI. VAAfiFinnish-speaking Ps in Vaasa. VAAsw+Swedish-speaking Ps together and Finnish- and Swedish-speaking Ps in VAA.

Fig. 5. Participants grouped by gender identity

Legend: JNSfi Finnish-speaking Ps in Joensuu. HKIfi Finnish-speaking Ps in Helsinki. HKIsw+Swedish-speaking Ps and Finnish- and Swedish-speaking Ps in HKI. VAAfiFinnish-speaking Ps in Vaasa. VAAsw+Swedish-speaking Ps together and Finnish- and Swedish-speaking Ps in VAA.

It is apparent from Fig. 4 that the five subgroups differ in size, but they are all large enough to allow for reliable results. As is often the case for surveys, more women than men were ready to participate in the study. As a result, female Ps are overrepresented in all subgroups (Fig. 5). In both bilingual places of the survey, HKI and VAA, the overrepresentation of women is of a similar extent in the different language groups (54–9 % female Ps) and only in JNS is the proportion of female Ps (67 %) higher than in the other subgroups.

←26 | 27→

The composition by age is similar across the subgroups, but again JNSfi and also HKIsw+ display exceptions (Fig. 6). For most groups, the youngest Ps are aged 14–5, only in HKIsw+ are they 17 years old. The Ps’ main occupation is tied closely to their age. The main occupation of a majority of Ps (students at school and in vocational training; cf. Appendix 1) is reflected in the narrow age range of 50 % of the Ps in each subgroup. The age of half of the Ps in the subgroups (quantiles Q1 and Q3 in Fig. 6) ranges from 15 to 23 years and from 15 to 29 years in the JNSfi group, while it is 18 in the HKIsw+ group. The age range is smaller in HKIsw+ than in the other subgroups and wider in JNSfi.

Fig. 6. The age range of the subgroups

←27 | 28→

2.5 Results

As mentioned above, the Ps’ responses from the matched-guise test and the questionnaire will be analysed in separate subchapters. Chapter 2.5.1 will focus on the question of whether the language that a person uses influences the way that other people perceive them. Thanks to the division of the Ps into five subgroups, this question can be answered in a differentiated manner for the assumed influencing factors place of the survey and first language. For each P of a subgroup, the difference is determined between the evaluation of the Swedish guise (Fig. 7: 1.a, c) and the Finnish guise (1.b, d) of the same guise speaker (G) on the same scale. In a following step, the arithmetic mean of all differences within one subgroup is calculated. The resulting mean differences (2.a, b) are tested for significance. For each subgroup, the results show how the evaluations of the guises differ within the group:

What is the Ps’ evaluation on the scales for the two guises of one G?

←28 | 29→

If one guise is evaluated with a higher rank than its counterpart:10 For which traits are the mean differences between the evaluations of one G’s two guises significant?

Is the influence of the language used strong enough to be interpreted as relevant?11

The question that follows here is whether the differences in evaluation of the guises that have been identified before, differ by place and/or first language of the Ps (chapter 2.5.1.8). To find this out, the mean differences found in the subgroups between the evaluations of the G’s two guises (2.a and 2.b) are compared, and the divergence between the mean differences is tested for significance (3).

Fig. 7. Analysis procedure for the matched-guise test

Mean evaluation of the two guises by Ps of two subgroups on the same scale (1.a–d), calculation of the mean differences (2.a, b) and their difference (3).

In chapter 2.5.2, I analyse and compare questionnaire responses to a number of questions on language policy for the five subgroups. One of the issues addressed is the P’s view on common stereotypes on the Swedish-speaking ←29 | 30→community. After separately analysing the results from the matched-guise test and the questionnaire, the findings gained by the two methods will be compared in chapter 2.5.3. There are three points of comparison between the matched-guise test and the questionnaire. The Ps’ views on three common stereotypes elicited by the questionnaire can be put into relation to the Ps’ evaluation of the guises on the three scales labelled ‘well-of,’ ‘educated’ and ‘influential.’

2.5.1 Language attitudes—the influence of language on how a person is perceived

In a first step, I analyse for all five subgroups whether the assessment of a person depends on the language that this person uses. If a subgroup’s perception of a person is independent of the language they use (Finnish or Swedish), there is no difference between the perception of the two guises by the subgroup; i.e. the average difference between the perception of the two guises is not significant. For each subgroup, significance of the differences is determined by a two-tailed paired t-test.

Only relevant differences, i.e. differences in perception that can be assumed to be important in everyday situations, are of interest here. Such a minimal value of difference is necessarily arbitrary since social situations are too complex and individual to determine a standard value.12 Relevance always requires significance (strictly in a mathematical sense).

2.5.1.1 Finnish-speaking participants in Joensuu.

We will first look at the subgroup of Finnish-speaking Ps in JNS (JNSfi) and examine whether they perceive the two guises of one and the same G differently depending on the language used. To find this out, we compare the mean evaluation of the Swedish-speaking guise recording (marked in light grey in the following figures) and the Finnish-speaking guise recording (dark grey) of the male speaker (Gm: Fig. 8) and the female speaker (Gf: Fig. 9), respectively. The distance between ←30 | 31→ ←31 | 32→the points on the scale represents the difference between the mean evaluation of the guise speakers regarding the respective trait.

Fig. 8. JNSfi, paired t-test—mean evaluation of Gm

Fig. 9. JNSfi, paired t-test—mean evaluation of Gf

It becomes clear at first glance that the mean evaluations of the Swedish and Finnish Gm differ with regard to only a few traits. For the Gf, in contrast, the mean evaluations are close with regard to only a few traits. The distance between the points on each scale visualises the mean difference as resulting from the paired t-test, between the evaluations of the guise of one G for one trait. The following table shows the absolute values of the mean differences and their statistical significance (significance and relevance in bold).13

Table 1. JNSfi, paired t-test—mean difference between the evaluations of the masks

←32 | 33→

As could be seen already from the visualisation above, the mean differences between the two guises of Gf are clearly greater than those between the two guises of Gm and the former are significant in more cases. For the Gm, the difference in perception of the two guises is statistically significant for three traits. The Finnish guise is perceived as more intelligent and more ambitious, and the Swedish is perceived as more confident. Because all three differences are below 0.4 points, they are interpreted as non-relevant (see footnote 12 for an explanation of the term ‘relevance’ in this study).

The difference in perception of the Gf ‘s two guises is significant for ten out of twelve traits. Half of these differences are above 0.4 points and therefore have to be interpreted as potentially relevant in most everyday situations. The traits in question are ‘friendly,’ ‘honest,’ ‘reliable’ and ‘ambitious.’ For all traits, the Swedish guise is ranked higher.

The influence of the language used by the guise is clearly stronger for Gf than for Gm, as the difference values show, which are higher for the Gf for almost all traits. The divide between the perception of the two guises of Gf by the Psfi in JNS is striking.

Another interesting result is the finding that the language can influence the perception for one trait in opposite directions for Gm and Gf, as the scales for ‘intelligent’ and ‘ambitious’ show. When it comes to being confident, it is the Swedish guise that receives a higher evaluation for both Gs.

2.5.1.2 Finnish-speaking participants in Helsinki.

We now turn to the Ps in HKI to see whether they perceive the two Gs differently depending on the language the Gs use. First, the perception of the guise by the Finnish-speaking Ps is in focus (HKIfi).

Fig. 10. HKIfi, paired t-test—mean evaluation of Gm

Fig. 11. HKIfi, paired t-test—mean evaluation of Gf

Table 2. HKIfi, paired t-test—mean difference between the evaluations of the masks

For the Gm, the differences between the mean evaluations of the guises vary (Fig. 10), while for the Gf, their values are similar on most scales (Fig. 11). Table 2 presents the absolute value and significance of the mean differences.

Statistically significant differences in the perception of the two guises of the Gm can be identified for seven traits. The differences for ‘social’ and ‘confident’ are above 0.4 points and can, therefore, be interpreted as relevant. For both traits, the Swedish guise received a higher-rank evaluation. For ten out of twelve traits, the difference in perception of the two guises of Gf is statistically significant. Out of these, at least those for ‘friendly,’ ‘reliable,’ ‘intelligent’ and ‘ambitious’ have to be interpreted as relevant. Again, the Swedish guise receives higher evaluations for all traits.

The difference in perception is significant and relevant for more traits for the Gf than for the Gm, which means that on the whole, the language spoken influences the perception of the female speaker to a higher degree than that of the male speaker. For the Gm, however, the influence is stronger for the two ←33 | 34→ ←34 | 35→relevant traits, as the larger differences show. The direction of influence is the same for all significant differences. Both Gs are evaluated with a higher ranking when they speak Swedish.

2.5.1.3 Swedish-speaking and both Finnish- and Swedish-speaking participants in Helsinki.

The following comparison of the two language groups addresses the question whether the Swedish-speaking Ps perceive the two guises differently than the both Finnish- and Swedish-speaking Ps in HKI (Psw+). As the figures below show, most differences between the mean evaluations of the two Gm guises are relatively small (Fig. 12). For the Gf, the mean evaluations are very close for some traits as well, while for others, they are notably further apart (Fig. 13).

Fig. 12. HKIsw+, paired t-test—mean evaluation of Gm

Fig. 13. HKIsw+, paired t-test—mean evaluation of Gf

Table 3 presents the absolute values of the mean differences between the evaluation of the guises as well as the differences’ significance.

Table 3. HKIsw+, paired t-test—mean difference between the evaluations of the guises

For the Gm, the differences are statistically significant and relevant for the two traits ‘educated’ and ‘successful.’ For both traits, the Finnish guise received a higher evaluation. The differences in perception of the Gf are statistically ←35 | 36→ ←36 | 37→significant for four out of twelve traits. Out of these, the differences for ‘friendly,’ ‘reliable’ and ‘ambitious’ can be interpreted as relevant. For all traits, the Swedish guise was evaluated higher.

The number of significant differences between the perceptions of the two guises by the Psw+ in HKI is higher for the Gf than for the Gm. The three significant and relevant differences are also bigger for the Gf. With the trait ‘successful,’ the difference in perception is significant for both guises. It is also relevant only for the Gm. The direction of influence is opposite in comparison with the two Gs. For the Gm, it is the Finnish guise, and for Gf it is the Swedish guise that is perceived as more successful. On the whole, the language spoken has a stronger influence on the perception of the Gf than on the Gm, with relevant differences being more frequent as well as of a larger extent.

2.5.1.4 Finnish-speaking participants in Vaasa.

The next group to be analysed for potential differences in perception of the two Gs depending on the language the Gs use is the Ps in VAA. Starting with the Finnish-speaking Ps (VAAfi), we gain the following picture.

Fig. 14. VAAfi, paired t-test—mean evaluation of Gm

Fig. 15. VAAfi, paired t-test—mean evaluation of Gf

←37 |
 38→

The differences between the mean evaluation of the guises of Gm (Fig. 14) are rather small. For Gf (Fig. 15), the situation is similar, with comparatively big differences for only a small number of traits. As Table 4 shows, only a few of the statistically significant mean differences are big enough to be interpreted as relevant.

Table 4. VAAfi, paired t-test—mean difference between the evaluations of the guises

←38 | 39→

While the differences in perception of the Gm by the Finnish-speaking Ps in VAA are significant for five traits, none of these is relevant. The Swedish guise of Gm was evaluated higher for all traits except for ‘social.’ For the Gf, the difference in perception of the two guises is statistically significant for nine out of twelve traits. Only for ‘friendly’ and ‘reliable’ are they also relevant. It was the Swedish guise that received higher evaluations for all traits.

There are more significant differences in the perception of Gf than for the Gm guises. The difference in the scale for ‘social’ is of similar size comparing the two Gs. This is also true for the traits ‘intelligent,’ ‘ambitious’ and ‘educated,’ but here the direction of influence is different. For the Gm, the Finnish guise was evaluated as more social, while for the Gf, it was the Swedish guise.

←39 |
 40→

2.5.1.5 Swedish-speaking and both Finnish- and Swedish-speaking participants in Vaasa.

The mean evaluations of the two Gm guises (Fig. 16) by the Swedish-speaking Ps in VAA (VAAsw+) display only small differences. The differences are greater for the Gf guises (Fig. 17).

Fig. 16. VAAsw+, paired t-test—mean evaluation of Gm

Fig. 17. VAAsw+, paired t-test—mean evaluation of Gf

The absolute mean differences and their statistical significance are compiled in Table 5.

Table 5. VAAsw+, paired t-test—mean difference between the evaluations of the guises

For Gm, three differences are statistically significant, but they all are too small to be interpreted as relevant. For the Gf, the differences for eight traits are statistically significant, of which four are also relevant, namely those for ‘friendly,’ ‘honest,’ ‘reliable’ and ‘well-off.’ The Swedish guise received a higher rating for all these traits.

While for both Gs, the differences in evaluation for ‘intelligent’ and ‘educated’ are significant, the direction of influence is not the same. For Gm, it was the Finnish guise, but for Gf the Swedish guise that was evaluated as more intelligent and educated. The larger difference for Gm shows that for these two traits, the influence of the language on the perception by the Ps is stronger for Gm than for Gf. Overall, however, the influence of the G’s language is stronger for the Gf ←40 | 41→ ←41 | 42→guises, which can be seen in the higher number of significant as well as relevant differences in mean evaluation.

2.5.1.6 The influence of language on perception in quantitative terms.

It became clear in the above analyses that the influence of the G’s language on the Ps’ perception is notably stronger for Gf than for Gm. This can be concluded, among others, from the number and size of the mean differences between the evaluations of the Finnish and the Swedish guise of the same G. In all subgroups taken together, 20 differences can be found between the Swedish and the Finnish guise of Gm and a total of 41 for the guises of Gf.

Looking only at the relevant differences (≥ 0.4 points), the intensity of the language’s influence on the Ps’ perception diverges even more. For Gm, the difference in perception was found relevant only for two traits each in the subgroups HKIfi and HKIsw+. In contrast, for Gf, relevant differences are found in the perception by all five subgroups for two to five traits, adding up to 18 relevant differences. Five of these differences are greater than 0.6 points, while the greatest difference between the Gm guises is 0.526 points. The findings on the different extent to which the language spoken influences the perception of the Gf in comparison to the Gm suggest that the influence of language and gender combines. By implication, conclusions can be drawn only by taking into account the gender of the Gs.

It is impossible to compile one simple ranking of the subgroups in terms of the intensity of influence that the G’s language had on the Ps’ perception. This is because firstly, we would need separate rankings for the Gm and the Gf and secondly, prioritising different parameters would lead to different orders within such a ranking. The values of all parameters mentioned in the following can be found in Table 1 to Table 5. An overview of the values of all subgroups is given in Appendix 2.1.

One of the parameters that can be used for a ranking is the sum score. It represents the sum of the mean differences between the evaluations of the two guises of the Gs, with only significant differences being taken into account in the calculation. The sum score (abbreviated ‘sc’ in the following) expresses the intensity of influence of the respective language used by the Gs on their perception by the Ps—in other words, it indicates how great the difference in evaluation is that the Ps assigned to the guises.14 The lower the sc, the lower the number ←42 | 43→of significant (but not necessarily also relevant) differences. Putting the sc and the number of significant differences in relation creates a second parameter: the average mean difference (Øsc). The number of significant and relevant differences are used as further parameters (number of Δsign, number of Δrelev—see Appendix 2.1). Keeping in mind that different parameters would produce different rankings for the intensity of influence, the following analysis also addresses the question of how the influence of language on the Ps’ perception may differ across the subgroups. Firstly, the focus lies on the issue how differently the Gs’ language influences the Ps’ perception of the Gm and the Gf in comparison.

For the Gm, the strongest influence of the language was registered for the subgroup HKIfi. This manifests in the sum score, which at 2.534 points is far higher than for the other subgroups. Additionally, differences in the Ps’ perception of the Swedish and the Finnish guise are found significant for a total of seven traits. In two cases, the differences are also relevant. The Ps’ perception is thus influenced by the G’s language with respect to a variety of traits, but the differences are of only intermediate size, with 0.362 points on average. For the subgroup VAAfi, the results are similar. Their sc is the second highest at 1.679 points, but the differences are significant for only five traits and are all below the relevance minimum. The mean difference is rather small at 0.336 points.

The results are considerably different for the HKIsw+ subgroup. While the sc is the lowest of all subgroups and based on only two differences, both of these are great enough to be relevant. It becomes clear that the language spoken by the Gm has an influence on the Ps’ perception only for a small number of traits, but the influence is strong. The differences amount to 0.440 points on average, the highest value across the subgroups.

Comparing the subgroups JNSfi and VAAsw+ yields very similar results: The Ps of both subgroups perceive the Gm significantly differently for three traits, with none of these differences being relevant at the same time. Due to the relatively small mean differences, the sc is rather low for both subgroups, with slightly higher values for VAAsw+.

The Gf ‘s language has the strongest influence on the perception by the Ps in JNSfi and HKIfi subgroups, with a sum score of 4.379 points, which is the highest for all subgroups, and 3.824 points, respectively. In the former subgroup, five out of ten significant differences are relevant—in the latter group, it is four. The mean differences are higher in JNSfi at 0.438 points than in HKIfi with 0.382 points. The influence on the perception of the guises happens not only with a high number of traits, but it is also remarkably strong for half of the traits.

For both VAA language groups, a high number of significant differences leads to relatively high sum scores with similar values for both groups (3.413 for VAAfi ←43 | 44→and 3.265 for VAAsw+). While for VAAsw+ four out of eight significant differences are relevant, it is only two out of nine for VAAfi. In the first subgroup, the mean difference is 0.408 points and in the second 0.379 points.

HKIsw+ has the lowest sc, at 2.091, but at the same time the mean difference is highest in this subgroup, at 0.523, and three out of four significant differences are relevant. The results indicate that the perception of the guises within this subgroup is influenced for only a few traits, but very strongly so.

Based on the results presented, the following general statements can be made on the influence of the G’s language on the perception by the Ps in quantitative terms. Ranking the subgroups by decreasing sum score (which simultaneously means a decreasing number of significant differences), the order is the same for both Gs for all subgroups but one: HKIfi – VAAfi – VAAsw+ – HKIsw+. From left to right, the influence of the G’s language on the Ps’ perception decreases. The exception is the JNSfi subgroup, where the influence is remarkably different for the two Gs.15 For Gm, the sum score is among the lowest, while for Gf, it is by far the highest value. (An explanation is given later in this chapter.)

Moreover, in all bilingual places of the survey, the G’s language has a stronger influence on the Psfi than on the Pssw+. For the latter group, it is a more common experience to (be obliged to) switch to the other language, which is the language of the linguistic majority—this observation is confirmed by questionnaire responses (Bindrim 2019: 455–7). It is also not unusual for them to find a person to be Swedish-speaking or both Finnish- and Swedish-speaking, whom at first they had perceived to be Finnish-speaking. Switching between Swedish and Finnish is thus not unusual for Swedish-speaking as well as both Finnish- and Swedish-speaking people. Based on this, it is plausible to assume that to Pssw+, language is not as much a determinant of individual behaviour and character traits and therefore used to a lesser extent to categorise people. For Finnish-speaking people, the experience is different. They switch between Finnish and Swedish less frequently. There are several reasons for this, which may indeed even reinforce each other: For the majority of Finnish-speaking ←44 | 45→people, occasions to communicate in Swedish are rare, e.g. because they live in a monolingual Finnish-speaking environment. Other reasons may be lack of confidence in speaking Swedish as well as the social status ascribed to Finland Swedish. It may be difficult for Finnish-speaking people to switch to Swedish because they lack skills of active language use or else they do not want to switch to Swedish because they regard the Swedish language or Swedish-speaking people in Finland as stigmatised. Both reasons can lead to Finnish-speaking people consciously or unconsciously avoiding situations potentially involving Swedish or both languages. In essence, the use of Swedish and switching to Swedish is more strongly marked for Finnish-speaking people, because it represents a deviation from the majority’s norm.16

Another interesting result is that the difference between the intensity of influence for the two language groups is higher in the HKI subgroup than in the VAA subgroup. The difference between the sc values of the Pfi and Psw+ for the Swedish and the Finnish guise of the Gm is at 1.662 points for HKI. For VAA, the difference is considerably smaller, at 0.502 points. This indicates that the language spoken by the G has an influence of similar intensity on the perception of the guises by the Ps in the two language groups in VAA, but an influence of different intensity comparing the language groups in HKI. The results for the perception of the Gf are similar for the two places of the survey. With this second finding, we can elaborate on the reasoning presented above concerning the markedness of switching languages. The town of VAA is characterised by living bilingualism that can be heard and seen on the streets. That is why the experiences of Finnish- and Swedish-speaking people are more alike here than in HKI, a city where bilingualism cannot be experienced to the same extent. And also in VAA, the Psfi are influenced more strongly in their perception of the guises than the Pssw+. This is in line with the previous findings. The former are speakers of the majority language and also in VAA, there are more monolingual Finnish-speaking people than monolingual Swedish-speaking people. ←45 | 46→For the latter, the Pssw+, switching language to speak the majority language is therefore less marked. As a Swedish-speaking person in Finland, one is highly likely to have this experience of switching to the majority language almost throughout the country. This interpretation raises the question why the HKIsw+ subgroup is apparently influenced less strongly by the guise’s language in comparison to the Ps of both language groups in VAA—even though the former is supposed to be more familiar with switching the language in everyday life. This question can be answered with a careful analysis of the statistical results on the influence on the Psw+ ‘s perception in HKI. Only in terms of the sum score does the intensity of influence appear to be the lowest of all groups. Using other parameters to create the ranking, however, the picture is likely to change, as pointed out above. A closer look at the influence registered for HKIsw+ shows that only in this subgroup, the perception is influenced for only very few traits (with the number of significant differences for both Gs being the lowest of all subgroups) without strongly affecting the overall impression of the G. But for these few traits, the influence is considerably stronger than in the other subgroups (the mean differences for both Gs being highest of all subgroups). In fact, HKIsw+ is one of the only two subgroups for which the difference in perception of the Gm is relevant, and even both significant differences are great enough (above 0.4 points) to be relevant. Two significant differences of 0.440 points on average can be considered to have a stronger impact than the three and five differences in the VAA subgroups that are significant, but not relevant. For Gf, the differences in perception by HKIsw+ are significant for four traits and relevant for three of them. The mean differences for Gf are even by far the highest, on average. These results can be interpreted in a very similar way to those of the Gm. Admittedly, one question has to be left open—would the sum score of HKIsw+ have been higher than in one or both VAA subgroups if the groups had been of equal size? Still, it became clear that the sc alone does not necessarily fully reflect the intensity of influence that the G’s language has on the perception of the guises among HKIsw+ Ps. This interpretation is backed up by the results, even though direct comparison of the groups is not possible with the present data.

The central question addressed here is whether the influence of the G’s language is of a different intensity across the five subgroups. After presenting some probable reasons for the differences in perception, the focus now shifts back to the statistical results. In the following, the sum scores of all subgroups will be looked at together and in comparison (see sc values in Appendix 2.1 for the data that the calculations are based on and see Fig. 18 for a visualisation). For the comparison, the first step is to establish the mean sum score for each G, ←46 | 47→which serve as reference points for the sc of the five subgroups. The mean sum score (Øsc-m) is 1.440 points for Gm and 3.394 points for Gfsc-f). In Fig. 18, the mean sum scores are represented by horizontal lines and the subgroups’ sum scores for the two Gs are represented by data points.

Fig. 18. Sum scores and mean sum scores for the two guise speakers

One finding is obvious right at first glance, and it has been described for the individual subgroups in chapters 2.5.1.1 to 2.5.1.5 already. Across all subgroups, the Ps’ perception of the Finnish in comparison to the Swedish guise differs more strongly for the Gf than for Gm. This is indicated by the fact that in each subgroup, the sum score is considerably higher for Gf than for Gm. The mean sum scores for the two Gs differ accordingly. We will first look at the sc deviation from the mean sc across subgroups for the two Gs, starting with Gm.

In three subgroups, the sc for Gm is below the mean sum score (ØSc-m). This means that for Ps in JNSfi, HKIsw+ and VAAsw+ subgroups, the difference in perception between the Swedish and the Finnish guises of Gm is smaller than for the Ps of the two other subgroups. In contrast, the difference is far above average with the HKIfi Ps. The language spoken thus has a stronger influence on the perception of the Gm by HKIfi Ps than on the perception by Ps in the other subgroups. For VAAfi, the intensity of influence is also above average, even if only slightly. For the Gf, the only subgroup with a difference in perception far below average is HKIsw+. Ps of this subgroup are influenced only to a small extent by the language ←47 | 48→that the Gf uses. The opposite is true for the JNSfi and the HKIfi Ps: the sc and the influence of the Gf’s language on their perception, accordingly, are above average. In-between are the results for VAAfi and HKIfi, whose Ps perceive the Swedish and the Finnish guise at a roughly average difference.

To shed light on the question of to what extent the intensity of influence differs between Gf and Gm within one subgroup, the sc values for the two Gs are compared per subgroup. The comparison is made for each subgroup between the sc for the Gm (the value that represents the sum of significant mean differences between the Finnish and the Swedish guise of the Gm), and the sc for the Gf from the same subgroup. The result is visualised in Fig. 18 by the vertical lines between the data points. The longer the vertical line, the greater is the difference in intensity of the language’s influence on the perception of the two G’s guises. Of the two sc values, the higher one shows for which G the difference in perception of the guises is higher. It is the Gf for all subgroups. In sum, the length of the vertical lines can be said to express the extent of the influence of gender, or rather of the combination of language and gender, on the difference in the Ps’ perception of the guises.

The largest difference between the sc values for Gm and Gf can be identified for the JNSfi participants: For Gm, the influence of his language on the Ps’ perception is the second lowest, while for Gf, it is the highest of all subgroups. This remarkable difference results from the contrast between a high number of significant as well as relevant differences in the perception of the Gf guises on the one hand and a low number of relevant differences and the lowest mean sc for the Gm on the other hand. This can be interpreted as extreme reactions to the G’s language by the Ps of the JNSfi subgroup. Their perception of the Gf ‘s Swedish guise differs most drastically from that of the Finnish guise, while it differs least for the guises of the Gm (at the same level as the HKIsw+ and VAAsw+ Ps). This relation can be expressed by the influencing factor, which is 4.61 for the JNSfi subgroup. The perception by these Ps thus differs 4.61 times more strongly between the Finnish and the Swedish guise of the Gf as compared to the two guises of the Gm. However, the JNSfi subgroup’s perception does not display higher differences between the Gm guises than the perception in HKIsw+ and VAAsw+ subgroups. In effect, the high difference between the Gf ‘s and the Gm ‘s perception among JNSfi Ps results exclusively from the strong difference in their perception of the Gf guises (further comments below).

The two language groups in HKI display a smaller difference between their perception of the two Gs than the VAA subgroups, as can be seen in Fig. 18 from the vertical lines connecting the sc values for the Gm and Gf that are shorter for the HKI language groups. The difference in perception between the guises of the two Gs are comparatively strong in the HKIfi subgroup. The sc values are ←48 | 49→relatively high for both G, but they are quite similar. The difference between the sc for Gm and Gf is comparatively small, at Δ = 1.290. The G’s gender thus influences the perception of the two guises only to a small extent.

The HKIsw+ subgroup’s result is very different from those of HKIfi in one respect, and very similar in another. Unlike in HKIfi, the sc values for Gsw and Gfi of both Gs are comparatively low, meaning that the difference in perception is small with regard to the G’s language. With regard to the difference between the two sc values per G, the results are similar to the other language group, with a low result of Δ = 1.219. Even though the sc values for Gm and Gf differ considerably comparing the two HKI language groups, the influence of gender on the Ps’ perception of the Gs is equally small in both groups.

In both bilingual places of the survey, the difference between the sc values for the two Gs’ guises is smaller in the Finnish-speaking group than among Psw+. The HKIfi Ps react 1.5 times more strongly to the Gf ‘s language compared to the Gm ‘s language, while the factor is 2.8 for the VAAsw+ subgroup. Gender or the combination of gender and language thus has a much stronger influence on the perception among VAAsw+ Ps than on HKIfi Ps. The factors for the other subgroups of bilingual places of survey range in-between the two mentioned above. Why is it, though, that the JNSfi Ps are influenced 1.5–3.0 times more strongly in their perception of the guises by the combination of language (Swedish) and gender (female) than the Ps of other subgroups?

As pointed out above, it is only the guises of the Gf that the JNSfi Ps differentiate markedly between; and they evaluate the Swedish guise higher than the Gfi (see Table 1 and Fig. 18). It is thus the G combining the features female and uses Swedish that stands out for the JNSfi subgroup. What I assume to be the reason for this has to do with the fundamental frequency that differs for the two languages. Just like all other Ps, the Ps in the JNSfi subgroups hear Finnish spoken by women in everyday life. Finnish is one of the languages that are known to be spoken with a comparatively low fundamental frequency—by both men and women. The fundamental frequency of Finland Swedish is relatively low, too, but higher than for Finnish, on average. I analysed the fundamental frequency for one sentence each of the Finnish and the Swedish guise recording of the Gf with the help of the software Praat (see praat.org) and found the following. For the Gf, the fundamental frequency is 189 Hz on average when the speaker uses Swedish and 175 Hz when she uses Finnish. A person, and a woman in particular, with a higher fundamental frequency in their normal speaking voice is more commonly associated with stereotypically female traits, and it is thus not surprising to find that the Gf ‘s Swedish guise is evaluated higher by all subgroups for the corresponding traits (on the scales friendly and reliable).

←49 | 50→

For the Ps in JNS, it is least common to hear women speaking at a comparatively high voice. They, therefore, show the strongest reaction to the phenomenon. The Ps of all other places of the survey are more used to hearing Swedish spoken at a higher fundamental frequency. To them, the fundamental frequency of the Gf ‘s Swedish guise does not differ to a noteworthy extent from the fundamental frequency that they would expect of a female speaker. Consequently, their reaction to the voice is least strong. However, the tonal aspect can only be part of the explanation for the fact that the Ps of all subgroups evaluated only the Gf ‘s Swedish guise higher; it has to be combined with the other explanations presented above.

This chapter addressed the question of whether the influence of the G’s language on the perception of the guises differs across the five subgroups. The analysis showed that the influence can come in different shapes: for some subgroups, it may be comparatively weak, but present for a high number of traits, or it may be strong for only a few traits. Despite the variation in ranking resulting from the variety of parameters that can be taken into account, we can identify the order in which the influence on the perception of the Gs decreases for four subgroups.

In the bilingual places of the survey, the influence of the G’s language is less strong on the Pssw+ than on the Psfi. Taking into account the ratio of significant to relevant differences in addition to the sum score, the Ps in towns with living bilingualism appear to be less strongly influenced by the G’s language. These two findings from the quantitative analysis both indicate the following. The higher the proportion of the Swedish-speaking population at the place of the survey, the weaker is the influence of language on the Ps’ perception. These findings probably owe to the fact that people who experience individual bilingualism in everyday life consider language less as a determinant of character traits.

The fifth subgroup, monolingual Finnish-speaking participants in JNS, provides the exactly opposite picture. They differentiate least between the Swedish and the Finnish guise of the Gm, but most strongly for the guises of the Gf. The high contrast between the languages can be explained by the noticeable divergence in the pitch of the Swedish guise from the Finnish one, as perceived by the Ps of this subgroup.

Across all subgroups, the influence of the G’s language on the perception of the guises is higher with the Gf than with the Gm. It is for the JNSfi subgroup that the combination of gender and language, and pitch, in addition, is of particular importance.

←50 |
 51→

2.5.1.7 The influence of language on perception in qualitative terms.

After looking at the number and size of the differences in perception and employing various quantitative parameters, this chapter now focuses on the qualitative influence of language on the perception of the guises. The following questions will be addressed:

For which traits do we find significant differences in the perception of the guises?

What is the direction of influence for the two Gs—i.e. is it the Gsw or the Gfi that receives a higher evaluation?

Comparing the subgroups, what is similar and what different with regard to these two questions?

How can the differences in perception be explained?

The previous chapter showed that the influence of the G’s language differs by gender, with the Ps’ perception of the guises being influenced to a different extent for the Gm and the Gf. Accordingly, the qualitative analysis of the influence on the P’s perception of the guises is done separately for the two Gs.

At first glance, there appears to be no common pattern in the perception of the guises of the Gm, looking at the subgroups’ results presented in the previous chapters (see 2.5.1.12.5.1.5). Across the subgroups, Ps differentiate between the Gm ‘s Swedish and Finnish guise for a variety of traits and give higher evaluations to different guises. Despite this variation, with a closer examination of the results, a number of details can be identified that all or at least several subgroups have in common or that constitute explainable differences. To start with, in none of the subgroups do the Ps differentiate the guises of the Gm with regard to perceived friendliness. The Gm ‘s language thus does not trigger a significantly different perception of how friendly the guises are.

For the other eleven traits, 20 significant differences in perception can be identified for the Gm among the Ps of all five subgroups. Four of these differences can be regarded as relevant, namely those for the traits ‘social,’ ‘confident,’ ‘educated’ and ‘successful.’ For all four, it is the Ps in one of the HKI subgroups for whom the difference in perception is relevant. The quantitative analysis already brought to light that the HKIsw+ Ps are influenced in their perception for only a small number of traits (‘educated’, ‘successful’), but most strongly so, compared to the other subgroups. We now see that also the HKIfi Ps are influenced more strongly in their perception than other subgroups, at least for the traits ‘social’ and ‘confident’ of the Gm. This can be explained in the same way as for the Pssw+. In Helsinki, language is more likely considered to be a determinant of personality ←51 | 52→as in Vaasa, for example, due to the capital’s linguistic situation that provides less inter-language contact. Here, Swedish is regarded as a deviation from the norm, and it is thus the marked variety, even though the city is officially bilingual, but with a comparatively small Swedish-speaking minority.

The next issue addressed here is whether either the Gfi or the Gsw is evaluated higher unanimously across the subgroups for any trait. It will be followed by an examination of the traits for which the G’s language has an opposite effect on the Ps’ perception, comparing the subgroups. For most traits, the same Gm guise was evaluated significantly higher by the subgroups. The Swedish guise was ranked higher in the evaluation for the traits ‘honest’ (1),17 ‘well-off’ (1), ‘confident’ (2), ‘respected’ (1) and ‘influential’ (1). In contrast, for the traits ‘intelligent’ (3), ‘ambitious’ (2), ‘educated’ (3) and ‘successful’ (1) it was only the Finnish guise that received the higher evaluation.

Some of the traits for which the Swedish guise exclusively received higher evaluations can be regarded as part of a disposition of features commonly associated with privileged people in society. These traits are either ‘hereditary’ in a social capital reading, i.e. they can be passed on or can be developed on the foundation of inherited social capital (‘confident,’ ‘well-off’), or they are typically ascribed to a person by others (‘respected,’ ‘influential’). Attributing respect and influence, for example, is based on a high standing that already exists and at the same time, it entails confirming and further improving the standing. In this way, privileges can be self-sustaining rather than arising from an individual’s achievements.18 In contrast, the Gfi is more often evaluated higher for traits that can be achieved through one’s own work, namely ‘ambitious,’ ‘educated’ and ‘successful.’ The contrastive pair ‘successful’ and ‘influential’ exemplifies particularly well the difference between privilege and achievement—while someone can be successful without having any professional or political influence, it is impossible to be influential without being successful. In order to be influential, it is necessary to be in the relevant high position that makes it possible to exert influence.

The results for the Gm from the subconscious evaluation elicited by the matched-guise test are, to a certain extent, in line with the stereotype of Finland ←52 | 53→Swedes being privileged. The interpretation of the findings as a contrast of privilege and achievement appears less reliable if considering that for five of the traits mentioned (‘honest,’ ‘well-off,’ ‘respected,’ ‘influential,’ ‘successful’), significant differences in the perception of the Gm ‘s guises were registered in only one subgroup each. These subgroups are all Finnish-speaking Ps, except for the trait ‘successful’. In contrast, the traits ‘intelligent,’ ‘ambitious’ and ‘educated’ are important for several subgroups, and they all evaluate the Finnish guise higher on these scales. This shows that it is a widespread unconscious judgement that the Gm ‘s Finnish guise was more likely to have had to achieve something by his own work (‘ambitious’ and ‘educated’). It is clear from the results that it is not exclusively a matter of linguistic in- or out-group, which backs up the interpretation involving the stereotypes of privilege and achievement.

For the traits ‘reliable’ and ‘social,’ the influence of the Gm ‘s language works in opposite directions. On the social scale, evaluations are significantly different for the Gm ‘s two guises in three subgroups. The Psfi in HKI and VAA give a higher evaluation to the Swedish guise. This matches with the stereotype of Finland Swedes having a closer social network. It contrasts with the stereotype that Finnish-speaking Finns, especially men, tend to be quiet and less sociable. Despite the stereotypes, VAAsw+ participants perceive the Gm Finnish guise as more social. This cannot be explained by the data alone. In a cautious attempt at explaining this, I suggest that it is the cultural differences between the Finnish- and the Swedish-speaking population as well as between urban and rural regions that play a role. In the Swedish-speaking tradition, social get-togethers are strongly characterised by fixed routines, e.g. by the singing of drinking songs (Swed. snapsvisor) at Crayfish parties (Swed. kräftskiva) and other special occasions or by the procession and singing of traditional songs for St. Lucy’s Day. In VAA, traditional events like these are commonly celebrated and are part of a lively culture. Traditional celebrations often take place at home with the family, and the Swedish-speaking population of VAA tends to have family ties within the municipality or the surroundings that are characterised by a strong Swedish-speaking culture (vaasa.fi 1: 7). HKI attracts people from all over Finland, making it the city with the highest influx of residents (hel.fi 1: 34), and also VAA loses the vast majority of people moving away to the capital (vaasa.fi 1: 8). With traditional celebrations being held rather in the home municipality than in the new place of residence, in everyday life, Swedish-speaking people in HKI socialise in the same urban, modern forms as Finnish-speaking people do. Considering this situation, we can assume that participants in the VAAsw+ subgroup regard the Finnish-speaking, non-ritualised forms of being together as more authentic social situations than those that are marked by fixed routines.

←53 |
 54→

A more straightforward explanation can be given for the fact that for the trait ‘reliable’, VAAfi Ps give a higher evaluation to the Gm ‘s Finnish mask, while it is the Swedish guise for the HKIfi Ps. The explanation lies in the different ways in which Finnish-speaking Ps are confronted with Swedish in the two towns, despite the commonalities of the towns’ bilingual status and Finnish as the majority language. In VAA, Swedish is present everywhere, it can be seen not just on local and national public signs, but also on private signs, and it can be heard in the streets. The language thus does not represent a deviation from the norm and is not a marked variety. In HKI, in contrast, the Finnish-speaking population has less contact with the Swedish language than with Finnish, and it is, therefore, a marked choice. Swedish can be seen almost exclusively in official settings, most commonly on written signs by local and national authorities. In consequence, Swedish becomes associated with public institutions, bestowing a certain sense of reliability on the language. The HKIfi Ps apparently—subconsciously—transfer this reliability onto people speaking in Swedish, which is expressed in their perception of the higher reliability of the Swedish guise.

While the differences in perception for the Gm ‘s guises vary greatly across the five subgroups, their perceptions are often unanimous for the Gf ‘s guises: In all subgroups, differences between the guises can be identified for a large number of traits and additionally, Ps evaluate the Gsw higher for all of these. The language’s influence on the perception of the guises is strongest for the three traits ‘friendly’, ‘reliable’ and ‘ambitious’, with the Gsw being ranked higher by Ps of all subgroups. For the former two traits, all differences registered are not only significant but also relevant, i.e. above the minimal value of 0.4 points on the 6-point scales used in the matched-guise test. In four groups, the two guises are evaluated differently for the traits ‘honest’, ‘intelligent’ and ‘educated’ and the Ps of three subgroups make a difference on the scales ‘educated’ and ‘well-off’. None of the traits is of importance for only a single subgroup; all differences in perception are found in at least two subgroups.

Dividing the Ps into language groups, we see that the Psfi perceive the Gf ‘s guises differently for all traits, while for the Pssw+, the traits ‘social’, ‘confident’ and ‘respected’ have no effect. (The same is true, except for ‘social’, for the evaluation of the Gm.) The perceptions of the Gf ‘s guises are more alike among the Psfi than among the Pssw+. This manifests in the fact that all three Finnish-speaking subgroups show significant differences for the same seven traits, while the evaluations by Pssw+ agree for only three traits. This result indicates that the ←54 | 55→Swedish language evokes very similar associations among the Finnish-speaking participants.19

The perception of the guises by the language groups for the trait ‘social’ is congruent with the stereotypes. The Ps’ evaluations in the matched-guise test always build on their frame of reference, their social environment or themselves. Among the Psfi, the Swedish language unconsciously activates the stereotype of Finland Swedes being regarded as particularly social, leading to their perception of the Gsw as more social than the Gfi, which rather represents their own social environment. When the Pssw+ evaluate the guises with their own social environment as a point of reference, the Swedish language does not activate the stereotype, and their perception is not different for the Gf ‘s two guises.

It is not surprising to see that the perception of the guises is influenced in a similar way for the three traits ‘social,’ ‘confident’ and ‘respected,’ considering that confidence and respect are relevant in the social context. For each of the three traits, the Gsw receives higher evaluations by at least two of the three subgroups JNSfi, HKIfi and VAAfi.

The two VAA subgroups are the only ones that are influenced in their perception of the Gf for the trait ‘influential’. The results from the matched-guise test reflect the linguistic composition of the towns in which the survey was conducted. In VAA, the only relatively large town in Finland with a comparatively high proportion of Swedish-speaking population, career perspectives are considerably better for bilingual people than for monolinguals. In such a bilingual municipality with a Finnish-speaking majority, only people speaking Swedish in addition to Finnish enjoy the prospect of working in a high position in a local or national institution or in a position involving customer contact or management in the private sector. The situation is different in JNS, a monolingual municipality, in which the Psfi are very unlikely to meet local Swedish-speaking people in general, let alone in high positions. In HKI, Ps of both language groups have the same common experience that Swedish-speaking people or bilinguals are not visible in official contexts since Finnish is the more usual language of communication and languages are not switched within a communicative situation. This interpretation is based on the understanding of Swedish-speaking people as at least functionally bilingual. This notion is very common, particularly in Helsinki and in fact in the whole Helsinki-Uusimaa region. Especially for the Pssw+ in HKI, language is unconsciously less considered a determinant of ←55 | 56→personal characteristics. This stems from the fact that an important part of the Swedish-speaking population is bilingual to a certain extent and that as part of the linguistic minority, they often find themselves in situations that demand to switch into Finnish.

To sum up, we find that the Gf ‘s Swedish guise was evaluated higher for each trait by at least two subgroups. Together with the finding that it was only the Swedish guise that was evaluated higher, this result reflects a high agreement among the Ps of all subgroups with regard to their perception of the Gf. Comparing the language groups, the Finnish-speaking subgroups stand out with an even higher agreement in their perception than can be identified across the groups of Swedish- and both Swedish- and Finnish-speaking Ps. As explained in chapter 2.5.1.6, the reason for this lies in the fundamental frequency of the female speaker’s voice that is perceived as unusually high by the Finnish-speaking participants.

2.5.1.8 Differences in influence.

The previous chapter first focused on the question of whether Ps of one subgroup perceive each G’s Swedish and Finnish guise differently for the various traits. Indeed, differences in perception were registered in all subgroups for both Gs. Based on this result, the question can be asked whether the differences in perception are the same across subgroups or if they are greater for some subgroups than for others. If the latter were the case, it would mean that the Ps’ perception was influenced to a varying extent by the G’s language according to their first language or the place of survey.

To find an answer to this question, this chapter is no longer about the comparison of the perceptions of the two guises of one G per trait and about checking the mean differences for statistical significance. Instead, I will now compare the mean differences per trait for each G across subgroups and calculate if they differ significantly from one another—see Fig. 7 for a visualisation of the calculation, and see Tables 15 for the differences included in this comparison.

In the previous chapter, differences in perception of the guises were identified for several traits, for both Gs and in all subgroups. This shows that the G’s perception is indeed influenced by the language the G uses. The manifestation of the influence can vary, however, with the Ps of one subgroup being influenced with regard to only certain traits and in a specific direction (i.e. either the Swedish or the Finnish guise is evaluated higher), while in another subgroup, the influence affects other traits. The influence may also be present for the same trait across subgroups, but of different intensity or working in the opposite direction. In these cases, it is possible to compare the influence for individual subgroups and to find out whether the difference in influence is statistically significant.

←56 | 57→

We can expect the language’s influence on the subgroups to be significantly different primarily for those traits that display opposite directions of influence. This is the case if, for one trait, the two subgroups compared give their respective higher evaluation to different guises of one G. Chapter 2.5.1.7 already showed that such an opposite direction of influence is noticeable for some of the traits, with subgroups being influenced in opposite directions by the G’s language. In addition, significant differences may be found in such cases where two groups give higher evaluations to the same guise, but where the mean differences in the two subgroups’ perceptions diverge considerably. Furthermore, significant differences are possible whenever the Ps of one subgroup perceive the guises markedly different, while there is no significant difference in the perception by another subgroup.

The comparison of the differences found for the five subgroups was carried out in SPSS with a one-way analysis of variance (ANOVA), comparing more than two independent groups by performing a series of comparison of pairs. The ANOVA produced five significant differences between the differences of perception for four traits. They are assembled in Table 6.

Table 6. Differences in the influence on perception, ANOVA results

←57 | 58→

As can be read from the table, differences are significant for cases in which the G’s language has a significant influence on the perception by two subgroups that works in opposite directions for the groups (①). In two other cases, significant differences are found in the comparison of pairs of which only one subgroup shows a significant difference in perception of the Gm ‘s guises (②). In the respective other subgroup, the difference is not significant, i.e. the Gm ‘s language does not have a significant influence on the Ps’ perception of the guises.

For the Gf, all subgroups agree in giving a higher evaluation to the Swedish guise—the direction of influence is thus the same, and the mean differences are too similar to be significant. As a result, significant differences can be identified exclusively for the Gm.

On the scale of ‘reliability,’ the higher rating goes to the Gm ‘s Swedish guise for the HKIfi Ps and to the Finnish guise for the VAAfi Ps. The differences diverge by 0.591 points. For the social trait, the Gm ‘s Swedish guise is evaluated higher by the VAAfi Ps, while it is the Finnish one for the VAAsw+ Ps. The differences diverge by 0.768 points. The differences between the mean evaluations compared here are all significant, but at less than 0.4 points, they are not relevant. Nonetheless, the subgroups are influenced to a significantly different degree by the Gm ‘s language because the influence works in opposite directions for the subgroups compared.

For ‘social,’ the Gm ‘s Swedish guise is evaluated higher also by the Psfi in HKI, contrasting with the VAAsw+ Ps. The differences diverge by 0.921 points. Both differences compared here are significant and for the HKIfi subgroup, the difference is also relevant. Taken together with the opposite directionality of the language’s influence on the Ps in the two subgroups, the differential between the differences represents the largest difference in this comparison.

Regarding ‘confidence,’ the Swedish guise receives a significantly higher evaluation by the HKIfi Ps that is also relevant. The perception by VAAsw+ participants, in contrast, is not marked by significant differences. Nevertheless, the differences of these subgroups diverge to the extent of 0.776 points, which is large enough to be a significant difference in the influence on the perception of the two subgroups.

VAAsw+ Ps perceive the Finnish guise as significantly more educated, while for the HKIfi subgroup, their perception is not significantly different for the Gm ‘s two guises. Even though the difference between the mean evaluations by the HKIfi Ps is not relevant, the differential between the differences compared is high enough, at 0.562 points, to qualify as a significantly diverging influence on the perception of the Gm ‘s guises by the respective subgroups.

←58 |
 59→

Looking at the results, it becomes clear that differences in perception exist not only along the lines of a single background variable that was used to divide the Ps into groups (Ps’ first language and place of the survey). For the reliability trait, the influence of the Gm ‘s language is indeed significantly different for the same language group (Psfi) compared for two places of survey (HKI and VAA). The first language variable then is important for the perception of the social trait by the VAA participants, with VAAfi and VAAsw+ being influenced significantly differently by the Gm ‘s language in their perception of the guises. In three other cases, however (for the traits ‘social,’ ‘confident’ and ‘educated’), the subgroups whose perception of the Gm ‘s guises is influenced with a significant difference are dissimilar both in their first languages and in the place of survey (HKI, VAA).

Moreover, differences in perception are visible not only in cases where two significant or even relevant differences between the mean evaluations are compared. Instead, we can see that the differences compared can diverge to a significant extent even if only one of the differences itself is significant.

Another remarkable finding is the fact that only the two subgroups HKIfi and VAAsw+ show significant differences in the influence on their perception; for three traits the difference is between these two subgroups and in one case each it is in comparison to another subgroup. In other words, it is primarily the perceptions by the participants in HKIfi and VAAsw+ that diverge strongly from one another. This can be explained, as mentioned several times above, by the very dissimilar experiences that Finnish-speaking participants have with Swedish-speaking population in HKI and VAA. It is also between these two subgroups that the largest significant differences can be identified for two traits (‘social:’ 0.921 points, and ‘confident:’ 0.776 points).

Comparing the VAAsw+ Ps with the two subgroups mentioned above, HKIfi and VAAfi, we find a difference in influence on their perception for one trait each (‘reliable’ / ‘social’). The participants in the subgroups HKIfi and VAAfi appear to be strongly influenced by the Gm ‘s language, resulting in high differences between the mean evaluations of the guises for the trait ‘social’ (see Table 2 and 5). A comparison of the differences shows that they do not diverge significantly, meaning that the influence on the two subgroups is similar in type and intensity. Looking at the VAAsw+ Ps, we see that they are influenced by the Gm ‘s language as well, but their perception is affected in the opposite direction. The difference of the VAAfi subgroup on the one hand and that of HKIfi and VAAsw+, on the other hand, diverge strongly enough to be statistically significant. Out of the three subgroups, it clearly is the perception by VAAfi participants that is influenced differently.

←59 | 60→

Furthermore, it is obvious that in most cases, it is the same subgroups whose perception is influenced significantly differently by the Gm ‘s language in comparison to the other Ps. The Ps of the JNSfi subgroup (which is the only subgroup in a monolingual Finnish-speaking town) and the HKIsw+ subgroup are not influenced significantly differently from any other subgroup. Accordingly, these Ps are not significantly more or less strongly biased towards the Gm than the Ps of the other subgroups. The absence of significant differences in the influence on perception can be explained by the relations between the evaluations on each scale. The effect becomes clear by looking at the statistical results: in total, the JNSfi Ps were influenced less strongly in their perception of the Gm ‘s guises, as reflected in the mean sum score that is among the lowest of all five subgroups (see Appendix 2.1). Moreover, the influence on their perception works in the same direction and for the same traits as in other subgroups, where Ps perceive the guises with significant differences (see Appendix 2.2). As a result, for this subgroup, no important differences can be found in any of the comparisons of pairs.

For the HKIsw+ subgroup, the similarity to other subgroups lies in the fact that it is the same Gm guise that receives a higher evaluation in most subgroups—the Finnish guise. In addition, the mean difference of the evaluation of the guises is too small to be significant for HKIsw+, but at the same time large enough to be on a level close to the other subgroups. This is why we find no major divergence between the differences in this subgroup and those of the other subgroups when compared in pairs. Another factor to be considered is group size. The HKIsw+ subgroup is comparatively small (see Fig. 4), and so differences would need to be greater than in other subgroups to be significant (see footnote 15 above).

In contrast to the two subgroups just focused on, there are at least two subgroups for which the influence on the Ps’ perception of the Gm ‘s guises was significantly different in more cases than for the other subgroups. It is the subgroups HKIfi, VAAsw+ and, in fewer cases, VAAfi.

For the two subgroups HKIfi and VAAsw+, the high number of significant differences can be explained by the fact that they both perceive the two guises differently for the same three traits (‘social,’ ‘confident’ and ‘educated’), but do not give their higher evaluations to the same guises. The direction of influence on their perception is thus opposite from the other groups in the comparisons of pairs. The influence on VAAfi Ps differs from that on the other two subgroups for only two traits (‘reliable’ and ‘social’), for which they give the higher evaluation to the Gm ‘s respective other guise.

The results presented in the previous chapter allowed the conclusion that within each subgroup, it was mostly or exclusively one of the Gm ‘s guises that ←60 | 61→received the higher evaluation, but that it differed across subgroups. The aim of the present chapter was to compare the subgroups with regard to the influence on the perception of the guises for each trait. It was statistically tested whether the five subgroups were (un-)biased towards the Gm ‘s guises to a significantly different degree.

The language that the Gm uses influences the Ps’ perception for a variety of traits and to differing extents across the subgroups. The Ps’ bias is linked to the subgroup they belong to, but it does not depend exclusively on either their first language(s) or the place of survey.

For the Gf, in contrast, all subgroups are influenced in a very similar way in their perception of the guises. The subgroups who perceive the Gf ‘s guises with a significant difference do so for mostly the same trait, and they all give a higher evaluation to the Swedish guise. In other words, the Gf ‘s language influences the Ps of all subgroups in the same direction. The divergences of the differences between the mean evaluations of the Gf ‘s Finnish and Swedish guise are too small to be significant. The conclusion that can be drawn from these results is that the Gf ‘s language not only has a stronger influence on the Ps of all subgroups than the Gm ‘s language has but that all subgroups are also biased against the Gf in the same way and to a similar extent.

It needs to be pointed out, however, that this result (i.e. that the influence on the perception of the Gf ‘s guises does not vary across subgroups) is taken only from the comparison of pairs of subgroups per trait. Regarding the sum scores (see Appendix 2.1), the extent of influence, measured purely in quantitative terms, does indeed differ, and quite considerably so in some cases.

2.5.2 Opinions on language policy

This chapter presents the participants’ (Ps) responses to specific questions and subquestions in the questionnaire. The choice of questions is centred around the Ps’ opinions on the Swedish language and speakers of Swedish in Finland. The first two questions concern the legal status of Swedish in Finland and ask the Finnish-speaking Ps (Psfi), which rights they personally would grant the Swedish-speaking population in Finland. The subsequent three subquestions are part of a larger question, which confronts the Ps with various statements on the relation between the language groups. The Ps are asked to put a cross beside those statements that they consider to be true. We will here look at the Ps’ agreement with three stereotypical statements that are connected to three traits from the semantic differential used in the matched-guise test. It is via these three statements and the three traits that a comparison of the direct and the indirect method is possible (chapter 2.5.3). ←61 | 62→Apart from constituting the points of comparison, these subquestions presented here also provide the context in which the Ps’ responses in the matched-guise test have to be considered. In the last question presented here, the Ps were asked to assess the relationship between the language groups.

2.5.2.1 The legal status of Swedish in Finland—no more than a label?

The first question in the questionnaire’s part on language policy was directed to all Ps and concerns the official status of Swedish in Finland. Already in Finland’s constitution of 1919 (FFG 94/1919 § 14), its first constitution as an independent country, Finnish and Swedish are recognized as the national languages. However, Finland’s bilingualism and the status of Swedish have never been uncontested. Currently, it is the Finns Party (formerly known as True Finns, Finn. Perussuomalaiset, Swed. Sannfinnlandärna) in particular, who question the country’s bilingualism or the role of the Swedish language in Finland. In part, they oppose the status of Swedish as a mandatory school subject and lately the Finns Party youth wing even objected to the understanding that Finland Swedes belong to the Finnish people (HS 12.1.2019). As the critical voices that oppose Swedish are very loud, it is intriguing to find out how to assess their importance against the background of a more comprehensive survey.

The responses to the question what the official status of Swedish should be in today’s Finland reveals a deep divide between the two language groups: in JNS (Fig. 19), the majority of Ps does not support the status of Swedish as a national language. Only 37 % are supporters of Finland’s bilingualism in its current form. A similar proportion of the Psfi is in favour of granting Swedish the status of a minority language. A fifth of JNS Ps would rather grant Swedish no legal status at all. In the bilingual towns that were surveyed, the Psfi have a more positive stance towards the status as a national language, but only in HKI is it a majority that supports this status (Fig. 20–1). Considerable percentages of Psfi in all three places of the survey are against any legal recognition of the Swedish language in Finland. To them, Swedish would have the same status as any other language in the world that has no connection at all to Finland.

Among the Pssw+, the highest agreement to the status quo is registered in VAA, where at 92 % it is even higher than in HKI (85 %), which can be explained by the notably higher proportion of Swedish-speaking population in VAA. In Vaasa, speakers of the minority language can use their first language almost anywhere in their everyday life, despite the clear minority position (22.6 % Swedish-speaking population). Finnish and Swedish appear to be indeed equal in many ←62 | 63→aspects of everyday life, which is why the Pssw+ do not find it plausible to grant Swedish only the status of a minority language. It is different in HKI, where Swedish can be used in only very few everyday situations and is considerably less present overall. Among the linguistic out-group, however, the percentage of Ps agreeing with the status of Swedish as a national language is higher in the capital than in VAA. This can be related to the common trend of out-groups in urban centres of culture or education having a generally more accepted stance towards linguistic and other minorities compared to out-groups in smaller towns and rural areas (see, for example, e2 2018: 31, for a self-assessment of the Ps’ liberalness of values on a scale).

A small percentage of JNS Psfi chose the response option that Swedish should be given a status different from that of a national or minority language, but should still have official status. The Ps’ comments provide no idea for such a different status, however. A positive correlation can be identified for the subgroups between the proportion of the linguistic majority population and the support for the status of Swedish as a national language among the Ps of the linguistic majority. If, however, the results of the Psfi in JNS were representative of the total Finnish-speaking population in monolingual Finnish-speaking municipalities, Swedish would not be supported by the majority of the population. Of the 311 municipalities in Finland, 262 are monolingual Finnish-speaking (kuntaliitto.fi 1).

Fig. 19. JNS—the status of the Swedish language in Finland

‘In your opinion, which legal status should Swedish hold in today’s Finland?’

nfi = 124. y-axis per cent of n answers.

Fig. 20. HKI—the status of the Swedish language in Finland

‘In your opinion, which legal status should Swedish hold in today’s Finland?’

nfi = 78; nsw+ = 41. y-axis per cent of n answers.

Fig. 21. VAA—the status of the Swedish language in Finland

‘In your opinion, which legal status should Swedish hold in today’s Finland?’

nfi = 91; nsw+ = 78. y-axis per cent of n answers.

←63 | 64→

Most of the Finnish-speaking Ps responded critically of the status of Swedish as a national language. But would they actually deny the Swedish-speaking population the linguistic rights that they currently enjoy thanks to the status of Swedish as a national language equal to Finnish (Language Act ←64 | 65→423/2003)?20 The Psfi were asked which rights they would grant the Swedish-speaking population depending on the linguistic status of the municipality. Should they be entitled to use Swedish, their first language, when communicating with state or local authorities as well as with private businesses? The Psfi also had the possibility to tick the response option that none of these should apply, i.e. that the Swedish-speaking population in a municipality with a certain linguistic status should not be entitled to use Swedish in their communication with any of the entities mentioned above (see Figures 224 for all response options and the municipality’s linguistic status). In those areas where the Language Act grants the Swedish-speaking population linguistic rights (inner box in Figures 224) a majority of 57–92 % of the Psfi in all three places of the survey would grant the Swedish-speaking people those same rights. We can identify commonalities between the responses by the Psfi depending on the linguistic status of a municipality and between the places of the survey.

In monolingual Swedish-speaking municipalities as well as in bilingual municipalities with a Swedish-speaking majority, 82–92 % of the Psfi in all three subgroups agree to allow the Swedish-speaking population to use their first language in the communication with state authorities. Contrary to what one would expect, the lowest percentage is found in VAA. In almost all subgroups, the percentages of agreement go down by about 12 % as soon as Swedish is the language of the minority of the municipality’s population. Only in VAA is the agreement roughly as high as for municipalities with a Swedish-speaking majority. For monolingual Finnish-speaking municipalities, different percentages of Psfi in the bilingual places of the survey would grant the Swedish-speaking population the right to communicate in Swedish with state authorities. It is highest in VAA at 76 % and considerably lower among the Psfi in HKI (62 %) and JNS (57 %).

Asking the Psfi about the communication with local authorities, the responses are similar to those to the previous question on state authorities. As long as Swedish is the language of the majority in a municipality, 80–91 % of the respondents in ←65 | 66→all three towns agree to allow the Swedish-speaking population to use their first language. For bilingual municipalities with Finnish as the majority language, the percentage drops by 10 % in VAA and, more markedly, by 20 % in HKI and 25 % in JNS. For situations for which the Language Act does not grant the Swedish-speaking population any linguistic rights in the communication with local authorities (in monolingual Finnish-speaking municipalities), 27 % of the Psfi in JNS and about 45 % of those in HKI and VAA would still allow them to communicate in Swedish.

The Language Act does not regulate communication between private individuals and private businesses. Nonetheless, high percentages of Psfi would grant the Swedish-speaking people the right to service in their first language in municipalities with Swedish as the majority language (of Psfi: 60–73 % in JNS and HKI, 67–76 % in VAA would grant the respective right). For municipalities with Finnish as the majority language, the percentages are considerably lower in all three subgroups (16–32 % in JNS and HKI, 22–40 % in VAA).

In all three places of the survey, only a small proportion of Ps would not grant the Swedish-speaking population in bilingual or monolingual Swedish-speaking municipalities any of the rights discussed here. The highest percentage can be identified, as one would expect, among the Psfi in the monolingual Finnish-speaking town of JNS (3–14 %) and the lowest among the Psfi in VAA (2–4 %), being the place of the survey with the highest proportion of Swedish-speaking population. For monolingual Finnish-speaking municipalities, the highest agreement to the response option of not entitling the Swedish-speaking population to communicate in Swedish in any of the contexts mentioned is registered again in JNS at 35 %. In HKI, the percentage is only slightly lower, at 29 %, but among the Psfi in VAA it is considerably smaller, at 15 %.

There is a close connection between the issue of the status of the Swedish language in Finland and that of the linguistic rights which speakers of Swedish should enjoy, depending on the linguistic status of the municipality. Despite these being intertwined, the Ps’ respondents to the respective questions differ considerably. While the majority of Psfi would not give the Swedish language the current status of a national language, large shares would still grant the Swedish-speaking population linguistic rights even for domains where the Language Act does not. This is true in particular for the private sector in municipalities with Swedish as the majority language. How can these contrary impressions of the Finnish-speaking population’s relation to the Swedish-speaking, as expressed in the Psfi ‘s responses, be explained?

The most striking difference between the two questions is their prominence in the discussion about aspects of Finland’s bilingualism. As mentioned in the introduction, the legal status of the Swedish language in Finland as an equal national ←66 | 67→language has been repeatedly contested in the course of history, although with different motivations at different times. The controversy over the legal status of Swedish becomes prominent in the media and political discourse every now and again, sparked by various events or issues (e.g. the discussion on the status of Finnish in Sweden) and as a result, most of the Ps already have an opinion on the issue. The opinion is ready to be stated, without the need to rethink the issue and considering various options. Thus, the responses to this first question presumably reflect how the majority of Ps position themselves in the debate based on the arguments presented in public discourse in favour of and against Finland’s bilingualism with both Finnish and Swedish as national languages. (See Bindrim 2019: 221–2, for more on these arguments and an assessment of their validity). The public discourse appears to have an encouraging effect on many Finnish-speaking people to question the status of Swedish as a national language in Finland. The fact that there is no collective memory of a debate that had led the population to the consensus of setting down Finnish and Swedish as equal national languages in the constitution may be another encouraging factor. Even in the latest important revision of the Language Act (in 2004), the public and the media were portrayed as a potential problem and tried to be excluded from the debate (Ihalainen/Saarinen 2014: 47). As a result, for most of the Finnish-speaking population, the mantra of Finland’s bilingualism remains an abstract ideology—partly due to the lack of personal experience with the bilingualism.

The legal status of a language has immediate consequences for the linguistic rights of the speakers of that language. However, the latter are not commonly part of the debate on bilingualism on the Finnish-speaking side. We can assume that the Psfi first had to reflect on the response option when answering the question on the linguistic rights depending on the linguistic status of a municipality. The fact that the responses were often more positive in comparison with those to the first question can be explained by considering that in a society striving for consensus and that for a long time has been understood as homogeneous and characterised by equality, the psychological hurdle of denying someone fundamental rights (as far as the Psfi are familiar with the linguistic rights) or plausible rights is higher.

It becomes apparent particularly with the high percentage of Psfi in VAA, who would grant the Swedish-speaking population linguistic rights in the private sector, that the Ps do not or only to a small extent, take the current legal situation as a reference for their response. Instead, they appear to base their responses on their personal experience in everyday life in VAA—service by private businesses, such as supermarkets, bars, cafés and restaurants, shoe and clothing shops etc. is given in both languages without problems. Despite the smaller proportion of Swedish-speaking population, Finnish is not considered to be the default language; instead, the staff uses Finnish and Swedish in their initial contact with customers, who then ←67 | 68→can answer in their respective first language. Swedish is not a marked linguistic option in VAA. In this respect, the two bilingual places of the survey, VAA and HKI, differ notably from one another, with the different linguistic situation being reflected in the responses by Psfi. The visual and auditive prominence of minorities contributes to reducing the markedness of the feature that connects the members as a minority community. With stronger prominence, the feature is no longer dominant and stops pushing a person’s other personal characteristics in the background.

In JNS in particular, there is virtually no contact between the language groups, because it is almost impossible there, considering the town’s homogeneous linguistic composition. It is, therefore, easier for the Psfi from JNS to deny the unknown other certain rights in their response to the survey question. Likewise, with the question on the language’s legal status it is easier not to think primarily of the language users, but to conceptualise the Swedish-speaking population in terms of their percentage of the total population. In addition, a large proportion of the Ps probably understands the concept of minority above all in terms of quantity. If, on top of that, respondents lack the awareness that the legal status of a language and the linguistic rights of the language users are immediately connected, it is highly likely that they consider the legal status of a language as a mere label. They might reject the legal equality of about 5 % of the population and the majority population for the reason that this ‘label’ does not reflect reality, in their view.

Fig. 22. JNSfi—granting linguistic rights

nfi = 124. Values: per cent of n answers. Legend: Outer box—domains included in the Language Act. Inner box—domains in which the Swedish-speaking population enjoys linguistic rights.

Fig. 23. HKIfi—granting linguistic rights

nfi = 78. Values: per cent of n answers. Legend: Outer box—domains included in the Language Act. Inner box—domains in which the Swedish-speaking population enjoys linguistic rights.

Fig. 24. VAAfi—granting linguistic rights

nfi = 91. Values: per cent of n answers. Legend: Outer box—domains included in the Language Act. Inner box—domains in which the Swedish-speaking population enjoys linguistic rights.

←68 |
 69→ ←69 | 70→

2.5.2.2 Stereotypes of the Swedish-speaking population.

Many of the stereotypes of the Swedish-speaking population in Finland can be traced back to the historical fact that for several centuries, the upper class of society in Finland was Swedish-speaking. During the time when Finland belonged to the Swedish Empire, many relevant institutions were established—the first bishopric was founded in the 12th century (arkkihiippakunta.fi 1), the first Dominican monastery was established in 1249 and with it the first schools (Ikonen 2015: 12). In the 17th century, the first Court of Appeals was set up (oikeus.fi 1) and the first university (Royal Academy of Turku, Swed. Kungliga Akademien i Åbo, Finn. Turun Akatemia; helsinki.fi 1) on Finnish territory was founded. Since the bishopric was moved to Turku in 1300, all of the institutions mentioned above have been located in Turku. They were all run by the Swedish Empire and were Swedish-speaking. When Finland became an autonomous grand duchy within the Russian Empire in 1809, Turku gradually lost its importance, with the capital being moved to Helsinki in 1812. After the Great Fire of Turku in 1827, the university, which was still the only university in Finland at the time, was moved to Helsinki as well (oikeus.fi 1). It becomes clear from this very short historical account alone that it were Swedish-speaking people (Swedes) who established the basic administrative structures of all important domains of society in Finland. What is often neglected, however, is that the converse interpretation of the whole Swedish-speaking population belonging to the upper class does not hold. On the contrary, the majority of them were fishers or farmers; later a working class developed (Saarela 2004: 84). The historical, socio-economic stratification with Swedish-speaking members of the educated class being on the top of the Finnish society is often projected onto all Finland Swedes up until today, even though this part of the population has always been very heterogeneous as well.

In a meta-study, the Canadian professor of Political Science Kenneth McRae analysed to what extent the Finnish-speaking and the Swedish-speaking population of Finland have actually differed in the course of time. In his study, he compares, among others, the standard of living and the socio-economic status of the two language groups in Finland. For practical reasons, he focuses on the 20th century, for which he could use already existing statistical data. McRae’s comparison shows how deep the divide between the language groups was at the beginning of the 20th century, but it also becomes clear how quickly the circumstances and standard of living have improved and markedly converged for the two language groups, especially since the 1970s. (McRae 1997: 141–2.)

Via the questionnaire in my study, the Ps were confronted with a number of well-established stereotypical statements about the Swedish-speaking population ←70 | 71→that have to do with the historical, socio-economic status of the latter. The Ps’ task was to choose those statements, out of eight statements in total, that are generally true in their opinion. In the following, the percentage of the Ps’ agreement to three of the subquestions is discussed:21

(6) Finland Swedes generally get a better education (schools, university etc.).

(7) Finland Swedes generally have a higher income.

(8) Finland Swedes disproportionately often hold senior positions.

The aim in asking for the agreement to certain stereotypical statements22 was not primarily to find out how high a proportion of the population considers them to be true (for this perspective see Folktinget 1997, Magma 2008, Samforsk 2014, e2 2017). Instead, the subquestions were intended to facilitate the comparison between external image and self-image: While the Psfi assessed the validity of the stereotype from the out-group perspective, the Pssw+ responded from the in-group perspective.

Comparing the agreement expressed in the questionnaire with the results elicited by the matched-guise test produces interesting findings (see chapter 2.5.3). First, however, the stereotypes are presented and against this background, the Ps’ way of responding, which varies considerably across the five subgroups, will be explained.

The percentage of Ps agreeing with all three stereotypical statements (Fig. 25–7) is small in all five subgroups: at 16.67 % the share is highest among the Psfi in the monolingual Finnish-speaking town of JNS. In both VAA subgroups, at 13.19 % (VAAfi) and 12.82 % (VAAsw+), respectively, the percentages are only slightly lower than in JNS. The high similarity between the percentages in the two language groups in VAA can be explained by the fact that Finnish-speaking and Swedish-speaking people are likely to have similar experiences living in this bilingual town. In HKI, 7.69 % of Psfi and only 4.88 % of Pssw+ consider all ←71 | 72→three stereotypical statements to be true. In this comparison, the place of survey emerges as the most important factor for the agreement, with the first language being less decisive.

The stereotype of Finland Swedes being better educated (No. 6) is considered true by 23 % (HKI) to 29 % (VAA) of Psfi in the three places of the survey. Among the Pssw+, the percentage is notably lower in HKI, at 17 %, while it is almost twice as high in VAA, at 35 % (see Figures 257).

In fact, Swedish-speaking secondary schools rank disproportionately high in the list of Finland’s best secondary schools.23 In spring 2015, for example, seven out of the 37 Swedish-speaking secondary schools ranked among the country’s 50 best secondary schools.24 Four of these are located in the capital region comprising Helsinki, Espoo, Kauniainen and Vantaa; and one is in Ostrobothnia, in VAA. The best Swedish-speaking secondary school of the capital region is on rank 9 in the 2015 list and the best in Ostrobothnia on rank 19. The top of the ranking has been occupied by the same Finnish-speaking secondary schools from Helsinki or the capital region in varying order for the past years. One may criticise the rankings for various reasons, but they are unquestionably prominent in the media twice a year and are firmly established in the Finns’ consciousness.

In the Finnish-speaking media news, the focus is on the finding which secondary schools made it to the first 3–5 ranks (again). The disproportionately high representation of Swedish-speaking secondary schools among the upper 25 % of the ranking is not a news topic, however. The second focus in the media is on the potential weaknesses of such rankings and the danger that they may be overrated (e.g. HS 25/05/2015, HS 30/05/2016).

The Swedish-speaking media, in contrast, for instance Hufvudstadsbladet for the exemplary year 2015 and the year after, reports on the performance of the secondary schools in general and in the Swedish-speaking regions of Finland in addition to presenting the highest-ranking schools (HBL 25/05/2015, HBL ←72 | 73→30/05/2016, yle.fi 25/05/2015 a; cf. yle.fi 25/05/2015 b). It is thus only in the Swedish-speaking media that schools are compared more or less explicitly by the language of instruction.25

Against this background, the Ps’ responses to the subquestions on Finland Swedes being better educated can be interpreted from two distinct perspectives. One possible interpretation is that it is primarily the Swedish-speaking media that give the Ps the impression that Swedish-speaking secondary schools (being seen as representative of education in other schooling types) were better. Secondly, the image of the historical Swedish-speaking educated class might have left traces until today.

←73 | 74→

The percentages of Psfi that consider the stereotype of the Finland Swedes being better educated to be true are very similar across the three places of the survey. From a question on the use of Swedish-speaking media, we know that it is very low across all age groups.26 It can be therefore ruled out almost completely that this could have influenced 23–9 % of the Ps. Consequently, the interpretation that the Psfi are influenced by the traditional stereotype is more likely.

The question arises, however, how it can be explained that in VAA compared to HKI, twice as high a proportion of Pssw+ consider the stereotype to be true? Neither has the centre of Swedish-speaking education ever been located in VAA (see the historical outline at the beginning of this chapter) nor are the local or regional Swedish-speaking schools at the top of the rankings mentioned. It is therefore unlikely that the Ps thought of the Swedish-speaking educational institutions in the municipality of VAA or the surrounding region when they ticked the questionnaire response, expressing agreement to the stereotype. It is moreover safe to assume that the Pssw+ both in HKI and VAA are influenced in their opinion by Swedish-speaking media. According to the survey, they use Swedish-speaking media considerably more frequently than the Ps in Finnish-speaking subgroups, as can be expected.27 This can be interpreted to indicate that a higher percentage of Pssw+ in VAA is informed about the good results of Swedish-speaking secondary schools by the publication of rankings in the Swedish-speaking media. This may at best help to explain the higher percentage of agreement among VAAsw+ Ps, but it cannot fully explain the large difference to the Pssw+ in HKI and their conspicuously lower percentage of an agreement to the stereotype. Instead, we need to take the hypothesis into account that the traditional stereotype of the Swedish-speaking educated class had an influence. As mentioned earlier, the centre of the Swedish-speaking education in Finland moved from Turku to HKI in the early 19th century. Part of the explanation for the Pssw+ in VAA may be the fact that those who consider the stereotype to be true do not think of the educational institutions in their region, but that instead the image of the Swedish-speaking educated class in HKI is so firmly established in their minds that they locate the best schools in the capital region. In addition, the best secondary schools—both the Finnish- and the Swedish-speaking ←74 | 75→ones—have rather been schools in the capital region than in VAA or Ostrobothnia for quite some time. The contrast between the rural region of Ostrobothnia and the urban capital Helsinki, which is the most important educational centre in Finland, may additionally reinforce this image. The Ps in HKI, on the contrary, make the assessment from their own perspective, with their own schools as the frame of reference and so they know, being more familiar with the schools in their region and their performance in the rankings, that the best secondary schools traditionally are Finnish-speaking ones.

The stereotype of Finland Swedes having a higher income (No. 7) is considered to be true by 24 % of Psfi in HKI, which is the lowest share, and by somewhat higher proportions in VAA and JNS at 32 % each. The percentage of agreement among the Pssw+ is considerably lower at 21 % in VAA and 21 % in HKI. In both language groups in HKI thus about 7 % to 9 % fewer Ps than in the other surveyed towns agree with the stereotype (see Figures 257).

In the questionnaire, the stereotype is worded as generalising as it is often expressed. But how does the Ps’ subjective evaluation relate to reality? In a study on this topic for the Finnish Institute of Occupational Health (Finn. Työterveyslaitos, Swed. Arbetshälsoinstitutet), Moilanen analyses data from 2007 of more than 1,500 Finnish persons, 4.3 % of whom are Swedish-speaking. In her study, no differences in income can be identified comparing Finnish- and Swedish-speaking people. However, the study shows differences in income by gender that are equally large in both language groups. (Moilanen 2010: 108.)

Moilanen’s results do not confirm the findings by the authors Saarela and Finnäs on differences in income between the language groups in Helsinki for the period 1987 to 1999.28 Their study found that in Helsinki, Swedish-speaking men have an income 17 % higher than that of Finnish-speaking men. For women’s income, the difference is only 2 %, with the Swedish-speaking women having a lower income on average. (Saarela/Finnäs 2004: 43–4.)

Saarela takes on a different approach to the data, widening the scope from Helsinki to all Swedish-speaking regions in Finland. The corresponding data analysis produces more differentiated findings. As a first general result, Saarela finds that the percentage of persons with a very high annual income (more than 150,000 FIM, corresponding to € 25,230 at the currency value of 1999) is higher among the Finnish-speaking than among the Swedish-speaking population. This ←75 | 76→is true both for men and women, albeit with a difference in income by gender between 15 % and 20 % for both language groups. (Saarela 2004: 83.)

In the next step, Saarela differentiates the incomes by region and looks at the data for the male inhabitants of HKI only. This shows that the distribution of high incomes varies strongly by region. In the capital region, the percentage of persons with high income is considerably greater among the Swedish-speaking men, while in Ostrobothnia it is higher among the Finnish-speaking men (Saarela 2004: 92). It becomes clear from the results of these studies that the stereotype of Finland Swedes having higher incomes can be disproved by a simple calculation of average, but that by closer examination it can be partly confirmed.29

Despite the fact that Moilanen’s data are considerably more recent than the data used by Saarela/Finnäs and Saarela, and although they could be a plausible consequence of the convergence of the socio-economic status of the two language groups in Finland as pointed out by McRae in his meta-study—despite all this, Saarela/Finnäs’ and Saarela’s findings are more meaningful for several reasons. Firstly, their studies are based on comprehensive statistical data (stat.fi 3), and secondly, the more in-depth analysis can shed light on the effects that neutralise each other. In addition, the studies by Saarela/Finnäs and by Saarela provide several potential reasons why it can actually be expected that the Swedish-speaking people have a higher income on average than the Finnish-speaking population (for the investigation period 1987–99). The reasons include the averagely higher education, older age, a higher degree of bilingualism as ←76 | 77→well as the strong regional family rootedness (Saarela/Finnäs 2004: 44; Saarela 2004: 88). Some of the reasons that Saarela/Finnäs present for higher average incomes among the Swedish-speaking population are as valid today as they were at that time, such as the higher degree of bilingualism and being very firmly rooted in a specific region. According to the data by the centre for statistics for 2010 (stat.fi 4) and 2013 (stat.fi 5), the Swedish-speaking population was educated on a higher level, on average, than the Finnish-speaking population (stat.fi 6; stat.fi 7).30 Saarela’s and Saarela/Finnäs’s findings are thus to be considered more reliable than Moilanen’s results, despite being based on older data, and in part confirm the stereotype of the richer Finland Swedes.31

According to Saarela’s study, the stereotype of the Finland Swedes having higher incomes is true at least regionally speaking. Why, however, is the percentage of Ps considering the stereotype to be true smaller among the Pssw+ in the bilingual places of the survey than among the three Finnish-speaking subgroups? And why is it even smaller in HKI than in VAA? What is likely to be a reason for the low agreement among the Pssw+ subgroups is that the statement of Finland Swedes having higher incomes is one of the strongest stereotypes. For this reason alone, the Ps of the linguistic in-group would not want to confirm it when directly asked for their agreement. It would bring them into a situation in which they could feel obliged to justify themselves. At the same time, the stereotype may cause a feeling of injustice among the Ps of the linguistic out-group, who are therefore more inclined to agree with the statement.

The stereotype that Finland Swedes are overrepresented in senior positions (No. 8) is considered true by 30 % of Psfi in JNS and by 36 % in VAA. In HKI, the proportion is considerably lower among the Psfi, at 22 %, and slightly lower than among the Pssw+, at 20 %. In VAA, 23 % of the Pssw+ consider the stereotype to be true (see Figures 2527).

←77 | 78→

In her study, Moilanen finds that the language groups are distributed unevenly across socio-economic groups. Of the Swedish-speaking Ps in her survey, higher proportions are farmers and other entrepreneurs or employees, while more of the Finnish-speaking Ps are represented in the working class. Moilanen (2010: 107) did not explain these differences by language, however, but by differences in the average level of education, among others. Saarela (2004: 93) and Saarela/Finnäs (2004: 44) identify a higher socio-economic status of the Swedish-speaking population for HKI as well.

An analysis conducted by the Finnish newspapers Helsingin Sanomat (HS 29/08/2011) in 2011 showed that at the time of the investigation, more than 23 % of Finnish board members of the 50 largest companies listed at the Helsinki stock exchange were Swedish-speaking.32 The Swedish-speaking board members who were interviewed for the analysis see one of the reasons for the disproportionate representation of Finland Swedes in the executive boards of major companies in the historical situation that it was the Swedish-speaking bourgeoisie in Finland that owned large trading companies and dominated trade. Moreover, the German- and Russian-speaking people doing business or immigrating into Finland at the turn of the 20th century, assimilated to the Swedish-speaking upper class. An additional factor brought forward is that still today, the share of people deciding to pursue a career in the economic, legal or political domain was disproportionately high among the Swedish-speaking population. The fact that property is disproportionately owned by Swedish-speaking people is explained by the historically high social status—accompanied by what is called old money—the impacts of which are relevant still today and by the close social network of Finland Swedes.

The percentages of agreement with the stereotype of Finland Swedes’ overrepresentation in senior positions vary considerably across the five subgroups. They make up a consistent picture, however, as soon as the correlations with the prominence of Swedish in the surveyed towns are taken into account. For bilingual towns, it is safe to assume that the Ps tend to get a more complete impression of how often it is Swedish-speaking persons that are in senior positions. State authorities and also local authorities in bilingual communities have certain linguistic obligations towards the two language groups. In consequence, language skills in the respective minority language are required for many ←78 | 79→positions, and for senior positions in particular.33 Whether and to what extent these language skills can actually be used depends greatly on the municipality’s linguistic composition. Individual bilingualism is more common in VAA than in HKI, with the linguistic environment more frequently demanding to switch between Finnish and Swedish. Individuals that are indeed bilingual, whether they are in senior or other positions, can be perceived as bilinguals more easily in VAA than in HKI. In the Helsinki-Uusimaa region (Uusimaa/Nyland), (functional) bilingualism is probably common mostly among people with Swedish as their (main) first language.34 Originally Finnish-speaking persons are probably found more frequently in senior positions due to the region’s demographic composition alone, and they are more likely to reach the respective qualification by learning Swedish institutionally, e.g. as part of the vocational training or university studies. If Ps identify people speaking Swedish as Finland Swedes, they meet them more frequently in VAA than in HKI, regardless of their professional position. This is an explanation for the agreement with the stereotype in VAA. In the monolingual Finnish-speaking town of JNS, the situation differs dramatically. The Ps have no contact with Swedish-speaking people neither on the local level nor in the private domain. It is, therefore, more probable that the stereotype is activated, which is then projected onto all Swedish-speaking people, based on the few cases of Swedish-speaking persons in high positions.

Summarising the results, we find that in no subgroup a majority of Ps considers one of the stereotypes to be true. Only in one case do more than a third agree with one of the statements. In both bilingual places of the survey, the percentage of agreement is higher among the Psfi than among the Pssw+, and likewise, the percentage among the Psfi in JNS is above that among Pssw+. The only exception to these two observations is the VAAsw+ Ps’ agreement with the stereotype that Finland Swedes generally get a better education.

Pssw+ agree less with the stereotypical statements. This can probably be explained by considering that all three statements imply that the linguistic ←79 | 80→in-group was privileged in comparison with the majority population. Privileges per se represent injustice—something that members of the in-group would like to avoid to take on them because it would put them under pressure to justify themselves. This becomes especially clear with the statement on the higher income, which receives the smallest percentage of agreement of Pssw+ in one town.

For all three statements, the agreement is lowest among the Ps in HKI. In both language groups, a smaller proportion of Ps agrees with the stereotypes than in the respective language group in JNS and VAA. In VAA we find the highest agreement with the three statements. The most probable explanation for this response pattern is the varying visibility of Swedish-speaking people in the two bilingual towns, being considerably higher in VAA than in HKI. It is likely that in HKI, bilinguals are frequently not identified as such, because in the public sphere Finnish is the more obvious choice in contexts potentially involving both languages, due to the demographic composition of the population.

Biographical notes

Stephan Kessler (Volume editor) Marko Pantermöller (Volume editor)

Stephan Kessler holds the chair of Baltic Studies at the University of Greifswald. His research topics are in linguistics as well as in literary studies. Marko Pantermöller is professor of Finnish philology at the University of Greifswald. He puts the main stress on language policy, contact linguistics and morphology.

Previous

Title: The Social Status of Languages in Finland and Lithuania