<<

Dialect Change in Western – a comparative study of the project Processes of Change

Lieke Maier 10534245

MA thesis (18 ECTS) MA Language and Society University of August 2019

Supervisor: prof. dr. A.P. (Arjen) Versloot Second reader: prof. dr. P.P.G. (Paul) Boersma Number of words: 19.935

Table of contents

1. Introduction ...... 3 2. Background ...... 3 2.1 Dialect areas ...... 4 2.1.1 Western Norwegian ...... 7 2.2 Previous dialect studies ...... 9 2.3 Processes of Dialect Change-project ...... 10 2.4 Hypotheses and predictions ...... 11 3. Method ...... 13 3.1 ...... 13 3.2 Classification of informants ...... 23 3.3 Linguistic variables ...... 24 3.4 Presentation of linguistic variables ...... 25 4. Analysis ...... 40 4.1 Category A ...... 41 4.1.1 Traditional vs a new variant ...... 42 4.1.2 High-status variety vs low-status variety ...... 44 4.2 Category B ...... 46 4.2.1 Traditional vs new variant ...... 46 4.2.2 High-status variety vs low-status variety ...... 49 4.3 Category C ...... 49 4.3.1 Traditional vs new variant ...... 50 4.3.2 High-status variety vs low-status variety ...... 51 4.4 Categories per linguistic variable ...... 54 5. Discussion ...... 68 6. Conclusion ...... 70 References ...... 70 APPENDIX ...... 74

2

1. Introduction

Dialect change in Norway and especially the question in which direction the change is going has in the last decades been the subject of a number of books and studies, and the interest for this subject is still undiminished. In recent years, several dialects in have been studied and analysed within a project called Dialektendringsprosessar, ‘Processes of Dialect Change’ with the aim to get a better insight into the actual changes in the Western Norwegian dialect area. In order to reach this goal, dialect studies have been carried out in places where a similar study was carried out a generation ago. The results from the two studies are compared to see how the dialects have developed over these 30 years (Sandøy et al. 2007: 4).

The results of the studies within the Processes of Dialect Change-project serve as the source for this paper. In the present study it is not one particular dialect, but rather the complete set of results that are used to answer the following question: what is changing in the language use of speakers of Western ? Do the same linguistic variables turn up in different dialect studies? Is the change happening at the same pace or are some dialects ahead in the process of change?

The structure of the paper is as follows. First, a background is given on the development of Norwegian dialects and important research that has been carried out, followed by a more extensive description of the project Processes of Dialect Change. Next, the hypotheses and predictions are presented, followed by the method that is used in this paper. As a consequence of using data from studies from the project, some methodological choices have already been made. The method elaborates on these methodological choices and how these issues are tackled. Then, the analysis is given, in which three categories are presented that each represent a different stage in the process of change of a linguistic variable. This chapter also raises the question on the direction of the changes within the dialects and the possible forces behind the changes. Lastly, discussion and conclusion complete the paper.

2. Background

In this chapter a background is given on the Norwegian dialects. Firstly, 2.1 discusses the history of the dialects and the classification of the different dialect areas. 2.2.1 expands on the dialect area that stands in focus for this thesis, namely Western Norwegian dialects. Then, in 2.2 some previous relevant dialect studies are mentioned before turning to the project Processes of Dialect Change in 2.3, followed by hypotheses and predictions in 2.4.

3

2.1 Dialect areas It is difficult to put a number on how many dialects there are in Norway. From dialect to dialect there are no clear boundaries as to where one dialect stops and the other one begins. Single linguistic features have clearer isoglosses and are easier to define (Jahr 1990: 10). These isoglosses reflect the linguistic developments that affected the in the .

In the 13th century, was more or less a homogeneous language, but around 1600 the dialect areas were established as they currently still exist (Berg 2018: 165). Within a few hundred years, a number of far-reaching linguistic changes happened. In Old Norwegian there were already some geographical differences, mostly phonological, that were typical for Western or Eastern Norwegian. For example, consonant compounds were realized as mn, ft and fs in (namn, eftir, refsing), and as fn, pt, and ps in Western Norway (nafn, eptir, repsing) (Berg 2018: 168). In the Middle Ages more considerable changes happened that affected the language structure at a deeper level. Most of the changes discussed below separate Eastern Norwegian from Western Norwegian.

Phonology

An important change that triggered many other developments is the quantity change. Old Norwegian had four possible stressed syllable combinations: a short vowel with a short consonant, a long vowel with a short consonant, a short vowel with a long consonant and a long vowel with a long consonant (VC, V:C, VC: and V:C:). In the quantity change, short syllables were lengthened and the overlong syllables were shortened, so that the options VC and V:C: disappeared. In general, in Western and Southern Norwegian the vowels were lengthened (V:C - /ve:t/) and in Northern and Eastern Norwegian, and also in the Western Norwegian , the consonants were lengthened (VC: - /vet:/), and in the most northern part of Norway, both options were possible (Berg 2018: 171).

One development that is crucial in the division of eastern and western dialects is level stress. In Eastern Norwegian, disyllabic words with a short consonant in the root syllable were under the effect of level stress, with an even stress on both the root syllable and the ending syllable. Because of this, the ending vowel was maintained and could even be lengthened. Disyllabic words with initial stress reduced the ending vowel to -e, or had apocope. In most modern dialects, level stress is no longer effective, but it can be traced back in the endings of infinitive verb endings and weak feminine substantives. In two areas in Norway the ending vowel was reduced to -e, independently of stress patterns. This happened in Northwestern Norway and in , probably around 1400 (Berg 2018: 174). The ending vowel was reduced in Bergen as well, but already a few centuries

4 earlier. The dialects with a reduced ending vowel are called e-varieties (e-mål), and the dialects that retained the a-ending are called a-varieties (a-mål).

A number of other phonological changes happened during the Middle Ages that affected the quality of consonants. Palatalization of velar consonants before a front vowel affected all dialects (gera > /jera/, kenna > /ҫen:a/). Palatalization of velars could also be found in medial position but was less widespread than in initial position (Berg 2018: 180). Another type of palatalization is palatalization of alveolar consonants that is typical for trøndersk, but it is also found in the surrounding northwestern dialects, most of northern-Norwegian and the counties and . long l and n became palatalized in words such as ball > /baʎ:/ and mann > /maɲ:/ (Berg 2018: 180).

A sound that developed from Old Norse short /l/ is the thick l, a retroflex flap that is found in Eastern Norwegian, trøndersk, romsdalsk and most of Northern Norwegian. This is also a feature that divides Western and Eastern Norwegian (Mæhlum & Røyneland 2012: 36).

Assimilation, a process in which two consonants become (more) alike (rl>ll, rn>nn, rs>ss, ld>ll, nd>nn, mb>mm), affected many dialects in the . The opposite of assimilation, differentiation, in which two consonants become more unalike (rn>dn, rl>dl) is specific for southwestern dialects (Berg 2018: 175). Inside of the region that has differentiation, a smaller area also has segmentation, a process in which two identical consonants become two different consonants (nn>dn, ll>dl) (Berg 2018: 182).

Morphology

Old Norse had a four-case system with nominative, genitive, dative and accusative. Some dialects retained a functional case system with four cases up until the 16th century, but in most dialects the simplification of the case system started in the 14th century (Berg 2018: 170). Firstly, in substantives, the cases nominative and accusative merged. This resulted in another difference between East and West, since, in singular indefinite nouns, western dialects retained the nominative form and Eastern dialects retained the accusative form. Active use of genitive nouns had in the beginning of the 20th century already completely vanished as a productive case (Larsen 1906: 108) and can only be found following certain prepositions. The dative, however, was in the beginning of the 20th century still full- fledged in certain dialects, but its use is currently decreasing (Skjekkeland 2005: 105).

5

Dialect areas

On the basis of some of the linguistic features discussed above, the dialects can be divided into several dialect areas. Linguists do not agree on how many dialect areas there are. There is consensus about the clear difference between eastern and western Norwegian dialects, but the difficulty lies in placing the northern dialects, since they have a lot in common with both eastern dialects and western dialects. (1864) was the first to make a dialect division. In his view, there was a western, an eastern and a northern dialect area. Aasen used level stress and thick l as criteria for separating west and east. The existence of apocope in the northern dialects was enough reason for Aasen to make a separate group for the northern dialects. At the end of the 18th century, a division between Western and Eastern dialects became dominant among linguists. The first to introduce this two-way-division was Amund B. Larsen (1897), with the northern dialects as part of the western area, although he was careful in placing the northern dialects with either the west or the east, since the northern dialects have as many salient, important features in common with the western dialects as they do with the eastern dialects (Skjekkeland 2005: 150). According to the view of other linguists after him, the northern dialects were in this two-division best seen as part of the western dialect area, a view that stayed dominant until the mid-twentieth century (Skjekkeland 2005: 150).

Up until that point, classification of Norwegian dialects remained based merely on the two criteria suggested by Aasen (1864): level stress as the most crucial one, and thick l as a second factor. Hallfrid Christiansen (1954) did not agree with using only two criteria and nor with placing the northern dialects together with the western dialects. Christiansen suggested that it did not make sense, geographically, to group the Northern dialects with the Western dialects. She proposed a division in four areas, made on the basis of a range of linguistic features such as apocope, tone and palatalization of velar consonants in medial position (Skjekkeland 2005: 151). In the division in four areas, the northern area based on Aasen’s division is split up in two: trønder dialects and northern dialects. In a division in four areas, the borders more or less match the administrative and geographical borders. An exception is Sørlandet, Southern Norway, which administratively is a separate area, but that is part of the western area in the dialect division.

In this thesis, ‘western Norwegian dialects’ indicates dialects in Western-Norway, using the division in four areas as proposed by Christiansen (1954), following Mæhlum & Røyneland (2012). And although there are key characteristics that group Western dialects together, they still differ on a number of grounds. Therefore, a background is given on the different subareas that exist in the group of western Norwegian dialects and the defining differences between them.

6

2.1.1 Western Norwegian The western Norwegian dialect area can be divided in three sub-areas: northwestern dialects, southwestern dialects and southern dialects. The northwestern region covers the counties Møre og and og Fjordane, the southwestern region covers the counties , and Vest-, and Aust-Agder belongs to the southern region. There are several dialectal differences between the three regions, but also within these regions the dialects differ, which is explained in the paragraphs below.

In this thesis, no dialects are discussed that belong to the southern region. Therefore, only the northwestern and southwestern dialect area are Figure 1 - Dialect areas in Norway (adopted from discussed below and the southern characteristics Mæhlum & Røyneland 2012: 179). are not expanded upon.

Northwestern dialects

Three dialect areas can be distinguished within the northwestern area, to know , Sunnmøre and Fjordane. What bounds them together as a group is that they are all e-varieties, which means that in both infinitive verb endings and weak feminine substantives the ending is on -e. A second common feature is the palatalization of alveolar consonants in both strong and weak position (maɲɲ, maɲɲeɲ). However, this is a feature that seems to be getting out of use among the youngest generation. The palatalized consonants are being replaced by long alveolars (mann, mannen) (Mæhlum & Røyneland 2012: 95). Other features that formerly helped to differentiate between the three northwestern dialect areas are also in decline. Split feminine inflection could be found in both Sunnmøre and Romsdalen, with an ending on -å in definite form singular for weak feminine nouns and an ending on -a in definite form singular for strong feminine nouns. Especially in Sunnmøre this feature is no longer stable. Palatalization of velar consonants /g/ and /k/ in medial position is another example of a formerly common feature that is disappearing (Mæhlum & Røyneland 2012: 96).

Some other features have proven to be more stable across dialects. While in most dialects the consonant compounds nd, ld, mb and ng are assimilated, in Sunnmøre and Fjordane both consonants of the compound are pronounced. Another example, also found in Sunnmøre and Fjordane, is the

7 maintenance of the epenthesis in present tense of strong verbs and in adjectives (kje:me (pres.), fi:ne (adj.)). In present tense endings of weak verbs, the -r is remitted (kaste) (Mæhlum & Røyneland 2012: 96).

Some features are spreading outside of their original area of use. Romsdalen borders on the trønder dialect area, which can be traced back in the dialect. Romsdalen has a thick l and retroflexes, which are typical features for the trønder dialect area, with the difference that Romsdalen only has thick l stemming from Old Norse /l/, not from Old Norse /rð/. The occurrence of thick l is limited to Romsdalen only, but the retroflexion of consonant compounds r + d/l/n/s/t is finding its way into the other northwestern dialects, especially in Sunnmøre (Mæhlum & Røyneland 2012: 95).

In the use of pronouns, there are a lot of differences between the different dialects in the northwestern area, all variants of the personal pronoun not starting with j-. Romsdalen in the north has i: for first person singular, Fjordane most south has e:(g), and in between these areas, Sunnmøre has e:, ai or æi. In the first person plural the areas also have different forms, with me as the most common form in Romsdalen, vi in Fjordane, and again most variation in Sunnmøre, where both vi, me and oss can be found, although vi is most frequently used (Mæhlum & Røyneland 2012: 96).

Southwestern dialects

The southwestern dialect area is made up out of four regions, from north to south Indre Sogn, (with and Indre ), and Rogaland (with Vest-Agder). All dialects in this area are a-varieties with endings on -a in infinite verbs and weak feminine substantives (with the exception of Bergen that has endings on -e). The personal pronoun is the same throughout the region, with eg for singular and me for plural. The exception to this can be found in the that also have high-status variants jæi and vi. Another feature that is found in the whole southwestern region is the skarre-r, or uvular r, which has been expansive over the last century, first in the western coastal cities and currently spreading to more rural southwestern dialects (Røyneland 2009: 23).

A difference with the northwestern region is found in the historic development of long alveolar consonants. While they got palatalized in northwestern dialects, in the southwestern dialects long alveolar l and n were segmented in dl and dn ( > fjedl, finna > fidna). A feature that is closely linked to segmentation is differentiation, traditionally also a common feature in southwestern dialects. Through differentiation, two different consonants become more distinctive (barn > badn, karl > kadl) (Mæhlum & Røyneland 2012: 93). Both segmentation and differentiation are now used at a lesser extent by the youngest generation (Mæhlum & Røyneland 2012: 97).

8

Within the southwestern region, certain features are characteristic for specific areas. Split feminine inflection is part of dialects in Indre Sogn, Nordhordland, Voss and Hardanger, although it is used less and less. This feature also exists in some northwestern dialects, but the endings in the southwestern dialects are different, with -i/-ei in Indre Sogn and -æ/-e in Nordhordland, Voss and Hardanger. Dialects in Sunnhordland and Rogaland never had split feminine inflection, and have the same ending for both weak and strong feminine definite singular forms, either -o or -å (Mæhlum & Røyneland 2012: 97). Together with northwestern Sunnmøre and Fjordane, southwestern dialects have epenthesis in the present tense of strong verbs on -e (kje:me, sø:ve), but dialects in Voss and parts of Hardanger have an epenthesis -u (kje:mu, sø:vu). In present tense of weak verbs, in most of the southwestern dialects the endings-r has disappeared (kasta) with the exception of Hardanger, Bergen and parts of that have maintained the -r (kastar). Dialects in the southern coastal area, from southwards, have lenition of p, t, k after a long vowel, softening the consonants to b, d, g. But again, this is a feature that seems to be disappearing, especially in the cities and towns (Mæhlum & Røyneland 2012: 98).

Within the southwestern dialect area, Bergen forms a big exception. Contrary to the surrounding dialects, Bergen is not an a-variety but an e-variety. Also, while Norwegian dialects in general have three genders, Bergen only differentiates between masculine and neuter, feminine words have morphologically merged into masculine. Other deviations are the ending of weak a-verbs in the past tense and participle on -et, contrary to the ending on -a in the surrounding dialects (kastet / kasta), no umlaut in present tense of strong verbs (kåmmer; så:ver / kje:me; sø:ve) and the lack of segmentation or differentiation (Mæhlum & Røyneland 2012: 99). These are all features that are also found in East Norwegian dialects.

2.2 Previous dialect studies Norwegian dialectology has a long tradition of studying its wealth of dialects, starting with Ivar Aasen in the mid-19th century. These were diachronic studies, often trying to uncover the ‘real’, traditional dialect. Language varieties spoken in the cities were mostly neglected in dialectology. Exceptions are the studies in (Larsen 1907), Bergen (Larsen & Stoltz 1911/1912) and Stavanger (Larsen & Berntsen 1925) in the beginning of the 20th century, but until the 1970’s, the focus was mainly on rural dialects. In the 1970’s there was a turn towards sociolinguistic studies. Variation in language use between speakers of different age, gender, social class and other social factors became a central point of investigation (Skjekkeland 2005: 18). With the advent of sociolinguistics, the interest in urban dialects grew and several quantitative project were set up, such as Talemålsundersøkelsen i Oslo (TAUS 1971-1976) (‘Language study in Oslo’) and Talemål hos ungdom i Bergen (TUB 1983-1984) (‘Language study among adolescents in Bergen’). More recent projects are national projects such as

9

Talemålsendringar i Noreg (TEIN 1999-2002) (‘Language change in Norway’) and Utviklingsprosesser i urbane språkmiljø (UPUS 2005-2009) (‘Developments in urban language communities’) and also Dialektendringsprosessar (‘Processes of dialect change’), which will be elaborated on in 2.3.

While many other European show a clear tendency towards a standard spoken variety, in Norway, dialects are widely used and excepted, and there is no official spoken variety. There is an ongoing debate about the assumed existence of an unofficial standard variety. Mæhlum (2009) pleads for the acknowledgement of south-eastern Norwegian as the standard variety, while Akselberg (2009) feels that ‘standard variety’ is not the right term for the Norwegian linguistic situation (Akselberg 2009: 75). Sandøy (2009) recognizes that the south-eastern variety has obtained the mental function of a prestigious variety, but has trouble calling it the standard variety. The south- eastern is the cultural dominant variant but although Mæhlum emphasizes the differences between a source of influence and the outcome of that influence (Mæhlum 2009: 19), Sandøy has never found data that validate south-eastern Norwegian as a possible source of language change other than for lexical items (Sandøy 2009: 42). The debate about the standard variety is tightly linked to the discussion about the development of the dialects and the driving force behind the changes and whether this is a process of regionalization or standardization (Sandøy et al. 2007: 2). Mæhlum means that the source of the dialect changes is the ‘standard’ south-eastern variety and calls the process standardization, but the outcome regionalization, since the outcome of the change is different because of the linguistic differences in dialects (Mæhlum 2009: 20). Linguistic regionalization is a process of dialect leveling, in which geographical limited features are replaced by regional or standard features (Røyneland 2009: 9). The local dialects are not disappearing, but local features are used less and less, while regional and national features are expanding (Akselberg 2005: 1716). There is still a lot unclear about the driving forces behind dialect change in Norway, and getting more insights in these processes is a hot topic among linguists.

2.3 Processes of Dialect Change-project The main goal of the Processes of Dialect Change-project is to “develop insights into the modern processes of dialect change and the relation between changes in society and language change” (my translation) (Sandøy et al. 2007: 1). The initial aim was to analyse four dialects, two of which should be Bergen and Stavanger, but the project resulted in fifteen dialect studies between 2010 and 20181.

The innovating aspect about the project is that it analysed dialect changes in ‘real time’ (actual changes of a generation), as opposed to ‘apparent time’ (when changes of a generation are considered to be equal to changes over time) (Sandøy et al. 2007: 4). Over the last decades, several

1 https://folk.uib.no/hnohs/DEP/ 10 individual, solid dialect analyses were carried out, and the project took this source of data as the starting point for the new dialect studies. Every study within the project studied a dialect of which a description of the last decades was available. In the selection of the informants, the aim was to have several participants that also participated in the previous study. Comparing the old data with the new data created the opportunity to study the dialect changes in real time.

There were four different approaches to studying the dialect changes in real time, each analysing a different aspect of language change:

A – Language change and society typology;

B – Language change and life stories;

C – Regionalisation, levelling, standardisation, simplification;

D – Linguistic consciousness.

A takes a macro-approach to the analysis, studying the relation between dialect societies and language change by making a typology of the society. Its research question was whether certain societies have better conditions for quick changes than others (Sandøy et al. 2007: 4). B adds a more qualitative notion to the overall quantitative project. In this micro-approach the life stories of informants were taken into consideration to explain the language change of individuals over generations (Sandøy et al. 2007: 6). C aims at finding patterns that explain why dialects in different parts of the country have different processes of language change. Finally, D concentrates on the possible role of culture and unconscious assumptions about dialects as an influence on the way people talk and the change of their dialect (Sandøy et al. 2007: 7).

In Method, the way the project is used for this thesis is explained with regard to dialects, informants and linguistic variables.

2.4 Hypotheses and predictions Based on the findings of earlier studies, there are several expected outcomes. The first two hypotheses are expectations about the linguistic process of language change and the latter two concern external influences on language change.

H1: Geographically limited features are decreasing in use in favour of more regional or national features.

Language contact often leads to simplification and levelling of the language varieties. In the levelling of dialects, linguistic features that are specific to a certain dialect are likely to be replaced by features

11 that are used in a wider geographical area, either regional or national. The supposition is that the new variant of the linguistic variables is a regional or national variant.

H2: Linguistic variables from deeper linguistic levels are changing at a slower pace than variables from lighter linguistic levels.

The expectation is that there is a difference in the pace in which linguistic items from different linguistic ‘levels’ go through the process of language change. There is an expected hierarchy in which variables from lighter and more salient linguistic levels such as lexical items are more amenable to language change than variables from a deeper, more structural level, such as morphological items.

H3: Urban dialects are further ahead in the process of dialect change than rural dialects.

According to Trudgill, the amount of contact impacts the rate of change, “with higher levels of contact leading to faster rates of change” (Trudgill 2011: 6). Since the amount of contact is higher in urban areas than in rural areas, variables in urban dialects are expected to be in a further stage of language change than the same variable in rural dialects.

H4: Cities or towns with a centre function influence the direction of dialect change of the periphery.

The fourth hypothesis supposes a hierarchy in the process of language change, in which language change starts in the centre and then spreads to the surrounding areas. A centre offers services such as jobs, school and shopping, and is a central point for communication in a region. H4 hypothesizes that this central role also extends to language and that linguistic changes, starting in the regional centre, are likely to be adopted by the surrounding periphery.

12

3. Method

3.1 Dialects This chapter presents a short historical and demographic background of each of the 13 dialect locations. How the dialect studies are distributed over the Western Norwegian region is illustrated in Figure 2. The dialect studies are not evenly distributed over the region. There is a noticeable gap in , where not one single dialect study was carried out, while in other regions several dialects were studied in close geographically vicinity. This is the case for , Midøya and Brattvåg and Hildre, a bit further south for Øygarden, Bergen and , and thirdly for , Stavanger and Hå.

Chapter 2.1.1 elaborated on the Western Norwegian dialect area, mentioning different subarea’s within the overarching Western Figure 2 - The distribution of dialects over the Western Norwegian dialect area. How the dialects from Norwegian region. the studies from the project are divided over these subareas is listed in table 1.

From north to south, every location will now be introduced with the aim of illustrating the kind of community the dialect is spoken in. On the one hand, there is the kind of community that has a central function and offers certain services, that people travel to. On the other end there are the peripheral communities, that have to travel to a centre for services such as shopping, work and administrative functions (Sandøy 2003: 232-233).

13

Table 1 - List of dialects.

Dialects from Dialect region Dialect subregion north to south Molde Northwestern Romsdalen Midøya Northwestern Romsdalen / Sunnmøre Brattvåg and Hildre Northwestern Sunnmøre Åram Northwestern Sunnmøre Øygarden Southwestern Nordhordland Bergen Southwestern Nordhordland Fana Southwestern Nordhordland Southwestern Nordhordland (Hardanger) Åkrafjorden Southwestern Sunnhordland Randaberg Southwestern Rogaland Stavanger Southwestern Rogaland Hå Southwestern Rogaland Lista Southwestern Vest-Agder

Molde

Molde is a city in the county Møre og Romsdal in Northwestern Norway. It is the administrative centre of the Molde with a population of 20.957 (2018)2 and it is the 23th biggest city in the country. In the Late Middle Ages there was an active timber trade with the and England. In the mid-17th century, Molde became the county capital, and in 1742 the city got trading rights. This led to the settlement of many upper-class families, creating a rich social life. In the 19th century the economic focus was on the industry and in addition tourism became an important source of income. During the period of industrialisation, many factories arose, creating a lot of and as a consequence, a lot of immigration. At the start of the Second World War in 1940, the whole city was bombed. When the war was over, the city was slowly rebuilt. The period after WWII is characterized by a large growth in population and the development of the city into the administrative and economic centre the city is today (Rød 2014: 6-8). Molde has a hospital, two ‘ungdomsskoler’ (junior high school), two ‘videregående skoler’ (upper secondary school) and a

2 Store Norske Leksikon. De største tettstedene I Norge. [22-05-2019] 14

‘høyskole’ (college). The city is well connected by bus, ferries and , and also has an airport with direct flights to other parts of Norway as well as abroad (Rød 2014: 15).

The dialect of Molde is part of the Romsdal dialect area. The periods with large growth in population also left their traces in the dialect, especially when large groups moved to the city speaking a dialect from another part of the country3. As a result, the dialect in Molde stands out from other dialects in the Romsdal area, characterized by a simplified , with for example simplified inflection in nouns and verbs. Also, the dialect cannot be described as one dialect, but is divided in a high-status variant and a low-status variant. The high-status is characterized by elements from the written standard bokmål (and formerly riksmål), while the low-status variant is closer to other Romsdal dialects (Rød 2014: 20-21).

Midøya

Midøya is a small island in municipality in the county Møre og Romsdal, about 30 km south of Molde. The island, only 13,9 km² big, has a population of around 410 people (2015)4. Over the last decades, the population of the island has been shrinking (Fossheim 2010: 8).

In table 1, not one but two dialect regions are mentioned for Midøya. This is the result of a former administrative border that, up until 1965, separated the island in two (Fossheim 2010: 1). The north and south side of the island were separated both administratively and ecclesiastically, with their own schools, clubs, national day celebrations (Fossheim 2010: 8). As a consequence, they each had different dialects. On the north side they spoke a romsdals dialect and on the south side a sunnmørs dialect. In 1965 there was a restructuring of , resulting in the merger of the two parts of the island into one municipality, Midsund. After 1965 the amount of contact between the two sides of the island increased, and today the island no longer has separate schools and clubs (Fossheim 2010: 11). Also, the ways of transport changed since then, with a bridge to the neighbouring island Otrøya (1969) and ferries and bus routes to Molde and Ålesund. Nowadays, ferries and busses are going more frequently to the east in the direction of Molde than to the southwest in the direction of Brattvåg and Sunnmøre (Fossheim 2010: 4). The majority of the citizens travel to the next island, Otrøya for work. Only a small percentage works on the island itself, either on farms, in construction or as teachers. There is one private primary school on the island, but for high school, pupils have to travel to Otrøya.

3 Store Norske Leksikon. Norske bymål. [22-05-2019] 4 Store Norske Leksikon. Midøya. [22-05-2019] 15

Brattvåg and Hildre

Brattvåg and Hildre are a town and a village in Haram municipality in the county Møre og Romsdal. Haram is the most northern municipality of Sunnmøre, right underneath Midsund municipality. Brattvåg and Hildre have a centre-periphery relation. Hildre is an old, small farming and fishing village. Brattvåg was founded in 1911. The earliest immigrants came from Hildre to build mills and hydroelectric power plants, and Brattvåg soon developed into an industry town (Hildremyr 2006: 6). With the municipal merger in 1965, Brattvåg became the municipal centre of Haram. As a result, in the following decades many public services and functions were established in Brattvåg, leading to a growth employment and population. Most immigrants came from other places in Sunnmøre. In 2014, Brattvåg had a population of 24125. The population in Hildre is a lot smaller with 535 citizens in 2015, and decreasing (Hildremyr 2006: 10). Both Hildre and Brattvåg have a primary school, but children from Hildre have to commute to Brattvåg for junior high school and secondary high school (Hildemyr 2006: 13).

The dialect in both Brattvåg and Hildre belongs to the Sunnmørs dialect area. Hildremyr presents only one dialect in her thesis, indicating that the language varieties in Hildre and Brattvåg are very much alike.

Åram

By ‘Åram’, a coastal area is meant that stretches from Hakallestranda, via Åram to Sørbrandal. Fishing and farming have historically been the most important source of income. The population has always been small, with a current size of circa 300 people (Lianes 2013: 11-14). Since 1. January 2002, Åram belongs to municipality, in southwestern Møre og Romsdal. Previously, Åram was part of Sande municipality, with Larsnes as the administrative centre. Larsnes is located on the other side of the and for the citizens of Åram only accessible by ferry. In 2002, a referendum was held which led to the reorganization of the municipal borders. Fiskå then became the new administrative centre, only a short drive away (Lianes 2013: 18). Åram offers only a few permanent jobs and most people have to travel for work. Children have to travel too because, in the last decades, both the high school and primary school were closed. The dialect on Åram is part of the sunnmørs dialect area.

5 Store Norske Leksikon. Brattvåg. [22-05-2019] 16

Øygarden

Øygarden is a municipality northwest of Bergen, consisting of a string of islands in the North with a population of 48896. Øygarden municipality is a result of the 1964 merger of Hjelme municipality and a part of municipality. For centuries, Øygarden was depending on the ferry-connection to travel to the mainland, but between 1956 and 1986 twelve bridges were built in a project to improve the connection between Bergen and its surrounding areas. In 1988 another big change happened for Øygarden when the Sture terminal was opened, which is an oil terminal, and in 1996 Kollsnes, a gas processing plant, was opened. This led to an increase in guest workers, but also created jobs for citizens of Øygarden (Villanger 2010: 3-5). Before these big changes, the population had been decreasing, but ever since the coming of the connection to the mainland and the industries, the population has been increasing7. Previously, most of the people worked on farms or in the fishing industry, but nowadays the secondary and tertiary sector are the biggest sectors. More than half of the working people travel outside the municipality for work, of which 25% travels to Bergen for work.

The dialect on Øygarden is in the nordhordland dialect area, and together with other dialects around Bergen called strilemål. The dialect is not identical on all islands in the municipality. The variation is a result of the former municipal borders, and because of the fact that before the building of the bridges, the islands used to be isolated from each other (Villanger 2010: 12).

Bergen

Bergen is the second biggest city of Norway with a population of 255 464 in 20188 and likes to call itself Vestlandets capital. Bergen has an important import- and export harbour, an airport, many ferry- and boat connections to both Norway and abroad, plus a direct train connection to Oslo by Bergensbanen9.

Bergen was founded by Olav Kyrre and got city status in 1070. Trade was important from the start, with an active trade in fish with the areas northwards and in goods such as wine, clothing and corn with cities abroad10. In 1350, a group of German merchants from the Hanseatic League settled in Bergen, and was established as a Hanseatic office. The Hanseatic League dominated life in Bergen especially in the 14th and 15th century. The merchants lived in Bryggen, and although there

6 Øygarden commune. Folketalsutvikling. https://www.oygarden.kommune.no/om- oygarden/folketalsutvikling/ [22-05-2019] 7 Store Norske Leksikon. Øygarden. [22-05-2019] 8 Store Norske Leksikon. De største tettstedene I Norge. [22-05-2019] 9 Store Norske Leksikon. Bergen – samferdsel. [23-05-2019] 10 Store Norske Leksikon. Bergen – historie. [23-05-2019] 17 was intensive contact between the merchants and Bergen citizens, they lived in two separated societies. As a result, Bergen was influenced by Low-German, but the languages did not mix11.

Bergen’s population kept on growing and Bergen was Norway’s biggest city up until the 1830s. Through the centuries, Bergen has always had a lot of immigration, both from abroad as from other parts of Norway, especially from Vestlandet. Nowadays, many of the people who move to the centre of Bergen move within the municipality or are from other parts of the (Nornes 2012: 10).

The dialect study in Processes of Dialect Change focusses on informants from the city centre, . Bergenhus has been the administrative centre of Bergen municipality since the merger of the districts Sentrum and Sandviken in 2000. Bergenhus is not only an administrative centre, within Bergenhus, many important institutions are located, such as Haukeland university hospital, many media headquarters, the , a campus of the Western Norway University of Applied Sciences and the Norwegian School of Economics. Many cultural significant organisations are found in Bergenhus as well, for example art KODE, concert hall Grieghallen and the National Theatre. Around 51000 people are employed within Bergenhus, of which circa 90% works in the tertiary sector. Many people travel to work every day, mainly within the borders of Bergen municipality (Nornes 2012: 15).

The dialect of Bergen stands out from other dialects from the same nordhordlands dialect area. Cities tend to develop differently than small dialects and this is no different for . The high amount of language contact to other languages or dialect varieties led to simplifications. The biggest differences between bergensk and the surrounding dialects is that bergensk is an e-variety, and that it has a simplified grammatical system with only two genders. The surrounding dialects have masculine, feminine and neuter, but bergensk does not differentiate between masculine and feminine and only has masculine and neuter (Nornes 2012: 17).

And, as is common in cities, there is a high-status variety and a low-status variety. But contrary to many other cities, the two varieties are linguistically not very different and not clearly separated. For example, both the high-status and low-status variety have two genders and both varieties have an ending on -et in past tense of weak a-verbs (kastet) (Mæhlum & Røyneland 2012: 100). The varieties serve an identity marker, and forms from both varieties are used by all informants (Nornes 2012: 19).

11 Store Norske Leksikon. Dialekter I Bergen. [23-05-2019] 18

Fana

Fana is one of the eight urban districts of Bergen with a population of 38.317 (2011)12. Fana was a separate municipality until it merged into Bergen municipality in 1972. In the 1870’s, a railway was built from Bergen to Voss, going straight through Fana. There was a station build in Hop and the railway started to function as public transport between Fana and Bergen (Doublet 2012: 19). Many privileged citizens from Bergen then decided to move to Fana, while working in Bergen. At first it was mainly the upper class who moved to Fana, but after a while also the middle class followed. This led to an enormous growth in population at the beginning of the 20th century, increasing from 4000 citizens in 1875 to 11600 in 1920 (Doublet 2012: 19). Moving to Fana had its advantages, since houses were cheaper than in the city centre and taxes were lower. Fana’s population kept increasing, especially right after the Second World War, when there was a housing shortage in Bergen which was resolved by building more houses in Fana.

The dialect study from Processes of Dialect Change focusses on informants from , which is the centre of Fana and the former municipal centre of Fana. In Nesttun, the same dialect is spoken as in Bergen, also with the variation of high-status forms and low-status forms, but with one difference. High-status forms are much more frequent and common in Fana than they are in Bergen (Doublet 2012: 11).

Ullensvang

Ullensvang is a municipality in Hordaland county and includes both sides of Sørfjorden, as well as a bigger area east of Sørfjorden, with a population of 3363 people. Most people live on the coastlines of the fjord, with 36% on the west side and 62% on the east side. There are two towns on the east coastline, the municipal center with a population of 551, and Lofthus with a population of 521 in 201613. The population in the municipality was stable up until the 1990s, but in the last decade the population is going down by annually 0,1%.

Unlike the closest towns of and , Ullensvang has no hydroelectric power industry. Ullensvang has the biggest fruit cultivation in the country and produces both apples, pears, plums, cherries and berries. The orchards draw many tourists to Ullensvang, which in the last decades has given in impulse to the tourist industry (Lid 2015: 18). Around 39%14 travels outside of the municipality for work, of which more than half works in Odda.

12 Store Norske Leksikon. Fana – bydel i Bergen. [23-05-2019] 13 Store Norske Leksikon. Ullensvang. [23-05-2019] 14 In 2013 (Lid 2015: 19). 19

In 1860, the first permanent schoolroom was opened, which substituted the omgangsskole, which was a school that traveled from place to place. More schools followed, and, up until the 1980’s, there were three school districts on the west side of the fjord and three on the east side (Lid 2015: 11). Up until this day, there is a separation between the two sides of the fjord. Students from different sides of the fjord do not often meet until they start attending high school in Odda. There are strong feelings of identity connected to the fjord side (Lid 2015: 13).

Åkrafjorden

By ‘Åkrafjorden’, the villages are meant that lie on the northern and southern shore of Åkrafjorden in Hordaland county. On the northside the informants are from Baugstranda and Åkra and on the southside from Kyrping, Rafdal, Markus and Teigland. Until 1965, both sides of the fjord belonged to the municipality Skånevik, with the town Skånevik on the south side as the administrative centre. People on both sides of the fjord belonged to one community. Although children went to different schools, there was one church in Åkra were people from all along the fjord gathered, and there were daily ferry connections. Nowadays, the ferries are no longer going, and there is almost no communication anymore between the fjord sides.

In 1965 the fjord sides got separated as a result of a municipal reorganization. The north side is part of municipality, with Rosendal and as the important centres. The south side belongs to municipality, with Etne as the administrative centre. Both sides of the fjord are outskirts of the municipality they belong to, and both sides struggle with depopulation. In 1910, the northside had a total population of 488 people. In 2016, the number had decreased to 199. On the southside there were 270 inhabitants in 1910, which had decreased to 169 in 2016 (Tjelmeland 2016: 11). On the north side, many farmsteads are already completely abandoned.

On the north side it takes about 45 to 60 minutes by car to get to the towns Husnes or Rosendal, what many people do on a daily basis to get to work. At the time of the dialect study, there was still a junior high school, so pupils did not travel on a daily basis to go to school. But for many extracurricular activities such as sports, children go to Husnes or Rosendal several times a week. Also, there is no senior high school. When children finish junior high, they have to move out to go to senior high, since there is no daily buss connection to senior high in Husnes (Tjelmeland 2016: 17).

Life in the villages on the south side is closely connected to Etne. Most people drive 20/30 minutes several times a day to go to work, school, kindergarten or to do grocery shopping (Tjelmeland 2016: 10). The three schools that once existed on the south side are all shut down.

20

Formerly, both sides of the fjord spoke the same dialect and was linguistically homogeneous. Nowadays, there is still a strong feeling of identity connected to the fjord, but only to one side of the fjord, which also has its consequences for the dialect (Tjelmeland 2016: 20).

Randaberg

Randaberg is a small municipality of 24 km² at the northern end of Jæren in Rogaland county. It is surrounded by water and in the south it has one border to Stavanger. Traces of inhabitation are found dating back to 6000 b.C. After the Middle-Ages, Randaberg developed as a farming municipality, with Stavanger as the most important trade market (Doublet 2015: 12). Still, 65% of the land in Randaberg is farming land.

After 1965, Randaberg changed a lot. In that year, an oil base was opened in Dusavika, creating new jobs and leading to a growth in population. This growth has been constant ever since, growing from a small town with 2000 inhabitants to a city with over 10000 inhabitants (Doublet 2015: 13). The largest part of the population, 71%, works outside of the municipality, with 43% traveling to Stavanger on a daily basis15. Many pupils also travel to Stavanger every day. There are three schools in Randaberg and since 1984 also a high school, but despite this, many children choose a high school in Stavanger over the one in Randaberg (Doublet 2012: 12).

Stavanger

Stavanger is the third biggest city in Norway with a population of 133.140 inhabitants16. It was founded in 1125 and stayed a small village throughout the Middle Ages, serving as a local trading centre for the farmers in Jæren. After the Middle Ages, there was a growth in population with people coming from both Norway and abroad. More than half of the newcomers was from Rogaland and Stavanger became a melting pot of all kind of Rogaland-dialects. In the 19th century, Stavanger flourished both economically, politically and culturally (Aasen 2011: 8) before it shifted to becoming a typical industry town in the beginning of the 20th century. The next big turning point came after the Second World War. In 1972, Parliament declared Stavanger as the city were the oil industry would be located. This was the start of another period of welfare for “oil city” Stavanger. Within ten years, Stavanger had a higher net profit per tax payer than the capital Oslo (Aasen 2011: 9). The growth in population that followed was, again, mostly people moving in from the area around Stavanger. Over the last decades, Stavanger has developed as a central city. More public buildings were built, as well as the , which opened in 2005 (Aasen 2011: 13).

15 Store Norske Leksikon. Randaberg. [28-05-2019] 16 Store Norske Leksikon. Stavanger. [28-05-2019] 21

The dialect of Stavanger has a high-status variety and a low-status variety. The high-status variety, that probably originates from the welfare period in the 19th century, has its roots in the written standard, conservative bokmål. The low-status variety developed from the local dialects that were spoken by immigrants that moved to Stavanger from other parts of Rogaland (Aasen 2011: 15). In Stavanger, the high-status and low-status variety are more separated than in Bergen, where the varieties historically served as a kind of social tools. In Stavanger, the varieties did indicate class and status.

Hå is a municipality at the southern end of Jæren. In the dialect study of Bøe (2013), the focus is on three towns in the southern corner of the municipality, , and Sirevåg. Hå has a population of 18762 inhabitants17, of which approximately 2400 people live in Ogna, Brusand and Sirevåg (Bøe 2013: 12). The municipality is one of the most important farming lands of the country, yet, the southern part where the dialect study was carried out is the exception to the rest of the municipality. Hå is a rural, very stable municipality. Over the last decades, there has been a steady, natural without much migration. Also, more than half of the people work within the municipality (Bøe 2013: 18).

Lista

Lista is a peninsula in municipality in Vest-Agder county. Up until 1965, Lista was a municipality of its own. Of the 9758 inhabitants of Farsund, 54 percent lives on Lista (Heradstveit 2018: 12). The population has been slowly decreasing since the 1980’s.

Lista is a flat area of land were signs of inhabitation have been found dating back many thousands of years. With its for Norway unusual long stretched white beaches, Lista is often compared to . In the time when transport happened mainly by boat, the peninsula was on the route of everyone passing the southern Norwegian coast, but now that the means of transportation have changed to the , Lista is in the periphery, with a 40 minutes’ drive to the closest main road (Heradstveit 2018: 13).

Historically, most people in Lista had small farms that could never cover all expanses. Incomes were always complemented by fishing and timber. When times were low in the middle of the 19th century, many people from Lista, both men and women, went to America to work and send the money home. Most of them returned back to Lista after a few years. Up until the 1960’s, there was a constant flow of people moving to and from America. This affected cultural life in Lista, as well as the language.

17 Store Norske Leksikon. Hå. [28-05-2019] 22

There is an American inspired restaurant, a shop, American street names and a yearly America Festival (Heradstveit 2018: 17). Because of the incomes from their work in America, the farms were not modernized as elsewhere in Norway. After the Second World War, the farms were bought by people moving to Lista from different parts of Jæren. This also left its traces in the dialect (Heradstveit 2018: 18).

Lista is still a very rural community, with most jobs in the primary and secondary sector. The senior high school in Lista only offers practical tracks and for general studies, pupils have to travel to the senior high school in Farsund. Compared to the national average, there are less people in Lista with a higher education.

3.2 Classification of informants In the studies of the project Processes of Dialect Change, informants are categorized in two ways. The informants are either classified on the basis of ‘livsfaser’, stages of life, or on the basis of ‘årsklasser’, age groups.

The age groups are static. The year of birth of the informant determines the age group and that does not change over time. Stages of life are not static and change as the informant gets older. Not the year of birth, but the age at the moment of the recording determines the stage of life.

Table 2 - Classification of informants in age groups.

Age groups Year of birth Age group 0 1860-1889 Age group I 1890-1919 Age group II 1920-1949 Age group III 1950-1979 Age group IV 1980-2009

In stages of life, the informants are categorized in three groups: the elderly, the middle-aged and the youth. In the group elderly informants the age ranks from 65 to 80, if necessary from 60 to 80. In the group middle-aged informants the age ranks from 35 to 50, and 30 to 59 if necessary. The group of young informants is in a lot of studies the 9th or 10th grade of a secondary school, called ‘ungdomsskole’ in Norwegian. Students in the 9th or 10th grade are 14 to 16 years old.

23

Table 3 - Stages of life

Every stage-of-life group fits into a certain age group. Stages of life Age Youth fits in age group IV, Middle-aged fits in age Elderly 65-80 years old (60-80) group III, Elderly fits in age group II. The two different Middle-aged 35-50 years old (30-59) ways are useful when recordings are made at two Youth 14-16 years old periods in time. An informant born in 1960, for example, will fall into age group III, both in the older study and in the recent study from Processes of Dialect Change. But in terms of stages of life, the informant born in 1960 will fall in the group of young informants in the older study, while in the more recent study the same informant will be categorized as an middle-age informant.

In the analysis (chapter 4), two groups of informants are repeatedly mentioned. The first group is the total group of informants that was part of the most recent dialect study. The generations in total are made up out of either the three stages of life, Elderly, Middle-aged and Younger, or the three age groups II, III, and IV. The second group that is mentioned in the analysis is the group of Younger informants, or age group IV. The language use of the youngest group of informants is compared to the language use of the generations in total.

3.3 Linguistic variables

The thirteen dialect studies from the Processes of Dialect Change-project that serve as the source for this paper each analyse a number of linguistic features. The study with the least number of variables analyses five items, the study with the largest number of variables analysed thirteen items, but on average eight linguistic items are studied per dialect study. All studies together generate a total of 108 linguistic items. Of this total amount, only the variables that have overlap in two or more dialect studies are analysed in this thesis. This resulted in a set of 17 linguistic variables, with a total amount of 76 tokens. These are the variables presented in 3.4.

Every token holds information about a linguistic variable in a specific dialect. Each token consists of two, sometimes more variants. Most tokens vary between a traditional and a new variant. By a traditional variant, the variant is meant that one would expect to find in the dialect on the basis of dialect descriptions in literature. A new variant is a variant that earlier was not part of the dialect where it is now observed and used. This can either be a derivative of the traditional variant, or a form that comes from another dialect. In the analysis, the traditional form is called variant 1 and the new form is called variant 2.

Another form of variation is found in urban dialects. These dialects have a high-status variety and a low-status variety. Besides variation between a traditional and a new variant, these urban dialects

24 can have variation between the high-status variant and the low-status variant. Marking the variants as variant 1 or variant 2 turned out to be more complicated for this variation than it was for the traditional and new variant.

Marking the high-status variant as variant 1 and the low-status variety as variant 2 would bring order, but a complication arises when the same linguistic variable occurs both in an urban dialect and in a rural dialect, which is illustrated in table 4.

Table 4 - Complication in structuring high-status and low-status variables.

Dialect Variation Variant 1 Variant 2 Stavanger High- vs. Low-status Infinite verb ending on -e Infinite verb ending on -a Øygarden Traditional vs. New Infinite verb ending on -a Infinite verb ending on -e Randaberg Traditional vs. New Infinite verb ending on -a Infinite verb ending on -e

The linguistic variable has three tokens, one token in an urban dialect and two in rural dialects. The variation that occurs is the same for all three tokens, with a variant on -e and a variant on -a, but with the high-status variant as variant 1, the tokens are not in the same column. The same issue arises – but in other variables - when structuring the variants the other way around, with the low- status variant as variant 1. To solve this problem, not all high- and low-status variants are structured in the same way. There is a cluster with the high-status variant as variant 1 and the low-status variant as variant 2, and there is a cluster with the low-status variant as variant 1 and the high-status variant as variant 2. The goal is to have the same variants of a variable in the same column for all tokens. The way the traditional and new variants were marked are leading. In case of types with only variation between a high- and low-status variant the decision is made on the basis of the data from the dialect studies. The form that was most dominant in the dialect study from a generation ago is then marked as variant 1, the ‘traditional’ or formerly most frequent form.

3.4 Presentation of linguistic variables Lexical variables

V01 Determiner /mye/

Number of tokens: 2 Dialects: Fana, Lista

For the quantifier mye, ‘a lot’, there is a token in the dialect of Fana and one in the dialect of Lista. Although the variable is the same, some differences can be found in the variation in the two dialects.

25

In Fana, the variable varies between a high-status variant and a low-status variant, while in Lista the variation is between a traditional variant and a new variant. Also, the variants do not fully overlap. The form [myә] is a variant for both dialects, but the competing variants differ.

Fana

High-status variant: [myə]

Low-status variant: [myçә, myʃә]

Lista

Traditional variant: møje, mø:e

New variant: my:e

V02 Determiner /noen/, /noe/

Number of tokens: 4 Dialects: Bergen, Fana, Lista, Lista

V02 consist of tokens from, again, Fana and Lista, with in addition a token from Bergen. Lista has two tokens in V02, since in the Lista-study the determiner was analysed separately for noen, ‘someone’ and noe, ‘something’. In the studies in Bergen and Fana, they were analysed as one variable, with noen as the masculine and feminine form, and noe as the neuter form of the paradigm. Again, in Fana the variation is between a high-status variant and a low-status variant, and in Lista between a traditional and a new variant. For Bergen, both forms of variation apply, but in the dialect study, the variable was analysed as variation between a high and a low form, so that will be the case in this study as well.

Bergen

High-status variant: någen/någet, noen/noe

Low-status variant: nåkken/nåkke

Fana

High-status variant: [nʊːən] / [nʊːə]

Low-status variant: [nɔkən] / [nɔkə]

Lista

26

Traditional variant: nɔ:en, nɔ:ne / nɔ:ge, nɔ:e

New variant: nu:en/nu:e

V03 Determiner /det/

Number of tokens: 3 Dialects: Åkrafjorden North, Åkrafjorden South, Ullensvang

V03 has tokens in Åkrafjorden and Ullensvang, which are dialects in two neighbouring regions, Sunnhordland and Hardanger. In both dialects, variation is found between the form /de:/ and /da:/, but which variant can be called the traditional form is not quite clear. In the Ullensvang-study, no information is given about which variant is the traditional variant, but from the data in the study it can be derived that a generation ago /da:/ was the dominating variant (Lid 2015: 80), and therefore variant /da:/ will be called the traditional variant and /de:/ the new variant for Ullensvang. In Åkrafjorden, the situation is the other way around. Variant /de:/ is mentioned as the traditional form here, although Tjelmeland remarks that there may be variation between the towns on both sides of the fjord (Tjelmeland 2016: 32).

Åkrafjorden (North and South)

Traditional variant: /de:/

New variant: /da:/

Ullensvang

Traditional variant: /da:/

New variant: /de:/

V04 Differentiation of /rn/

Number of tokens: 3 Dialects: Åkrafjorden North, Åkrafjorden South, Øygarden

With differentiation, two different phonemes become more distinctive. Old Norse /rl/, /rn/ and /fn/ differentiated to /dl/, /dn/ and /bn/. This is a common feature in southwestern dialects. In the dialects of Åkrafjorden and Øygarden it was analysed whether the original, differentiated phoneme is

27 still in use. Both studies were limited to the compound /dn/ > /rn/, since this is the most frequent combination. For V04, the variants align for both dialects:

Traditional variant: /dn/

Intermediate form: /n/

New variant: /rn/

V05 Segmentation of long /l/

Number of tokens: 3 Dialects: Randaberg, Åkrafjorden North, Åkrafjorden South

With segmentation, a long phoneme is split up in two different phonemes, creating a new phoneme compound. Old Norse long /n:/ became /dn/ and long /l:/ became /dl/. Segmentation is often found in dialects that also have differentiation. V05 has tokens in Randaberg and Åkrafjorden, that both limited the analysis to the segmentation of long /l/, in the case of Åkrafjorden because of the lack of relevant occurrences of long /n/ (Tjelmeland 2016: 34). Both studies analysed to what extent the traditional, segmented form is still in use.

Traditional variant: /dl/

New variant: /l/

V06 Loss of lenition

Number of tokens: 3 Dialects: Lista, Randaberg, Stavanger

The short, non-voiced consonants p, t, k get softened and become a voiced b, d, g after a vowel. This feature called lenition is found in the southern and southwestern coastal area of Norway. V06 has tokens in Lista, Randaberg and Stavanger. In all these dialects, the voiced consonants b, d, g are restored to p, t, k. Lenition is a lexical feature, and the process of change happens mostly in individual words. Therefore, it is said that this process of change can take quite some time (Sandøy 2018: 199).

Traditional variant: /b, d, g/

New variant: /p, t, k/

28

V07 Monophtongization of /ɛi/

Number of tokens: 3 Dialects: Brattvåg, Hildre, Åram

Brattvåg and Hildre belong to the northern sunnmørsk dialect area, Åram to the southern sunnmørsk dialect area. Originally, sunnmørsk is a dialect area with a lot of diphthongs, but, as in other parts of the country, many diphthongs are changing into monophtongs. V07 analyses the monophtongization of /ɛi/ to /e/. The Åram-study is specific in what exactly has been analysed, limiting itself to a small number of words that traditionally has a diphthong /ɛi/, originating from Old Norse long é: tre, ser, vêr, meter, brev and fele (‘tree, to see, weather, meter, letter and violin’) (Lianes 2013: 34). In the Brattvåg and Hildre-study no specifics are given about the variable, merely that in individual words the traditional pronunciation is with a diphthong, for example the word kne (‘knee’) (Hildremyr 2006: 26).

Brattvåg and Hildre

Traditional variant: /ɛi/

New variant: /e/

Åram

Traditional variant: /ɛi/

Intermediate form: /æ/

New variant: /e/

V08a + b Personal pronoun – 1. Person singular – stressed and unstressed

Number of tokens: 14 Dialects: Bergen, Fana, Midøya North, Midøya South, Molde, Stavanger, Åram

The personal pronoun first person singular is probably the word with the widest range of possible pronunciations in Norway, originating from the time when in a sentence, different forms were used in stressed and unstressed position. Over time, one or the other became generalised, resulting in the range of variants there are today (Sandøy 2018: 201). The different variants can be divided in two groups, either with or without a j- in onset position (eg / jeg). These are two different forms with a

29 different origin. The forms with a j- developed from Proto-Norse eka, whereas the forms without a j- developed from Proto-Norse ek (Skjekkeland 2005: 107).

V08 has 7 tokens for V08a – in stressed position, and 7 tokens for V08b – in unstressed position. Since all variants are the same in stressed and unstressed position, they are presented here simultaneously. In city-dialects of Stavanger, Bergen, Fana and Molde, V08 varies between a high- status variety and a low-status variant, with some intermediate forms. Even though all cities vary between a jeg-form and an eg-form, there is a difference between the southwestern cities and Molde. Belonging to the north-western dialect area, Molde, along with Midøya and Åram, have different eg-forms than the southwestern dialects.

Bergen and Stavanger

High-status variants: /jɛi/ /je:/

Low-status variants: /e:g/ /e:/ /ei/

Fana

High-status variant: /jɛi/

Intermediate form: /ɛi/

Low-status variant: /eg/

Intermediate form: /e/

Molde

High-status variant: /jei/

Low-status variant: /i:/

Nordmøre og Ålesund: /eː/

Midøya

Romsdalsk variant: /i/

Sunnmørsk variant: /ɛi:/

Intermediate form: /e:/

Åram

Traditional variant: /i/

30

New variant: /ɛi:/

Intermediate form: /e:/

V09 Negation “ikke”

Number of tokens: 5 Dialects: Bergen, Fana, Hå, Molde, Stavanger

The personal pronoun first person singular and the negation adverb ikke are strong solidarity and identity markers (Sandøy 1985: 139). These two words are very high-frequent and salient, and a change in the way you pronounce these words is very marked, since they express the belonging to a certain group. The (South-)Eastern Norwegian variant is /ikə/, the Western Norwegian variant is /iҫə/. In West-Norwegian towns that have a high- and a low-status variant, ikke is used as the high- status variant and ikkje (/iҫə/) is the low-status variant. V09 has five tokens, four of which are found in urban dialects, varying between ikke and ikkje. The fifth token in Hå does not have an opposition between ikke and ikkje, but between different ways of pronouncing ikkje.

Bergen and Stavanger

High-status variant: /ikə/

Low-status variant: /iҫə/

Fana

High-status variant: /ikə/

Low variety form: /iҫə/, /iʃə/

Molde

High-status variant: /ikə/

Low-status variant: /içə/

New youth variant/urban form: /iʃə/

Traditional variant: /icҫ͡ ə/

Intermediate form: /iҫə/

31

New variant: /iʃə/

Phonological variables

V10a The sj-sound

Number of tokens: 3 Dialects: Bergen, Hå, Stavanger

The development that is analysed in V10a is the change from /sҫ/ or /sj/ to /ʃ/. The /ʃ/-sound, originally a south-eastern Norwegian sound, started spreading among younger people in Western and Southern Norway, mostly in urban areas, while the older generation still distinguished /sj/. The three dialects with a token for V10a all have multiple variants, the traditional variant, the new variant and one or more intermediate forms. In the data it became soon very clear that the traditional variant /sj/ or /sҫ/ is hardly ever used any more, and that the variation is found between the intermediate form /ʃj/ and the new variant /ʃ/. The variant in the analysis that is called ‘traditional variant’ for V10a is actually the intermediate variant.

Bergen

Traditional variant: /sj/

Intermediate variant: /ʃj/

New variant: /ʃ/

Other forms: /ҫ/ and ҫ+

Stavanger

Traditional variant: /ʃj/

Intermediate variant: /ʃj/

New variant: /ʃ/

Traditional variant: /sҫ/

Intermediate variant: /ʃj/

New variant: /ʃ/

32

V10b The kj-sound / Merging of the kj-sound and sj-sound

Number of tokens: 5 Dialects: Bergen, Hå, Lista, Molde, Randaberg, Stavanger

In V10b two synonym processes are put together in one variable. The variable analysis the change of the pronunciation of the kj-sound, that is changing from the original pronunciation with /ҫ/ into the phoneme /ʃ/. What follows from this development is a merge of the kj- and sj-sound. This merge is only possible after a change in the pronunciation of sj-, as described for V10a.

The merge of the kj- and sj-sound was first observed in the bigger cities among the younger generations in most towns and cities as far north as , skipping the rural dialects in between (Skjekkeland 2005: 79). The occurrence of tokens in Hå and Lista in this study might be a sign that, after the city hopping, the phenomenon is now spreading around these cities like a wave.

V10b only analyses the pronunciation of kj- in onset position. In a number of words, kj- occurs in medial position, for example in the negation ikke, but this was in most studies made into an own variable (see V09).

Bergen

Traditional variant: /ҫ/

New variant: /ʃ/

Intermediate/other variants: /ʃj/, /cҫ͡ /, ҫ+

Traditional variant: /cҫ͡ /

Intermediate form: /ҫ/

New variant: /ʃ/

Lista

Traditional variant: /ҫ/

New variant: /ʃ/

Molde

33

Traditional variant: /ҫ/

New variant: /ʃ/

Stavanger

Traditional variants: /ҫ/, /cҫ͡ /

New variants: /ʃ/, /ʃj/

Randaberg

Traditional variant: /ҫ/

New variant: /ʃ/

V11 Retroflection of /rt/ (/r/ + alveolar)

Number of tokens: 5 Dialects: Brattvåg, Hildre, Midøya North, Midøya South, Åram

Traditionally, the dialects of Brattvåg, Hildre, Midøya and Åram separate the consonants in compounds of r + t, d, n, s, l and pronounce them as two consonants, such as in /gʉ:lt/ and /sʉ:rt/ (gult ‘yellow’, and surt ‘sour’). In other dialects in Eastern, Central and , the consonant compounds assimilate into a retroflex sound so that /gʉ:lt/ becomes /gʉ:ʈ/ and /sʉ:rt/ becomes /sʉ:ʈ/. The retroflection of r + an alveolar consonant is common within the borders of the area with thick l, but recently it appears to be spreading outside of the thick l-area, and also independently of thick l, for example in Sunnmøre (Skjekkeland 2005: 68). V11 has five tokens in the dialects of Brattvåg, Hildre, Midøya North and South, and Åram. The study in Åram was limited to only the compound /rt/, in Midøya to /r/ + alveolar t, d, n. For the study in Brattvåg and Hildre it is not specified further than /r/ + alveolar consonant.

Brattvåg and Hildre

Traditional variant: /r/ + alveolar

New variant: /r/ + dental

Midøya North and South

Sunnmørsk variant: /r/ + t, d, n

Romsdalsk variant: retroflex/postalveolar /r/ + t, d, n

34

Åram

Traditional variant: /rt/

New variants: /t/, /ʈ/

V12 Interrogatives with /kv-/

Number of tokens: 2 Dialects: Fana, Molde

Interrogatives such as hva, hvor, hvorfor hvem and hvordan (‘what’, ‘where’, ‘why’, ‘who’ and ‘how’) are in Western Norway traditionally pronounced with /kv-/ or /k-/. In the dialects of several Western Norwegian towns, such as Molde and Fana, there is a pronounced difference between the low-status variant and the high-status variant of interrogatives. In these urban dialects, /k/ is the low-status variant and /ʋ-/ the high-status variant.

High-status variant: /ʋ-/

Low-status variant: /k-/

V13 Loss of palatalization of alveolars

Number of tokens: 4 Dialects: Brattvåg, Hildre, Midøya North, Midøya South

Palatalization of alveolars is often mentioned as a trademark for Trønder dialects, but covers a far wider area, from Sogn og Fjordane up until in the north. Alveolar consonants nn, ll, dd, tt become palatalized, such as in mann > /maɲ/ and ball > /baʎ/. In the central area of this phenomenon, from Sogn og Fjordane until Trondheim, palatalization of alveolars occurs both in stressed and unstressed position, around the centre palatalization is found only in stressed position (Skjekkeland 2005: 74). Palatalization of alveolars is a widespread phenomenon in the northern part of Norway but appears to be dissolving.

In the dialects of Brattvåg, Hildre and Midøya, palatalization can occur both in stressed and unstressed position, but for Midøya, the analysis was limited to /ɲ/ in stressed position (Fossheim 2010: 21).

Brattvåg and Hildre

35

Traditional variant: with palatalization (palatal /ʎ/ and /ɲ/)

New variant: without palatalization (alveolar /l/ and /n/)

Midøya North and South

Traditional variant: /ɲ/ (palatal)

New variant: /n/ (alveolar), /rn/ (retroflex)

Morphophonological variable

V14 Loss of palatalization of velars

Number of tokens: 7 Dialects: Brattvåg, Hildre, Ullensvang, Øygarden, Åkrafjorden North, Åkrafjorden South, Åram

Palatalization of velars is an old feature where the velar consonants /g/ and /k/ become palatalized to /ɟj/ or /j/; /cҫ/ or /ҫ/, before the vowel /i, e/. There is a lot of variation in the realisation of palatalized velars in medial position (Tjelmeland 2016: 34, Lianes 2013: 36), and the morphological categories in which palatalization occurs also differ per dialect. The dialect in Ullensvang merely has two categories with palatalization, strong masculine and feminine substantives, while dialects such as in Brattvåg, Hildre and Åram have a wide range of categories, besides different substantive classes also adjectives and the past participle of strong verbs. Already in the 19th and 20th century this feature started to be abolished, category by category (Sandøy 2018: 218). In the dialect studies that analysed the loss of palatalization of velars, the results were not specified per category, so that the results only inform whether or not palatalization still occurs, and not in which categories this happens.

Traditional variant: with palatalization of velars

New variant: without palatalization of velars

Morphological variables

V15 Verb – Infinite verb ending

Number of tokens: 3 Dialects: Randaberg, Stavanger, Øygarden

36

The ending of the infinite verb is a significant feature in defining a dialect area as an a-language or e- language. Western Norway, up until Sogn og Fjordane, has an infinitive on -a as a common feature. In Stavanger and Bergen, cities with a high- and low-status variety, the a-infinitive is the low-status variety and the e-infinitive the high-status variety, the latter corresponding to the south-eastern Norwegian form. V15 has three tokens: Stavanger itself, the neighbouring town Randaberg, and Øygarden, which is not far from Bergen. For all three tokens it is analysed to what extent the use of the e-infinitive is increasing at the expense of the a-infinitive.

Traditional variant: a-infinitive

New variant: e-infinitive

V16 Verb – Present tense – weak verbs

Number of tokens: 4 Dialects: Hå, Molde, Ullensvang, Øygarden

In V16 four tokens our found with some differences in the various variants. Hå traditionally has an a- ending in the category of kasta-verbs, that now seems to be getting influenced by the e-ending that is used in Stavanger (kaste). In Ullensvang, the traditional ending is on -ar (kastar) but seems to be changing to an ending on -a, which is the ending in several neighbouring dialects (Lid 2015: 88). Øygarden originally has an ending on -a, but seems to be moving closer to Bergen that has an ending on -ar and -e. Further north, the dialect in Molde never rigidly had only one ending, varying between -a and -e, both low-status variants, and in addition a high-status variant on -er and a ‘new’ variant without a vowel-ending, of which the origin is unclear (Rød 2014: 73)

Traditional variant: a-ending - kasta

New variant: e-ending - kaste

Ullensvang

Traditional variant: -ar - kastar

New variant: a-ending (common in neighbouring dialects) - kasta

Other variant: e-ending (from døma-category) - kaste

Øygarden

37

Traditional variant: a-ending - kasta

New variant: -ar, -e – kastar/kaste

Molde

Low-status variant: e-ending - kaste

Low-status/Romsdalsk variant: a-ending - kasta

High-status variant: er-ending - kaster

New variant: 0-ending (uncertain origin) - kast

V17 Verb – Present tense – strong verbs

Number of tokens: 2 Dialects: Lista, Molde

V17 has tokens in two dialects that originally have a different system of inflection of strong verbs. In Western and Southern Norway, from up till Ålesund, present tense has i-umlaut and an e- ending: kjeme and søve (I come, I sleep). From Ålesund northwards, the present tense is a single syllable with i-umlaut (kjem, søv). In South-Eastern Norway, the present tense does not have i- umlaut and has an ending on -er: kommer, sover. In several cities and towns in the coastal area from Mandal till Ålesund, the South-eastern variant without i-umlaut seems to be spreading (Skjekkeland 2005: 120-121).

V17 analyses the use of present tense without i-umlaut in two dialects that originally do have umlaut. In Lista, the traditional variant is with i-umlaut and an ending on -e. The new variant has no umlaut, but the e-ending is preserved: kjeme, søve > kome, sove. The low-status variant in Molde is a one- syllable form with i-umlaut. The high-status variant corresponds to the South-eastern variant on -er without i-umlaut: kjem, søv > kommer, sover.

Lista

Traditional variant: i-umlaut + e-ending: kjeme

New variant: no umlaut + e-ending: kome

Molde

Low-status variant: Ø (i-umlaut): kjem

38

High-status variant: no umlaut + -er ending: komer

Other variant: -e ending (sunnmørsk): kjeme

Table 5 - List of linguistic variables.

Type N Linguistic variable Dialect V01 2 Determiner /mye/ Fana, Lista V02 4 Determiner /noen/, /noe/ Bergen, Fana, Lista (2x) V03 3 Determiner /det/ Ullensvang, Åkrafjorden North, Åkrafjorden South V04 3 Differentiation of /rn/ Åkrafjorden North, Åkrafjorden South, Øygarden V05 3 Segmentation of long /l/ Randaberg, Åkrafjorden North, Åkrafjorden South V06 3 Loss of lenition Randaberg, Lista, Stavanger V07 3 Monophtongization of /ɛi/ Brattvåg, Hildre, Åram V08a 7 Personal pronoun – 1. Person sg. - Bergen, Fana, Midøya South, stressed Midøya North, Molde, Stavanger, Åram V08b 7 Personal pronoun – 1. Person sg. - Bergen, Fana, Midøya South, unstressed Midøya North, Molde, Stavanger, Åram V09 5 Negation “ikke” Bergen, Hå, Fana, Molde, Stavanger V10a 3 The sj-sound Bergen, Hå, Stavanger V10b 6 The kj-sound & Merging of kj- and sj- Bergen, Hå, Lista, Molde, Randaberg, Stavanger, V11 5 Retroflection of /rt/ (/r/ + alveolar) Brattvåg, Hildre, Midøya North, Midøya South, Åram V12 2 Interrogatives with /kv-/ Fana, Molde V13 4 Loss of palatalization of alveolars Brattvåg, Hildre, Midøya North, Midøya South

39

V14 7 Loss of palatalization of velars Brattvåg, Hildre, Ullensvang, Øygarden, Åkrafjorden North, Midøya South, Åram V15 3 Verb – Infinite verb ending Randaberg, Stavanger, Øygarden V16 4 Verb – Present tense – weak verbs Hå, Molde, Ullensvang, Øygarden V17 2 Verb – Present tense – strong verbs Lista, Molde Total 76

4. Analysis

The overlapping variables are analysed in order to compare the processes of dialect change in the different dialects in Western Norway. The analysis is based on the frequency of use of variant 1 and variant 2 of the two group of informants. The language use of the youngest generation is set against the language use of the generations in total. As a result, three clusters emerge that will be called category A, B and C. Every category reflects the stage in the process of language change of the linguistic variable.

Table 6 - The process of linguistic change in 3 stages

A B C

The first stage of change is category A. In this category, variant 1 is the dominant variant, both among the generations in total and among the youngest generation. Although the youngest generation is starting to use variant 2, variant 1 is still the dominant form. The second stage of change is category B. In this category, the younger generation deviates from the generations in total. The generations in total still have variant 1 as the dominant variant, while the youngest generation has variant 2 as the dominant variant. The younger generation has adopted the new form, but the language change is limited to this generation, the older generations have not (yet) accepted the new form. The third and last stage is category C. Variant 2 is the dominant variant in category C, both among the youngest generation and among the total group of informants.

40

Table 7 - Categories

Three categories Dominant variant – Dominant variant – generations in total youngest generation A 1 1 B 1 2 C 2 2

As a side note, not all variables have two variants. It is possible that three or even more variants exist for one linguistic variable. But as it turns out, the variant 3 or 4 or 5 is an intermediary variant that only has a small percentage of use and would not change the results of the analysis. Therefore, only variant 1 and variant 2 are mentioned in the analysis, intermediary forms are left out.

In the following chapters, the variables of all three categories are discussed. A distinction is made in the analysis between variables with variation between a traditional and a new variant and variables with a high- and a low-status variant.

4.1 Category A Category A consists of 17 tokens from 12 types. For all tokens, variant 1 is the dominant variant, both among the generations in total and in the youngest generation of informants. For the tokens that vary between a traditional and a new variant, this means that the process of linguistic change is at a very early stage. Even though there is a new variant that is being used, the traditional variant holds a strong position, also among the youngest generation.

In terms of dialects that appear in this category, it is conspicuous that no large cities fall into this category. The urban centres like Bergen or Stavanger have no variables in category A. This does not mean that it is a ‘rural’ category, since there are urban dialects represented in this category, for example the towns Molde and Fana, but not the dialects of the biggest urban centres.

Another remarkable point regarding the different dialects in category A is that for the dialect in Lista, all the linguistic variables fall into this category. Lista is far from an urban centre, so that its variables fall into category A is not surprising, but to have all variables in the same category does not happen for any other dialect.

From the 17 tokens, 13 tokens vary between a traditional variant and a new variant, and 4 tokens have a high- and a low-status variety variant. First, the tokens will be discussed that have a traditional and a new variant, thereafter the high- versus low-status variants will be discussed.

41

4.1.1 Traditional vs a new variant The tokens with variation between a traditional and a new variant in category A are in an early stage in the process of linguistic change. The youngest generation is picking up on the new variant, but the traditional variant is still the dominant one, also among the youngest generation.

In the set of 13 tokens, not a lot of overlap is found between the linguistic types (see table 9). Overlap is found in types V01 and V02, which are both determiners. But rather than overlap in types, it is a cluster of tokens from the same dialect, all tokens being from Lista, presented in table 8. As mentioned above, it is the only dialect where all tokens fall into the same category. Moreover, five out of the six tokens are similar in the percentages of use of the old and new variant. Type V01, V02, V06 and V17 all have a percentage of use of variant 1 among the generations in total of around 80%. The youngest generation uses the traditional a bit less, with a frequency of circa 75% of variant 1. Only for type V10b the situation is different, with 96% of variant 1 for the generations in total. In line with the other tokens, in V10b the percentage of use among the youngest variation of variant 1 is about 5% lower than the generations in total, but overall variant 1 has a much more stable position in V10b than in the other tokens. V10b will be compared to similar tokens from the same type in 4.4.

Table 8 - Category A: tokens found in Lista.

Types Variant 1 Variant 2 % Var. 1 – % Var. 1 – % Var. 2 – generations youngest youngest in total generation generation V01 - Determiner /møje/, /mø:e/ /my:e/ 83 76 24 /mye/ V02 - Determiner /nᴐ:ge/, /nᴐ:e/ /nu:e/ 80 72 28 /noe/ V02 - Determiner /nᴐ:en/, /nᴐ:ne/ /nu:en/ 79 68 32 /noen/ V06 - Loss of lenition /b/, /d/, /g/ /p/, /t/, /k/ 83 76 20

V10b – The kj-sound /ç/ (for /kj/) /ʃ/ (for /kj/ and /sj/) 96 91 9 & Merging of kj- and sj- V17 - Verb- Present /kjeme/ (i- /kome/ (without i- 79 75 25 tense - strong verbs umlaut umlaut) +epenthesis)

42

In addition to the determiners found in the dialect in Lista, one other determiner is found in category A, type V03 - determiner /det/, ‘this’. The token in Åkrafjorden South seems to deviate from the rest of the category in that the frequency of use of variant 1 among the youngest informants is higher than the percentage among the generations in total, which shows that the traditional form /de:/ holds a stable position on the southside of the fjord. Type V03 also has a token in Åkrafjorden North, belonging to category C in this analysis and showing the opposite of the situation in the south, with a strong increase of /da:/. A possible explanation for the absence of change on the southside is mentioned by Tjelmeland. The closest town on the southside of the fjord is Etne, and in this town the traditional form /de:/ is the most frequent variant (Tjelmeland 2016: 32).

Another cluster in table 9 are morphological types about verb inflection. Since the tokens are from different types it is tricky to compare them, but as will be discussed further down in 4.4, most morphological types concerning verbs are in category A, with the exception of V16- the inflection of weak verbs in present tense (see also V16 in 4.4).

Category A shows the tokens with the first signs of a process of linguistic change. Within this category, it is interesting to see which variables are in a more advanced stage, showing a relatively high percentage of use of the new variant among the youngest generation and thus probably on their way to the next category. Interestingly enough, the token with the highest percentage of use of variant 2 is similar to the token with the lowest percentage of use of variant 2, which was V10b in Lista, at the southern end of the Western Norwegian dialect area. The token with the highest frequency, 42%, is found at the other end of the dialect area, in Molde. In Molde only cases of /kj/ in initial position are counted, words with /kj/ in medial position, such as ‘ikke’, are left out, or in the case of ‘ikke’ made into an own variable (Rød 2014: 72). Whether the same was done for V10b in Lista is not specified (Heradstveit 2018: 41, 55), but fact is that the number of hits in Lista is far higher than in Molde, with 2456 hits in Lista and only 753 in Molde. Since it is not clear what exactly is researched in Lista, it is difficult to say if this has had an effect on the results in this category.

Table 9 - Category A: traditional/new variant.

Type Dialect Variant 1 Variant 2 % Var. 1 – % Var. 1 – % Var. 2 – generations youngest youngest in total generation generation V01 - Determiner /mye/ Lista /møje/, /mø:e/ /my:e/ 83 76 24 V02 - Determiner /noe/ Lista /nᴐ:ge/, /nᴐ:e/ /nu:e/ 80 72 28 V02 - Determiner /noen/ Lista /nᴐ:en/, /nᴐ:ne/ /nu:en/ 79 68 32

43

V03 - Determiner /det/ Åkrafjorden /de:/ /da:/ 80 89 11 South V06 - Loss of lenition Lista /b/, /d/, /g/ /p/, /t/, /k/ 83 76 20 V09 - Negation - /Ikkje/ Hå /iҫe/ /iʃe/ 88 76 24 V10b – The kj-sound & Molde /ҫ/ /ʃ/ 92 58 42 Merging of kj- and sj- V10b – The kj-sound & Lista /ç/ (for /kj/) /ʃ/ (for /kj/ and 96 91 9 Merging of kj- and sj- /sj/) V11 - Retroflection of /rt/ Åram rt (without with retroflex /t/ 84 64 36 retroflex /t/) V13 – Loss of Midøya /ɲ/ /nn/, /ɳ/ 77 47 45 palatalization of alveolars North V15 - Verb- Infinite verb Øygarden a-infinitive e-infinitive 93 61 39 ending V15 - Verb- Infinite verb Randaberg a-infinitive e-infinitive 76 65 35 ending V17 - Verb- Present tense Lista /kjeme/ (i- /kome/ (without 79 75 25 - strong verbs umlaut i-umlaut) +epenthesis)

4.1.2 High-status variety vs low-status variety Four tokens vary between a high- and low-status variant, two of which are in Fana and two in Molde. In the Molde dialect, variant 1 is the low-status variant and variant 2 the high-status variant. In the Fana dialect the opposite is the case, with variant 1 as the high-status variant and variant 2 as the low-status variant.

Starting with Molde, two different types are shown in table 10. In both cases the high-status variant corresponds to the East-Norwegian form. For type V12 the dominant low-status variant appears to be in a quite stable situation with 14% use of the high-status variant among the youngest generation. Type V17 shows stronger variation between the low- and high-status variant, with 41% of the high- status variant among the youngest generation. After taking a closer look at the dialect study Hild Rød carried out (2014), the relatively high frequency of the high-status variant can be attributed to the frequency in the data of the verb å komme, ‘to come’. With the verb komme left out, the frequency of the low-status variant with umlaut was as much as 93,4% for the youngest

44

generation (Rød 2014: 93-94), showing an entirely different situation with a more stable position of the low-status variant.

Table 10 - Category A: Low/high-status variety in Molde. Type Dialect Variant 1 – Variant 2 – % Var. 1 – % Var. 1 – % Var. 2 – Low-status High-status generations youngest youngest variety variety in total generation generation V12 - Interrogatives with /kv-/ Molde /k/- /ʋ/- 89 86 14 V17 - Verb- Present tense – Molde With umlaut /er/-ending 75 59 41 strong verbs

The two tokens in Fana are the personal pronoun first person singular in stressed position (V08a) and unstressed position (V08b). Variant 1 and variant 2 both exist of two forms, of which the one in bold is the most frequent. As was the case for the high-variant in Molde, the high-status variants correspond to the East-Norwegian form.

Table 11 - Category A: High/low-status variety in Fana.

Type Dialect Variant 1 – Variant 2 – % Var. 1 – % Var. 1 – % Var. 2 – High-status Low-status generations youngest youngest variety variety in total generation generation V08a - Personal pronoun 1. Fana /jɛi/ /æi/ /e:g/ /e:/ 44 45 27 person sg. subject -stressed V08b - Personal pronoun 1. Fana /jæi/ /æi/ /eg/ /e/ 40 45 36 person sg. subject - unstressed

Whereas in Molde the low-status variant was the dominant form, in Fana it is the high-status variant that has the highest frequency of use. Moreover, not only is variant 1 the dominant variant among the youngest generation, it also has a slightly higher percentage of use of variant 1 than the generations in total. Though it seems to indicate that the youngest generation deviates from the older generations with an increasing frequency of use of the high-status variant, it might actually be a generation above that is the deviating generation. In the oldest generation it is also the high-status variant that is dominating, but in the middle-age generation the dominant variant is the low-status variant. Nevertheless, the youngest generation in Fana is a bit different in their use of high-status variants, as will become clear when analysing other tokens from the same dialect, and other tokens of the same type, which will be done in 4.3.2 and 4.4.

45

4.2 Category B Category B contains tokens where the generations in total have Variant 1 as the dominant form while the youngest generation has Variant 2 as the dominant form. In other words, it is the cluster with the biggest difference in language use between the youngest generations and the older dialect users. Only one token in this cluster has variation between a high- and a low-variant, the other 19 tokens vary between a traditional and a new variant. A number of linguistic variables have multiple tokens in category B. On the one hand there is the pronunciation of the sounds /kj/ and /sj/, typically described as a youth phenomenon, so to find such a youth phenomenon in the cluster with the biggest difference between generations is not surprising. On the other hand, a number of linguistic types occur in category B that attest to dialect levelling, such as differentiation, segmentation, loss of palatalization of velars and alveolar and loss of lenition.

The tokens in this category are in the middle of the process of language change. The new variant has spread among the youngest generation. Essentially it will become clear whether the variables spread to older language users as well, in which case the same variables will belong to Category C if the dialect studied are repeated again a generation from now, or the variables turn out to be youth phenomena that do not push out the original variants over time.

4.2.1 Traditional vs new variant As is mentioned in the paragraph above, all tokens except for one have variation between a traditional and a new variant. In this cluster the tokens are found where the youngest generation is deviating from the other dialect users. When the percentage of use of variant 1, the traditional variant, among the generations in total is sorted from high to low, it ranks from 83% till 50%. Conspicuously, it is amid the tokens with the highest frequency of use of variant 1 among the generations in total that at the same time show high percentages of use of variant 2 among the youngest generation. In other words, these are the tokens where the youngest generation is deviating strongest and rather than gradually changing to another variant, the youngest generations breaks with the older dialect users and has switched to a new variant. In table 12 the tokens are shown where the frequency of variant 2 among the youngest generation almost equals or even transcends the frequency of variant 1 among the generations in total. In V10b - Bergen and V14 the strongest deviation is visible with a frequency of use of 94% for the new variable in both cases.

46

Table 12 - Category B: traditional/new variant with a high percentage of use of variant 2 among the youngest generation.

Type Dialect Variant 1 Variant 2 Variant % Var.1 – % Var.1 – % Var.2 – 3 generations youngest youngest in total generation generation V06 – loss of lenition Randaberg /b/, /d/, /g/ /p/,/t/, /k/ 83 25 75 V10b – The kj-sound Bergen /ҫ:/ /cҫ͡ / /ʃ:/ /ʃj/ ҫ+ 62 6 94 & Merging of kj- and sj- V10b – The kj-sound Randaberg /ҫ/ /ʃ/ 69 26 73 & Merging of kj- and sj- V14 – loss of Øygarden With Without 79 6 94 palatalization of palatalization palatalization velars V16 – verb present Øygarden /kasta/ /kastɛ/ /kastaR/ 75 15 70 tense

Type V10a and V10b have multiple tokens in category B in addition to the two mentioned in the table above, as shown in table 13. The change in pronunciation of the /kj/- and /sj/-sound is a frequent discussed topic in the literature and is described as a youth phenomenon, therefore it is no surprise that these variables turn up in this category.

Table 13 - Category B: traditional/new variant - V10a and V10b.

Type Dialect Variant 1 Variant 2 Variant 3 % Var.1 – % Var.1 – % Var.2 – generations youngest youngest in total generation generation V10a - The sj-sound Bergen /ʃj/ /ʃ/ /sj/ 56 18 82 V10a - The sj-sound Stavanger /ʃj/ /ʃ/ /sj/ 58 25 75 V10b – The kj-sound & Bergen /ҫ:/ /cҫ/ /ʃ:/ /ʃj/ ҫ+ 62 6 94 Merging of kj- and sj- V10b – The kj-sound & Hå /ҫ/ /ʃ/ /cҫ͡ / 74 44 56 Merging of kj- and sj- V10b – The kj-sound & Randaberg /ҫ/ /ʃ/ 69 26 73 Merging of kj- and sj-

47

V10b – The kj-sound & Stavanger /ҫ/ /ʃ/ /cҫ͡ / 62 32 68 Merging of kj- and sj-

For type V10b, the variable is changing from the old variant /ҫ/ to the new variant /ʃ/, most prominently in Bergen. In type V10a the traditional variant /sj/ is not really an optional form for the youngest generation and the variation within this type is rather between a new (/ʃj/) and a newer form (/ʃ/). This is the case both in Bergen and Stavanger. The change from /sj/ to /ʃj/ that happened a generation ago was a bigger step than the current change, since the difference between /ʃj/ and/ʃ/ is far less than the difference between /sj/ and /ʃj/.

Apart from the youth phenomenon, another cluster appears in category B, consisting of a number of types all related to dialect levelling. These are old Western-Norwegian dialect features that are going out of use among the youngest generation, such as lenition, segmentation, differentiation and palatalization. It is among the rural dialects that these tokens occur.

Table 14 - Category B: traditional/new variant - signs of dialect levelling.

Type Dialect Variant 1 Variant 2 Variant 3 % Var.1 – % Var.1 – % Var.2 – generations youngest youngest in total generation generation V04 -Differentiation Åkrafjorden /dn/ /rn/ /n/ 58 26 74 of /rn/ North V04 - Differentiation Åkrafjorden /dn/ /rn/ /n/ 51 30 67 of /rn/ South V04 - Differentiation Øygarden /dn/ /rn/ /n/ 50 0 100 of /rn/ V05 - Segmentation Åkrafjorden /dl/ /l/ 56 16 84 of long /l/ North V05 - Segmentation Åkrafjorden /dl/ /l/ 51 9 91 of long /l/ South V06 - Loss of lenition Randaberg /b/, /d/, /g/ /p/, /t/, /k/ 83 25 75 V06 - Loss of lenition Stavanger /b/, /d/, /g/ /p/, /t/, /k/ 70 50 50 V13 - Loss of Midøya South /ɲ/ /nn/, /ɳ/ rn 58 28 66 palatalization of alveolars

48

V14 - Loss of Øygarden /sa:jɛ/, /sa:gɛ/, 79 6 94 palatalization of /bacҫen/, /bakken/, velars /ta:cҫe/ /ta:ke/

With a few exceptions, in most tokens the youngest generation is varying quite a lot between the traditional variant and the new variant. The exceptions are the earlier discussed V14 in Øygarden, F1 also in Øygarden and V05 in Åkrafjorden South. In the other tokens the traditional dialect form is not completely lost yet, but the tendency is rather clear and will only be a matter of time before the old variant will disappear. It is in these tokens that clear signs of dialect levelling become apparent, where the most local traditional form is being pressed out by a more regional or national form. This issue will be further discussed in 4.4.

4.2.2 High-status variety vs low-status variety Only one token in category B displays variation between a high-status variant and a low-status variant. Variable D, the negation adverb “ikke” in Molde has variation between a high-status variant, a low-status variant and a new urban variety. In the cluster of tokens of variable V09, the one in Molde is an exception. In line with the rest of the type, the low-status variant is the dominant one for the generations in total, but contrary to the other four tokens, the youngest generation in Molde has the high-status variant /ike/ as the dominant variant. The other tokens of this type are analysed in 4.3.2, and the variable is in further detail discussed in 4.4.

Table 15 - Category B: Low/high-status variation.

Type Dialect Variant 1 – Variant 2 – % Var. 1 – all % Var. 1 – % Var. 2- low-status high-status generations youngest youngest variant variant generation generation V09 - Negation Molde /içә/ /ikә/ 70 9 78 adverb

4.3 Category C Category C is the biggest cluster of the three categories, consisting of 40 tokens from 14 types. Tokens that have an overall higher use of Variant 2 are in this cluster. Regarding tokens with an old and a new variant, category C could be the least interesting cluster of the three, since there is not a lot of variation anymore. The use of the new variant is no longer limited to the younger generation, but also being picked up by older dialect users. The 31 tokens where the newer variant has become dominant are analysed in 4.3.1.

49

What is interesting in this cluster is the lower percentage of use in the youngest generation in some tokens. Often, the youngest generation is the driving force behind the change, so the youngest generation then has the highest percentage of use, as is the case in most tokens in the analysis. Seven tokens have a lower percentage of use in the youngest generation than in the generations in total, four of which have variation between a traditional and a new form and three between a high variant and a low variant. A possible explanation could be that this indicates that a new variant is on the rise and that these are the first signs of a new linguistic change. This is further discussed in respectively 4.3.1 and 4.3.2.

As stated above, category C might be the least interesting category, with the least variation and differences between generations. Where in category B the current youngest informants were the starters of the linguistic change, for category C it can be assumed that the process of change did not start with the current youngest generation, but that it was a generation earlier that introduced the new variant, the generation that is now classified as middle-aged. In this light, it is not odd that in a few instances there is a decline in use of that variant by the youngest generation, in favour of another form.

4.3.1 Traditional vs new variant Again, rural dialects cover most of the cluster with variation between a traditional and a new variant. All 25 tokens except for one are variables from rural dialects: Åkrafjorden, Åram, Brattvåg and Hildre, Hå, Midøya, and Ullensvang, with the token from Randaberg as the exception. A large part of the variables consists of old dialect features such as diphthongs, palatalization of velars and palatalization of alveolars that are dissolving. Tokens from these types occur in several dialects, indicating that the same process of dialect levelling is happening simultaneously in different parts of the dialect area.

In the paragraph above it appeared that no interesting variation between generation could be found in category C, but there might be a nuance to that. In this cluster variant 2 is the dominant variant, but when ranking the frequency of use of the generations in total from high to low, the lowest percentage starts at 39% followed by a number of other tokens below or around 50%, while the frequency in the youngest generations ranks from 86% up till 100%. The two tokens with a percentage below 50% both have three variants, which explains why a variant with a frequency below 50% still is the dominant variant. The low percentages do not only show the difference between the generations, moreover, it demonstrates that the categories A, B and C are more of a continuum than three very separate clusters. The set of tokens with low percentages in frequency of

50 use in the total group of informants are in category C, but nevertheless have a lot in common with tokens in category B.

As pointed out in 4.3, some variables have a decline in frequency of use in the youngest generation. Four of those variables are in this cluster, with three tokens for Midøya North and one for Ullensvang. The tokens in Midøya can be explained by the historic border between the romsdals and sunnmørs dialect area (see also 3.1). The north and south side of the island have different traditional variants for V08a+b, the personal pronoun first person singular. What seems to be the case is that the dialect use of the youngest generation from the North and South is getting closer to each other. In Midøya South these tokens show an increase in use of the variant from the north, and in Midøya North there is a slight decrease in frequency of use in favour of the variant from the south. The use of the southern variant is only minimal, a few percent, while the increase of the northern variant on the southside is extensive (Fossheim 2010: 65-67).

Table 16 - Category C: traditional/new variant in Midøya.

Midøya North Midøya South Category % Informants % Y Category % Informants % Y var. 2 Total Total – var. 1 V08a C 94 86 B 54 77 V08b C 92 86 C 39 86

4.3.2 High-status variety vs low-status variety General The fifteen tokens in this cluster are divided over the cities Stavanger, Molde, Bergen and the adjacent suburb/city Fana. In case of ten tokens, the low-variant is the dominant variant 2, and for the other six tokens the situation is the opposite, with a high-variant as the dominant variant 2.

Dominant high variant

To start with the latter, suburb Fana is quite prominent in this cluster. Three out of five tokens are variables from the dialect in Fana. Earlier studies showed more frequent use of high-status variants by residents of Fana, so this might not be a surprising result. Overlap between variables is only found in the determiners in Bergen and Fana. Interesting to notice is that the percentage for the high-status variant is higher in Fana, and also without variation in the youngest generation, whereas in Bergen the percentage is 78% for all generations and still some variation between the low- and high-status variant in the youngest generation. In two other variables, V12 and V15, the high-variant is not a

51

stable variant. With both 51% for the generations in total, it shows that the low- and high-status variant are very much competing.

Table 17 - Category C: Low/-high-status variation.

Type Dialect Variant 1 – Variant 2 – Variant 3 – % Var.2 – % Var. 1 – % Var. 2- Low-status High-status Traditional all youngest youngest variant variant variant generations generation generation V01 – Determiner Fana /myçә/ /myʃә/ /myә/ 100 0 100 “mye” V02 – Determiner Bergen /nᴐkə/ /nʊə/ /nᴐgət/, 78 15 85 “noen/noe” /nᴐkən/ /nʊən/ /nᴐgən/ V02 – Determiner Fana /nᴐkə/ /nʊə/ 94 0 100 “noen/noe” /nᴐkən/ /nʊən/ V12 – Interrogatives Fana /k/- /v/- 51 49 51 with ‘kv-’ V15 - Infinite verb Stavanger a-mål e-mål 51 36 64 ending

Dominant low variant

In the set of variables with a dominant low variant two lexical types are very prominent. Type V09 – the negation adverb “ikke” and V08a/b – the personal pronoun first person singular, stressed and unstressed.

The negation adverb (V09) is a variable in Stavanger, Bergen and Fana, with a high-variant equal to the Eastern-Norwegian form /ike/ and a low-status variant ‘ikkje’, either with a /ç/- or /ʃ/-sound. Even though the low-status variant ‘ikkje’ is the most dominant form in all three tokens, there is quite a lot of variation between the high-status variant and low-status variant. In Stavanger this is not the case, and ‘ikkje’ has a percentage of use of 100%, but in Bergen and Fana the situation is not as stable, especially in the youngest generation. Here we find a lower frequency of use of ‘ikkje’ than in the generations in total. V09 is explained in further detail in 4.4.

52

Table 18 - Category C: High/low-status variation - V09.

Type Dialect Variant 1 – Variant 2 – % Var. 2- all % Var. 1 – % Var. 2 – High-status Low-status generations youngest youngest variant variant generation generation V09 - Negation adverb Stavanger /ikә/ /içә/ 93 0 100 V09 - Negation adverb Bergen /ikә/ /içә/ /iʃә/ 83 32 68 V09 - Negation adverb Fana /ikә/ /içә/ /iʃә/ 68 43 57

Table 19 displays the variants of both types, or in case of V08a and V08b, variant sets. The personal pronoun has various possible pronunciations, especially in Bergen and Stavanger, with the high- variants /jei:/ or /je:/ and low-variants /e:g/, /e:/, /ei/ or /æ:g/. This also explains why the frequency of use of variant 1 (/jei:/ or /je:/) is zero percent, while the frequency of use of variant 2 is not 100 percent. In these cases, another low-status variant than the bold form in the table is used.

Table 19 - Category C: High/low-status variation - V08.

V08 - Personal Dialect Variant 1 Variant 2 Variant 3 % Var.2 – % Var. 1 – % Var. 2- pronoun 1. Person all youngest youngest Singular generations generation generation V08a – stressed Molde /jei:/ /i:/ /e:/ 98 0 100 V08a – stressed Stavanger /je:/ /e:g/ /e:/ 92 0 97 /ei/ V08a - stressed Bergen /jei:/ /je:/ /e:g/ /e:/ 64 0 80 /ei/ V08b – unstressed Molde /jei:/ /i:/ /e:/ 98 0 100 V08b – unstressed Stavanger /je:/ /e:g/ /e:/ 75 0 72 /jei:/ /ei/ /æ:g/ V08b – unstressed Bergen /jei:/ /je:/ /e:g/ /e:/ 65 0 73 /ei/

Both type V08a+b and V09 are high-frequent function words and seem to function as an identity- marker for being Western Norwegian (Sandøy 1985: 139). What these tokens seem to indicate is that speakers in Western Norway feel an urge to use these high-frequent function words to mark their Western Norwegian belonging.

53

4.4 Categories per linguistic variable The previous chapter introduced three categories and the clusters of tokens in those categories. The aim was to compare overlapping variables from dialect studies from the same project to see if the overlapping variables are developing in the same way in the different dialects.

In the tables below, three different percentages are given. The percentage of use of the dominant variant of all generations, the percentage of use of variant 1 by the youngest generation and thirdly, the percentage of use of variant 2 by the youngest generation. Since, depending on the category, the dominant variant of the generations in total can either be variant 1 or variant 2, it is not always clear to which of the variants the percentage of the dominant variety refers. To make this visible, the dominant variant of the generations in total is marked blue. Also, the tables only show two variants. Some tokens have more variants (see chapter 3.4), but since the actual variation only happens between two variants, the tables in the coming paragraphs are limited to those two variants.

Lexical variables

V01 Determiner /mye/

The token in Lista falls into category A, the token in Molde into category C. The two dialects from this variable are geographically far apart, one rural and one urban, the variation differs between variation between an old and a new variant and high-status and low-status variant, and also the variants itself are different, which makes the two tokens hard to compare. The results from V01 make clear that two different developments are happening on both ends of the dialect area.

Table 20 - V01 – Determiner /mye/

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Lista A /møje/, /mø:e/ /my:e/ 83 76 24 Molde C /myçә/, /myʃә/ /myә/ 100 0 100

V02 Determiner /noen/, /noe/

The two tokens in Lista are part of category A, the two tokens in Bergen and Fana are in category C. In V02 the same differences in variation are found as in V01, rural versus urban dialects, variation between an old and a new variant versus variation between a high-status and low-status variant, and also a difference in the variants.

54

Table 21 - V02 - Determiner /noen/, /noe/

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Lista A /nᴐ:en/, /nᴐ:ne/ /nu:en/ 79 72 28

Lista A /nᴐ:ge/, /nᴐ:e/ /nu:e/ 80 68 32 Bergen C /nᴐkə/, /nᴐkən/ /nʊə/, /nʊən/ 78 15 85

Fana C /nɔkən/, /nɔkə/ /nʊːən/, /nʊːə/ 94 0 100

V03 Determiner /det/

V03 has three tokens that all belong to a different category, with Åkrafjorden South in category A , Ullensvang in category B and Åkrafjorden North in category C. Since the dialects are geographically close and the variable is exactly the same in all tokens, this is quite an interesting outcome. All three vary between /de:/ and /da:/, but which of them is the traditional variant differs (see also chapter 4).

Table 22 - V03 - Determiner /det/

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Åkrajorden South A /de:/ /da:/ 80 89 11

Ullensvang B /da:/ /de:/ 67 38 62 Åkrafjorden North C /de:/ /da:/ 74 15 84

V04 Differentiation of /rn/

In this variable the same development is happening in the same pace. All three tokens fall into category B. The development in Øygarden is slightly ahead, the youngest generation already has a 100% use of the new variant, whereas on both sides of Åkrafjorden the traditional variant is still used to some extent by the youngest generation.

55

Table 23 - V04 - Differentiation of /rn/

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Åkrafjorden South B /dn/ /rn/ 51 30 67

Åkrafjorden North B /dn/ /rn/ 58 26 74

Øygarden B /dn/ /rn/ 50 0 100

V05 Segmentation of long /l/

The segmentation of long /l/ is a traditional dialect feature of southwestern dialects. In Randaberg, the segmentation is lost among the youngest generation, and also for the generations in total, the majority has no segmentation anymore. The same process is visible in the rural dialect of Åkrafjorden, but at a slower pace. Just over 50% of the generations in total has maintained the segmented long /l/, but the feature is decreasing. The youngest generation has already abandoned the traditional form.

Table 24 - V05 - Segmentation of long /l/

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Åkrafjorden North B /dl/ /l/ 56 16 84 Åkrafjorden South B /dl/ /l/ 51 9 91 Randaberg C /dl/ /l/ 86 0 100

V06 Loss of lenition

Although the variant with lenition is still the dominant variant for all three tokens, the loss of lenition is at a further stage in Stavanger and Randaberg than it is in Lista. The youngest generation in Lista still has a majority of forms with lenition, in Stavanger the division is even with 50% for forms with and without lenition. In Randaberg the forms without lenition are dominant among the youngest generation.

56

Table 25 - V06 - Loss of lenition

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Lista A /b,d,g/ /p,t,k/ 83 80 20 Stavanger B /b,d,g/ /p,t,k/ 70 50 50 Randaberg B /b,d,g/ /p,t,k/ 83 25 75

V07 Monophtongization of /ɛi/

Monophtongization of /ɛi/ is as good as complete in Åram, Brattvåg and Hildre. All three tokens belong to category C. Not only are all tokens in category C, the youngest generation in both Brattvåg and Hildre has a 100% use of the monophtong /e/.

Table 26 - V07 - Monophtongization of /ɛi/

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation

Åram C ɛi e 81 1 93

Brattvåg C ɛi e 71 0 100

Hildre C ɛi e 70 0 100

V08a + b Personal pronoun – 1. Person singular – stressed and unstressed

V08 is discussed in the previous chapters on the analysis of category A and C (4.1.2, 4.3.1, 4.3.2). The seven tokens of both V08a and V08b exhibit quite some variation. The northwestern dialects have a different set of variants than the southwestern dialects. Also, the tokens in Fana, Bergen, Stavanger and Molde vary between a high-status variant and a low-status variant, the tokens in Åram between a traditional and a new variant, and in Midøya variation is found between a form from the sunnmørs dialect and a form from the romsdals dialect.

In the southwestern dialects, a variant of the low-status variant /e:g/ is the dominant variant, with the exception of Fana where the high-status variant is the dominant form, although the variants have an almost even distribution of percentage of use. In the northwestern dialects, V08 show a clear

57 difference between Romsdalen and Sunnmøre. In Molde and Midøya, it is the romsdal variant that is dominating, while in Åram, the sunnmørs variant is the dominant variant.

Table 27 - V08a - Personal pronoun, 1st p. sg, stressed. Southwestern dialects above the line, northwestern dialects below the line.

V08a - Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – stressed variant- all youngest youngest Dialect generations generation generation Fana A /jɛi/, /æi/ /e:g/, /e:/ 44 45 27 Bergen C /jei/, /je/ /e:g/, /e/, 64 0 80 /ei/ Stavanger C /je:/ /e:g/, /e:/, 92 0 97 /ei/ Midøya B /ɛi/ /i:/ 54 2 77 South Midøya C /ɛi/ /i:/ 94 1 86 North Molde C /jei:/ /i:/ 98 0 100 Åram C - (no /ei/ 100 0 100 variation)

Table 28 - V08b - Personal pronoun, 1st p. sg., unstressed. Southwestern dialects above the line, northwestern dialects below the line.

V08b - Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – unstressed variant- all youngest youngest Dialect generations generation generation Fana A /jæi/, /æi/ /eg/, /e/ 40 45 36 Stavanger C /je:/, /jei/ /e:g/, /e:/, 75 0 72 /ei/, /æ:g/ Bergen C /jei/, /je/ /e:g/, /e/, 65 0 73 /ei/ Midøya C /ɛi/ /i:/ 92 0 86 North Midøya C /ɛi/ /i:/ 39 2 86 South

58

Åram C /i/ /ɛi:/ 55 4 96 Molde C /jei:/ /i:/ 98 0 100

V09 Negation “ikke”

In the set of tokens for V09, the token in Hå deviates. While the other four tokens show an opposition between a high-status variant (/ike/) and a low-status variant (/iҫe/), the token in Hå analyses, similar to V10b, the difference between /iҫe/ and /iʃe/, with /iҫe/ as the dominant, but declining variant. Similarly, in the other four tokens the dominant variant is also /iҫe/, but its position as dominant variant is unstable in some dialects. For Stavanger the situation is clear, with a 100% use of the low-status variant. The situation in Molde, Fana and Bergen is not this stable.

While the older generations in Molde almost exclusively use the low-status variant, the youngest generation has a sudden 78% use of the high-status variant. Rød (2014) gives a number of possible explanations for this abrupt change. Firstly, in the Molde dialect a /ҫ/ in medial position does not occur often, this happens only in three words: ikkje (not), bikkje (dog) and mykje (a lot), while /k/ is common in medial position. Secondly, the form /ike/ is the Bokmål variety and the dominant variant in several cities, for example the nearest biggest town Ålesund (Rød 2014: 113). Rød proposes that the form /ike/ signals something different for the youngest generation than it does for the older generations. While the older generation would notice the form as a high-status variant, for the youngest generation it rather marks an urban identity (Rød 2014: 114).

In Fana and Bergen, the youngest generation has not (yet) switched to the /ike/-variant. As can be seen in the table below, the percentage of use of variant 2 by the youngest generation is lower than the percentage of the generations in total, which means that also in these two dialects, the use of the low-status form is decreasing, while the high-status form /ike/ is expanding.

Table 29 - V09 - Negation "ikke"

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Hå A /iҫe/ /iʃe/ 88 76 24 Molde B /iҫe/ /ike/ 70 9 78 Fana C /ikә/ /içә/, /iʃә/ 68 43 57 Bergen C /ike/ /içә/ 83 32 68 Stavanger C /ike/ /iҫe/ 93 0 100

59

Phonological variables

V10a The sj-sound

The dialects in Stavanger and Bergen fall into category B, with the intermediate form /ʃj/ as the dominant variant, while the development from /sҫ/ to /ʃ/ in Hå already has been completed. Bøe (2013) notices this as well, and explains this by saying that while the development in Hå started after it started changing in Stavanger, the process of change was faster in Hå (Rød 2013: 102).

Table 30 - V10a - The sj-sound

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Stavanger B /ʃj/ /ʃ/ 58 25 75 Bergen B /ʃj/ /ʃ/ 56 18 82 Hå C /ʃҫ/ /ʃ/ 60 2 98

V10b The kj-sound / Merging of the kj-sound and sj-sound

In comparison with V10a, V10b is less far in the process of change. This is no surprise since the merge of the kj- sound and sj-sound is only possible after /sҫ/ has developed to /ʃ/. Or in other words, the change of V10a is a precondition for the merge of kj- and sj-.

The southwestern dialects are furthest ahead in the process of change and fall into category B. The two dialects on the outskirts of the southwestern dialect area belong to category A.

Table 31 - V10b - The kj-sound / Merging of the kj-sound and the sj-sound

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Lista A /ҫ/ /ʃ/ 96 91 9 Molde A /ҫ/ /ʃ/ 92 58 42 Hå B /ҫ/ /ʃ/ 74 44 56 Stavanger B /ҫ/ /ʃ/ 62 32 68 Randaberg B /ҫ/ /ʃ/ 69 26 73 Bergen B /ҫ/, /cҫ͡ / /ʃ/, /ʃj/ 62 6 94

60

V11 Retroflection of /rt/ (/r/ + alveolar)

Midøya, Brattvåg and Hildre are geographically very close. In the results of V11, they all attest a 100% use of variant 2 for the youngest generation.

While the other tokens fall into category C, the dialect in Åram has only minimal use of retroflection and is found in category A. Åram is located more south in Sunnmøre, below Ålesund. The results show clearly that there is a difference between the dialects of Åram on the one side and Hildre, Brattvåg and Midøya on the other side.

Table 32 - V11 - Retroflection of /rt/

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Åram A /rt/ /ʈ/ 84 63 36 Hildre C /rt/ /ʈ/ 50 0 100 Brattvåg C /rt/ /ʈ/ 53 0 100 Midøya C /rt/ /ʈ/ 80 0 100 South Midøya C /rt/ /ʈ/ 99 0 100 North

V12 Interrogatives with /kv-/

The tokens in V12 are in a very different stage in the process of change. In Molde the use of interrogatives starting with /k/ is still in a stable situation in category A. In Fana, the frequency of use is evenly distributed over the two variants, not clear which of the two will win.

Table 33 - V12 - Interrogatives with /kv-/

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Molde A /k/ /ʋ/ 89 86 14 Fana C /k/ /ʋ/ 51 49 51

61

V13 Loss of palatalization of alveolars

The youngest generation in Brattvåg and Hildre does not have any palatalized alveolars anymore, while in Midøya the feature is not fully extinct yet. On the contrary, in Midøya north, the palatalized forms have the highest percentage of use. However, palatalized forms do not make out the majority of the used forms. The youngest generation in Midøya also uses a third, differentiated variant /rn/.

Table 34 - V13 - Loss of palatalization of alveolars

Dialect Category Variant 1 Variant 2 Variant 3 % Dominant % Var. 1 – % Var. 2 – % Var. 3 – variant- all youngest youngest youngest generations generation generation generation

Midøya A /ɲ/ /nn/, /ɳ/ /rn/ 77 47 45 8 North

Midøya B /ɲ/ /nn/, /ɳ/ /rn/ 58 28 66 6 South

Brattvåg C /ɲ/, /ʎ/ /n/, /ɳ/, /l/ 76 0 100

Hildre C /ɲ/, /ʎ/ /n/, /ɳ/, /l/ 75 0 100

Morphophonological variable

V14 Loss of palatalization of velars

From the results it becomes clear that palatalized velars are disappearing from dialects in Western Norway. Six tokens are in category C, with a percentage of (almost) 100% of forms without palatalization for the youngest generation.

The only exception is the dialect in Øygarden. The youngest generation in Øygarden is changing in the same pace as the young generations in the other dialects, but the older generations in Øygarden deviate from the patterns found in the other dialects. It is unclear what the reason is for the slower pace of loss of the old dialect feature in Øygarden.

62

Table 35 - V14 - Loss of palatalization of velars

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Øygarden B /cç/,͡ /cç/, /k/, /g/ 79 6 94 /ɟj/ Åkrafjorden C /cç/,͡ /cç/, /k/, /g/ 51 12 88 North /ɟj/ Ullensvang C /cç/,͡ /cç/, /k/, /g/ 5 2 98 /ɟj/ Åkrafjorden C /cç/,͡ /cç/, /k/, /g/ 59 1 99 South /ɟj/ Åram C /cç/,͡ /cç/, /k/, /g/ 70 0 100 /ɟj/ Brattvåg C /cç/,͡ /cç/, /k/, /g/ 71 0 100 /ɟj/ Hildre C /cç/,͡ /cç/, /k/, /g/ 70 0 100 /ɟj/

Morphological variables

V15 Verb – Infinite verb ending

V15 marks a difference between centre and non-centre. In the western dialect area, the infinite verb has an a-ending. Only in the two centres, Bergen and Stavanger, there is variation between an a- ending and an e-ending. In Stavanger the results slow a slight majority of e-endings. In Randaberg and Øygarden, the a-ending is still the dominant variant, but the youngest generation has an increasing use of e-endings compared to the older generations (Doublet 2015: 65, Villanger 2010: 76). So even though Randaberg and Øygarden fall into category A while Stavanger is in category C, what all three tokens have in common is an increasing use of e-endings.

Table 36 - V15 - Infinite verb endings

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation

63

Randaberg A a-infinitive e-infinitive 76 65 35 Øygarden A a-infinitive e-infinitive 93 62 39 Stavanger C a-infinitive e-infinitive 51 36 64

V16 Verb – Present tense – weak verbs

Even though the tokens had different variants/did not have the same starting point/original form and different motivations for the process of change, the result is the same for all tokens. The e-ending is dominant among the youngest generations in all dialects. Ullensvang, Molde and Hå are all in category C with over 90% use of variant 2 in the youngest generation. The e-ending is less dominant in Øygarden, which falls into category B.

Table 37 - V16 - Present tense - weak verbs

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Øygarden B a-ending e-ending 75 15 70 Ullensvang C ar-ending e-ending 44 2 93 Molde C a-ending e-ending 65 1 94 Hå C a-ending e-ending 64 4 96

V17 Verb – Present tense – strong verbs

Lista and Molde are geographically far apart, with Lista is on the southern end of the western dialect area and Molde on the northern end of the western dialect area. The two very different dialects are difficult to compare. The linguistic change analysed in V17 is not the same for the two dialects and have different variants. Despite these differences, the process of change seems to be similar. The traditional variant is, with about the same percentage, still the dominant variant in both dialects. The youngest generation in Molde has a slightly more frequent use of the new variant.

Table 38 - V17 - Present tense - strong verbs

Dialect Category Variant 1 Variant 2 % Dominant % Var. 1 – % Var. 2 – variant- all youngest youngest generations generation generation Lista A kjeme kome 79 75 25

64

Molde A kjem kommer 75 57 41

Overview

From the results it becomes clear that not all tokens of a variable belong to the same category, or in other words, they are not in the same stage in the process of linguistic change. On the contrary, finding all tokens in the same category is the exception and is the case only for V07 and V17.

Table 39 shows the distribution of all variables over the three categories. The percentage reflects the percentage of use of variant 2 by the youngest generation. The urban centres are marked bold, darkest for the biggest cities Bergen and Stavanger, and a lighter shade Molde, Fana and Randaberg. The three categories reflect a continuum of language change, with category A as the starting phase and category C as the final phase.

A clear tendency becomes visible from table 39, whereby larger, more central dialects are ahead of smaller, rural dialects. The dialects marked bold are in most cases a category ahead, or further ahead in the same category, than other tokens from smaller dialects from the same variable. The biggest deviation from this pattern can be seen in V08a and especially V08b, where smaller dialects surpassed the centre dialects and have a frequency of (almost) 100% of variant 2. The divergence can be explained by the differences between the tokens of V08a+b. The dialects of Bergen, Fana and Stavanger are southwestern dialects, exhibiting variation between a high-status variant /jɛi/ and a low-status variant /eg/, whereas the dialects of Molde, Midøya and Åram are northwestern dialects vary between the forms /ɛi/ and /i/. Since the variation is different for the southwestern and northwestern dialects, the results of V08 should be seen separately for each region, with Molde, Midøya and Åram on one side, and Bergen, Stavanger and Fana on the other side.

Another conspicuous deviator from the general tendency is Hå in V10a, that is ahead of Bergen and Stavanger, which was shortly discussed earlier in this chapter. In the dialect study of Hå (Rød 2013) this was explained by assuming that, even though the change started later in Hå than in Stavanger, the change developed faster in Hå. Although this seems like a plausible explanation, no other examples of such a situation can be found in this analysis, and Hå in V10a is still a divergence from the rest.

65

Table 39 - Overview of division of the variables over the three categories A, B and C with the frequency of use of variant 2-Y.

A B C Var. N 0-25% 25-50% 50-75% 75-100% 40-60% 60-80% 80-90% 90-100% V01 2 Lista Molde V02 4 Lista Lista BERGEN Fana V03 3 Åkra-S Ullensvang Åkra-N V04 3 Åkra- South Øygarden Åkra- North V05 3 Åkra- North Randaberg Åkra- South V06 3 Lista STAVANGER Randaberg V07 3 Åram Brattvåg Hildre V08a 7 Fana Midøya-S BERGEN STAVANGER Midøya-N Molde Åram V08b 7 Fana STAVANGER Midøya-N Åram BERGEN Midøya-S Molde V09 5 Hå Molde Fana BERGEN STAVANGER V10a 3 STAVANGER BERGEN Hå V10b 6 Lista Molde Hå BERGEN STAVANGER Randaberg V11 5 Åram Hildre Brattvåg Midøya-S Midøya-N V12 2 Molde Fana V13 4 Midøya-N Midøya-S Brattvåg Hildre V14 7 Øygarden Åkra-N Ullensvang Åkra-S Åram Brattvåg Hildre V15 3 Randaberg STAVANGER Øygarden V16 4 Øygarden Ullensvang Molde Hå V17 2 Lista Molde Tot. 76

Table 39 indicates a pattern in dialect change, in which urban dialects are ahead of peripheral dialects. Although it shows a pattern, it does not hold any information about the impact of the factor urban/pheripheral dialect. In order to get a better insight in the impact, a multiple regression was

66 carried out in which three factors are taken into account. Factor X1 displays language level, X2 is the logarithm of the number of citizens and X3 is a combined factor of region and centre/periphy. The language levels in X1 rank from 0 till 3, with 0 for morphology, 1 for phonology, 2 for lexicon and 3 for morpho-phonology. The ranking is based on the average percentage of variant 2 among the youngest generation. X3 is a combined factor of region (southwestern; northwestern) and periphery (periphery; centre), with 0 for periphery x southwestern and 1 for any other combination (periphery x northwestern, centre x southwestern, centre x northwestern). The peripheral dialects in the northwestern region had a comparable average of the categories A/B/C and is therefore marked in the same way as the centre dialects.

The results are presented in table 40, showing the highest correlation with Y for X3 (0.412), followed by X1 (0.247) and X2 (0.238), with an adjusted multiple R2 of 0.2427 (p<.0001). The results for X3 strengthen the pattern that appears in table 39, indicating that centre/periphery does matter in the process of change. Nevertheless, there is a difference between peripheral dialects in the northwest and the southwest. It appears that the northwestern region is more sensitive to language change than the southwestern region.

Table 40 - Correlation matrix of a multiple regression of three factors in dialect change: X1: language level, X2: logarithm of number of citizens, X3: region/periphery, Y: Category A, B, C.

X1 X2 X3 Y b B B x r-xy

X1 1 -0.188 -0.093 0.247 X1 0.3112 0.3122 0.077

X2 -0.188 1 0.357 0.238 X2 0.125 0.1594 0.0379

X3 -0.093 0.357 1 0.412 X3 0.6966 0.3839 0.1581

Y 0.247 0.238 0.412 1 Multiple R2 = 0.273 Adjusted Multiple R2 = 0.2427 p <.0001

The factor language level (X1) is another determining factor in dialect change, after X3. A morpho- phonological variable is more sensitive to change than a lexical variable, which is in turn more sensitive to change than a phonological variable, with morphological variables as the least sensitive. In the graph below the level of is presented per linguistic level, divided in southwestern periphery and others (northwestern periphery, southwestern centre, northwestern centre), showing an upgoing line from phonology to morpho-phonology. Morphology is absent from the graph, since it

67 did not give any results. A possible explanation might be that there are as many morphological variables in category A as there are in category C and therefore does not show any trend. Another remark about the graph concerns the morpho-phonological items. It is important to note that there is only one morpho-phonological variable, V14 – loss of palatalization of velars. With seven tokens the variable has a higher amount of tokens than most other variables, but the results from the graph need to be interpreted carefully since it only involves one type.

Figure 3 - Linguistic levels as a driving force

5. Discussion

The current paper analysed 76 linguistic variables from 13 dialect studies in Western Norway, using three categories, A, B and C, that each represent a different stage in the process of language change.

I first hypothesized that geographically limited features are decreasing in use in favour of more regional or national features. Particularly category B and C contain tokens typical for dialect levelling, such as loss of lenition, loss of palatalization, loss of differentiation and loss of segmentation. These are all features that define a dialect of a geographically limited area. This geographically limited area for the typical dialect levelling tokens is the southwestern dialect area, with the exception of loss of palatalization of velars which is also found in the northwestern area. However, these are all regional features. Very local features, specific for particular dialects, are not found in the analysis. Possibly this is caused by the fact that only overlapping variables are analysed in this paper.

68

Secondly, I hypothesized that linguistic variables from deeper linguistic levels are changing at a slower pace than variables from lighter linguistic levels. In the analysis it was found that the linguistic level of a variable has an impact on the process of change, with morphological items as the least sensitive to language change, followed by phonological items, then lexical items, and morpho- phonological as the most sensitive to language change. Yet, there are a few remarks. Firstly, not all linguistic levels are represented in the data. The variables are limited to phonology, morphology, morpho-phonology and lexicon. No syntactic or pragmatic variables are found in the data. Secondly, as mentioned before, the morpho-phonological variable is only one variable, V16, and this might lead to a false impression of the results.

Third, I hypothesized that urban dialects are further ahead in the process of dialect change than rural dialects. This is accurate for all but one variable, V10a – the sj-sound, where the variable in Hå is in category C, whereas Bergen and Stavanger belong to category B. In the original dialect study, it was suggested that the change might have started later, but was completed sooner than in Stavanger because the language community is smaller (Rød 2013: 102). It seems like a plausible explanation, although no other examples of such a case are found in this paper. In table 39, several other urban dialects are surpassed by rural dialects, such as V08, but they are different in the sense that the ongoing change in the dialects is different. The surpassing rural dialects are northwestern dialects with different variants of V08 than the variants of the southwestern urban dialects. Another remark on the hypothesis is the difference between northwestern and southwestern peripheral dialects. Northwestern peripheral dialects are more often found in category C and appear to be more sensitive to language change than the southwestern peripheral dialects. Possibly this is a result of the nearness of the trønder dialect area.

The fourth and last hypothesis was that cities or towns with a centre function influence the direction of dialect change in the periphery. As became clear from the third hypothesis, urban dialects, which have a centre function, are ahead in the process of change. Although this alone does not confirm the fourth hypothesis, in combination with the idea that language change starts in cities before spreading to the surrounding area, it makes the fourth hypothesis plausible. Nevertheless, to either confirm or contradict the hypothesis, a more thorough investigation needs to be carried out, pointing out the possible sources of change for each of the 76 tokens and especially the tokens of peripheral dialects.

When I initially started comparing the results from the dialect studies of the Processes of Dialect Change project, I was hoping to find overlapping variables with tokens in more dialects than turned out to be the case. Many variables have tokens in only two or three dialects, that are sometimes

69 geographically and linguistically quite far apart, asking for a different approach in comparing the results than I initially intended to do.

In the cluster of variety between a high-status and a low-status variant, the majority of the variables showed a movement to use the high-status variety form, and this is also well-established with most tokens in category C. Two variables deviate from this general trend, V08 and V09, the personal pronoun first person singular and the negation adverb ‘ikke’. The low-status variant is the most frequent variant in these cases, with the exception of the dialect in Fana. For variable V09, the dominance of the low-status variant is unstable, with a decreasing use by the youngest generation. For V08, it is the low-status form that holds the dominant position. A probable reason for the dominating low-status variant is the function of the personal pronoun as an identity marker (Sandøy 1985: 139). Language users might use the low-status variant as a way of expressing their Western Norwegian identity.

6. Conclusion

The analysed linguistic variables in this paper show a number of aspects about the process of language change in Western Norwegian dialects. The biggest share of the tokens is in the process of change where not only the youngest generation, but the generations in total have adopted a new variant. The new variant is either a form that replaces a traditional dialect feature, or a high-status or low-status variant. The general tendency showed a higher use of high-status variants.

Several variables were found that signal dialect levelling, in which traditional features of a dialect area are diminishing. Regarding the pace of dialect change, dialects with a centre function are ahead of peripheral dialects. There is, however, a difference between northwestern and southwestern peripheral dialects. Northwestern peripheral dialects appear to be more sensitive to language change than southwestern dialects. Another factor that was found to have an effect on the pace of change is the linguistic level. The morpho-phonological variable showed to be most susceptible to change, followed by lexical items.

References

Aasen, Kristine Nymark. 2011. Stavanger-dialekten 30 år etter. Ei sosiolingvistisk oppfølgingsgransking av talemålet i Stavanger. (MA-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies.

70

Akselberg, Gunnstein. 2005. Dialects and regional linguistic varieties in the 20th century II: Norway. The Nordic languages: an international handbook on the history of the North Vol. 2. (red. Oskar Bandle & Lennart Elmevik). : De Gruyter. pp.1707- 1721.

Akselberg, Gunnstein. 2009. Er det fruktbart å applisera termen standardspråk på norske talemålstilhøve? Norsk Lingvistisk Tidsskrift, Vol.27(1), pp.67-79.

Anderson, Ragnhild. 2013. Metodisk opplegg for feltarbeid på Vestlandet. Available at: https://folk.uib.no/hnohs/DEP/Metodisk%20opplegg,%20versjon%206.pdf.

Berg, Ivar. 2018. Tida fram til 1600. Norsk Språkhistorie II: Praksis (red. Brit Mæhlum). Oslo: Novus AS.

Berntsen, Mandius & Amund B. Larsen (1925). Stavanger bymål. Oslo: Aschehoug.

Bøe, Per Sigmund Sævik. 2013. Språkendringer sør på Jæren. En sosiolingvistisk studie i virkelig tid fra Ogna. (Ma-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies.

Doublet, Maria-Rosa. 2012. Bare fra Bergen, eller fra Fana i Bergen? En intern språkkrig mellom Bergensvarietetene. (MA-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies.

Doublet, Ingrid. 2015. Randaberg, den grøne landsbyen. Ei sosiolingvistisk undersøking av talemålet i Randaberg. (MA-thesis.) : .

Fossheim, Marie. 2010. Språket på Midøya – en oppfølgingsstudie av talemålet på ei øy i Romsdalen. (MA-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies.

Heradstveit, Rut. 2018. Språkendringer på Lista. En sosiolingvistisk studie i tilsynelatende tid. (MA-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies.

Hildremyr, Helene. 2006. Dialekten I Brattvågen og på Hildre – 23 år etter. Ei sosiolingvistisk oppfølgingsgransking av talemålet i to sunnmørsbygder. (MA-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies.

Jahr, Ernst Håkon. 1990. Den Store Dialektboka. Oslo : Novus.

Kroken, Agnes Thrastardottir. 2014. Talemålet i Bærum. En sosiolingvistisk undersøkelse av talemålet hos unge i Bærum. (MA-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies.

71

Larsen, Amund B. 1906. The Dialects of Norway: A Survey of Their Characteristics. The Journal of English and , Vol.6(1), pp.99-114. Larsen, Amund B. 1907. Kristiana bymål. Vulgærsproget med henblik på den utvungne dagligtale. Kristiania: Cammermeyer. Larsen, Amund B. & Gerhard Stoltz. 1911-1912. Bergens bymål. Kristiania: Aschehoug & co. Lianes, Anna. 2013. Endringar i talemålet i Åram på Søre Sunnmøre. Ei sosiolingvistisk gransking i tilsynelatande tid. (MA-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies. Lid, Hallvard. 2015. “Men Odda e’ jo da naturle’ sentere’ fý oss, dao”. Ei sosiolingvistisk gransking av variasjon og endring i dialekta i Ullensvang herad. (MA-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies. Mæhlum, Brit. Standardtalemål? Naturligvis! En argumentasjon for eksistensen av et norsk standardtalemål. Norsk Lingvistisk Tidsskrift, Vol.27(1), pp.7-26. Mæhlum, Brit & Unn Røyneland. 2012. Det Norske Dialektlandskapet: Innføring i Studiet Av Dialekter. Oslo: Damm Akademisk. Nornes, Marianne Valeberg. 2011. Bergensk i Bergenhus – ei sosiolingvistisk oppfølgingsgransking av talemålet i Bergenhus bydel. (MA-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies. Rød, Hild Meland. 2014. Molde bymål. Ein kvantitativ korrelasjonsstudie i verkeleg tid. (MA-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies. Røyneland, Unn. 2009. Dialects in Norway: catching up with the rest of ? International Journal of the Sociology of Language, Vol.2009(96), pp.7-30. Sandøy, Helge. 1985. Norsk Dialektkunnskap. Oslo: Novus Forlag. Sandøy, Helge & Gunnstein Akselberg & Gjert Kristoffersen. 2007. Dialektendringsprosessar. Prosjektskisse. Bergen: Universitetet i Bergen. Available at: http://folk.uib.no/hnohs/DEP/Prosjektbeskriving,%20Dialektendringsprosessar.pdf. Sandøy, Helge. 2003. Kontakt og spreiing. Språkmøte. Innføring i sosiolingvisikk (red. B. Mæhlum, G. Akselberg, U. Røyneland and H. Sandøy). Oslo: Cappelen akademisk forlag. 224- 246. Sandøy, Helge. 2009. Standardtalemål? Ja, men ...! Ein definisjon og ei drøfting av begrepet. Norsk Lingvistisk Tidsskrift, Vol.27(1), pp.27-48. Sandøy, Helge. 2018. Geografisk og sosial variasjon. Norsk Språkhistorie II: Praksis. (red. Brit Mæhlum). Oslo: Novus AS.

Skjekkeland, Martin. 2005. Dialektar i Noreg. Tradisjon og fornying. : Høyskoleforlaget.

72

Tjelmeland, Liv Thorunn. 2016. Språket i Åkrafjorden – ei sosiolingvistisk gransking av dialekten i to fjordbygder. (MA-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies.

Trudgill, Peter. 2011. Sociolinguistic Typology. Social Determinants of Linguistic Complexity. Oxford: University Press.

Villanger, Silje. 2010. 'Da hadde vært litt kult visst vi hadde snakka strilsk om hondra år'. Ei sosiolingvistisk opfølgingsgransking av talemålet i Øygarden. (MA-thesis.) Bergen: Institute of linguistic, literary and aesthetic studies.

73

APPENDIX

V01 – DETERMINER /MYE/

Fana

Lista

V02 – DETERMINER /NOEN/, /NOE/

Bergen

74

Lista

Fana

V03 – DETERMINER /DET/

Åkrafjorden

Ullensvang

75

V04 – DIFFERENTIATION OF /RN/

Åkrafjorden

Øygarden

V05 – SEGMENTATION OF LONG /L/

Åkrafjorden

76

Randaberg

V06 – LOSS OF LENITION

Lista

Randaberg

Stavanger

77

V07 – MONOPHTONGIZATION OF /ɛI/

Åram

>>Brattvåg and Hildre down below

V08A – PERSONAL PRONOUN – 1. PERSON SINGULAR – STRESSED

Bergen

Fana

78

Stavanger

Molde

Midøya

79

Åram

V08B – PERSONAL PRONOUN – 1. PERSON SINGULAR – UNSTRESSED

Bergen

Fana

80

Stavanger

Molde

Midøya

81

Åram

V09 – NEGATION ‘IKKE’

Bergen

Fana

82

Stavanger

Molde

83

V10A – THE SJ-SOUND

Bergen

Stavanger

84

V10B – THE KJ-SOUND & MERGING OF KJ- AND SJ-

Bergen

Stavanger

Randaberg

85

Lista

Molde

V11 – RETROFLECTION OF /RT/ (/R/ + ALVEOLAR)

Åram

86

Midøya

>> Brattvåg and Hildre down below

V12 – INTERROGATIVES WITH /KV-/

Fana

Molde

87

V13 – LOSS OF PALATALIZATION OF ALVEOLARS

Midøya

>> Brattvåg and Hildre down below

V14 – LOSS OF PALATALIZATION OF VELARS

Åram

Øygarden

88

Åkrafjorden

Ullensvang

>> Brattvåg and Hildre down below

V15 – VERB – INFINITE VERB ENDING

Øygarden

89

Stavanger

Randaberg

V16 – VERB – PRESENT TENSE – WEAK VERBS

Molde

Øygarden

90

Ullensvang

V17 – VERB – PRESENT TENSE – STRONG VERBS

Lista

Molde

91

Brattvåg

Hildre

V07 – monophtongization of /ɛi/ : V09 V11 – retroflection of /rt/: V03 V13 – palatalization of alveolars: V01 V14: palatalization of velars: V02

92