1. Introduction
People commonly sort their experiences into kinds (Lakoff Reference Lakoff1987):
Every time we see something as a kind of thing, for example, a tree, we are categorizing. Whenever we reason about kinds of things – chairs, nations, illnesses, emotions, any kind of thing at all – we are employing categories. Whenever we intentionally perform any kind of action, say something as mundane as writing with a pencil, hammering with a hammer, or ironing clothes, we are employing categories. (Lakoff Reference Lakoff1987:5−6)
Categorization processes such as these are grounded in our body–world interactions and essential for making sense of the world (Lakoff Reference Lakoff1987). By means of them, we organize our experiences into categories (here defined as a number of entities that are considered equivalent; see Rosch Reference Rosch, Rosch and Bloom Lloyd1978:30), which are designated by the terms that speakers use (Rosch, Reference Rosch, Rosch and Bloom Lloyd1978:30). Categories may be sorted into cognitive taxonomies (i.e. kind-of-hierarchies based on kind-of relations; chairs and desks are kinds of furniture [Mark et al. Reference Mark, Smith, Tversky, Freksa and Mark1999:285], and Sweden and Norway are kinds of countries) and partonomies (i.e. part-of hierarchies based on part-of relations; seconds are parts of minutes, minutes are parts of hours, and hours are parts of days [Mark et al. Reference Mark, Smith, Tversky, Freksa and Mark1999:285]; Stockholm is part of Svealand, Svealand is part of Sweden, Sweden is part of Europe). However, entities that are part of the same geographical category are not necessarily categorized the same way by the speakers of a language. Consider the following och Norrland sentences from the three subcorpora of governmental, news, and social media texts (henceforth abbreviated GT, NT, and SMTFootnote 1 ) from the Swedish Korp corpus (Borin et al. Reference Borin, Forsberg and Roxendal2012). They all include the nounFootnote 2 Norrland, which denotes one of Sweden’s three main regions, Norrland, Svealand, and Götaland.
In sentence (1), Norrland is coordinated with nouns for towns within Götaland (Malmö and Simrishamn). The sentence specifies from what towns in Götaland there was only one representative (of something) without providing the same level of detail in the information about the representative from a place in Norrland. All we learn is that he or she is from Norrland. Sentence (2) reflects more complex and, indeed, incorrect categorization. Here, Norrland is coordinated with nouns for towns in Svealand (Stockholm and Falun-Mora), Götaland (Göteborg), and Norrland (Sundsvall) and with nouns for counties in Götaland (Skåne) and Norrland (Jämtland). From this sentence, we learn that the demand for train tickets with SJ is the biggest between towns and counties (within or outside of Norrland) and the region of Norrland. It suggests to the speaker that ‘Norrland’ is something other than that of the Norrland town of Sundsvall and the Norrland province of Jämtland. In sentence (3), Norrland is again coordinated with the nouns for the province of Skåne and the town Göteborg in Götaland, and with the town Stockholm in Svealand, but also with the counties or provinces Värmland and Dalarna in Svealand. It again shows that Norrland may be coordinated with nouns for more specific geographical categories within the other two regions of Sweden. As in sentence 1, places in Norrland are not categorized as instances of the more specific categories they are part of (e.g. as towns and counties) but sorted into the larger category of Norrland. Furthermore, from the sentence, we learn that to the speaker, the actual marks of a person from Stockholm are to consider Sweden to consist of the towns of Stockholm and Göteborg, the counties or provinces of Värmland and Dalarna, and then Norrland. In that sense, it provides information on the speaker’s attitude towards these geographical categories. When considering some geographical areas, categorization at the levels of towns (Stockholm and Göteborg) or counties or provinces (Värmland and Dalarna) is essential. However, regarding Norrland, categorization at the region level appears to be sufficient. Specific places within this region are not categorized as such but sorted into the broader category of Norrland.
Sentences (1)–(3) thus suggest a potential asymmetry in how geographical categories are categorized, particularly in how Norrland and the other two main regions of Sweden are categorized. Although Norrland, Svealand, and Götaland refer to the same geographical category (i.e. to regions), they do not seem to be treated equally.
This study investigates the categorization of geographical locations, as reflected in phrases including the nouns Norrland, Svealand, and Götaland and the Swedish conjunction och (‘and’). The study is based on corpus linguistic analyses of up to 1,000 random instances of sentences including each of the phrases och Norrland, och Svealand, och Götaland, Norrland och, Svealand och, or Götaland och from the GT, NT, and SMT subcorpora of the Swedish Korp corpus (Borin et al. Reference Borin, Forsberg and Roxendal2012). The purpose of the study was to test the hypothesis that the asymmetries reflected in sentences (1)–(3) are by no means unusual. Analyses of the representations of Norrland in contemporary newspaper texts (Eriksson Reference Eriksson2010) have shown that Norrland is discursively constructed as an abstract space with certain specific connotations. However, we do not know how common such usage patterns are and whether Norrland and areas within Norrland differ from Svealand and Götaland and areas within these regions in this regard.
The following research question were asked:
-
RQ1. Are Svealand and Götaland more commonly used with nouns for one of the three main regions compared with Norrland?
-
RQ2. Is Norrland more often used with nouns for geographical categories within regions than Svealand and Götaland?
We expected Norrland to be coordinated with nouns for regions less commonly than Svealand and Götaland, but more frequently with phrases for geographical places within regions.
The paper is structured in the following way. A short background to the regions of Sweden and a discussion of the functions of the nouns Norrland, Svealand, and Götaland is provided in Section 2, and a theoretical background to research on cognitive categorization processes in Section 3. Section 4 deals with the materials and method used and Section 5 the results. We conclude with a discussion of the results, limitations of the study, and prospects for future research in Section 6.
2. The regions of Norrland, Svealand, and Götaland
Sweden is the fifth largest country in Europe, and with its elongated form (1,500 km) similar in length and size to California (https://sweden.se/life/society/key-facts-about-sweden). It has twenty-five provinces (Sw. landskap) historically grouped into three regions (Sw. landsdelar): Norrland, Svealand, and Götaland (Teleman et al. Reference Teleman, Hellberg and Andersson1999). The entities of landskap and landsdel do not fill any administrative functions today. Instead, riksområde is used for statistical purposes in relation to the territorial units in the European Union, NUTS (Nomenclature des Unités Territoriales Statistiques) (Förordning (EU) 2019/2088 av den 27 november 2019 om hållbarhetsrelaterade upplysningar som ska lämnas inom den finansiella tjänstesektorn 2019), and län for administrative purposes. Still, references to the historical regions are commonly made in news and weather reports.
The climate in the north of Sweden is sub-arctic with long, cold winters and short summers, while in the south, the climate is semi-boreal with a damper climate and milder winters with few days of snow. However, Sweden has a milder winter climate than other areas on the same latitude (https://www.smhi.se/kunskapsbanken/klimat/sveriges-klimat/sveriges-klimat-1.6867).
Sweden has a long coastline on the east and southwest coasts, while the Scandes (Sw. Skanderna) mountain range delimits the country to Norway in the northwestern part. As much as 69% of Sweden is covered with forests (https://sweden.se/life/society/key-facts-about-sweden). Despite its size, Sweden is one of Europe’s most sparsely populated countries (with 21 inhabitants per square kilometer). Still, most Swedes live in urban areasFootnote 3 (Sw. tätorter). According to SCB (https://www.scb.se/hitta-statistik/sverige-i-siffror/miljo/tatorter-i-sverige/), 72% of inhabitants in the least populated province of Norrland live in such areas and 83% of inhabitants in the most populated province. The equivalent numbers for Svealand are 83% in the least populated province, and 97% in the most populated one, and for Götaland 64% and 91%, respectively.
The urban areas are primarily located along the Swedish coasts or around the lakes of Mälaren, Hjälmaren, Vättern, or Vänern. Norrland and Svealand are particularly unevenly populated, with substantially smaller numbers of inhabitants inland. The three biggest municipalities in Sweden are its capital Stockholm (with 1.6 million inhabitants), Göteborg (with 608,000 inhabitants), and Malmö (with around 325,000 inhabitants).
Norrland is the largest and northernmost of the three regions but has the smallest population. It covers about 58% of Sweden (261,292 square kilometers), and about 12% of the Swedish population (about 1.17 million inhabitants) live there. Still, six of Sweden’s thirty largest urban areas are in Norrland, two of which have more than 100,000 inhabitants (https://www.scb.se/hitta-statistik/sverige-i-siffror/miljo/tatorter-i-sverige/). They are generally located along the coast, while areas in the western part of Norrland are more sparsely populated.
Svealand is the smallest of the regions and is located between Norrland and Götaland. It covers about 20% of Sweden (91,098 square kilometers), and about 40% of Swedes (about 4 million people) live there. Ten of the thirty most populated municipalities are in Svealand. Eight of them have more than 100,000 inhabitants (https://www.scb.se/hitta-statistik/sverige-i-siffror/miljo/tatorter-i-sverige/). Several of them are located around the lake Mälaren. Historically, Svealand was Sweden’s administrative center (Asker Reference Asker2009), which is also reflected in contemporary Sweden.
Götaland is the southernmost and most densely populated of the regions. It covers about 22% of Sweden (97,841 square kilometers), and about 48% of Sweden’s inhabitants (about 4.78 million people) live there. Nearly half of the thirty most populated areas in Sweden are in Götaland. Nine of them have more than 100,000 inhabitants (https://www.scb.se/hitta-statistik/sverige-i-siffror/miljo/tatorter-i-sverige/).
The nouns Norrland, Svealand, and Götaland are all proper names, which prototypically have unique referents (Teleman et al. Reference Teleman, Hellberg and Andersson1999:117). This feature distinguishes proper names from appellatives, and therefore proper names are sometimes treated as a separate part of speech. In addition to grammatical differences, the main differences between proper names and nouns are semantic. Proper names, like nouns, have referents, but they are labels for unique referents rather than referents with functional traits or qualities in common. Different places may have the same place name even if they do not share any features besides those related to topography. Such meanings have often faded but are obvious in nature names with suffixes such as -vik (‘-bay’) and -berg (‘mountain’). Ainiala & Östman (Reference Ainiala and Östman2017:3) imply that etymology is largely irrelevant when names are used in context. A much-debated question reflecting the distinctive feature between proper names and appellatives is whether proper names have meaning or sense in addition to reference (for an overview see Helleland Reference Helleland, Helleland, Ore and Wikström2012:99).
Because place names are not associated with semantic meaning like common nouns, but primarily have a referential function, it is not always clear what referents the Norrland, Svealand, and Götaland instances have. The reference of a place, or speakers’ conceptions of this reference, may be very wide, as in Sweden, or very narrow, as in cultural names like Pekkantalo (‘the homestead of Pekka’). Norrland, Svealand, and Götaland denote large regions, and speakers might not always have the whole regions in mind when using them but may use the terms metonymically for more specific areas that make up parts of the regions (e.g. for towns and provinces within them). Such broad categorization may be imprecise (as reflected in sentence 1), incorrect (as in 2), or sweeping, as in (3), and involve asymmetry in speakers’ categorization of place. Furthermore, it is not always clear, or agreed upon, where the borders of some geographical areas really are. For example, where do mountains such as Kebnekaise and Mount Everest begin and end, and where were the borders of Rome in 410 AD? Moreover, geographical names may be used in reference to several overlapping areas and several different concepts. Previous research has shown that Norrland is not only used in reference to a Swedish region but about a combination of geographies, histories, climates, borders, and ideas (Eriksson Reference Eriksson2010:26). First, in territorial terms it is equal to the sum of Sweden’s five northernmost counties (Eriksson Reference Eriksson2010:28). Second, it is used for a somewhat larger area north of the climatological border, Limes Norlandicus (Sw. biologiska norrlandsgränsen) (Rune Reference Rune1995). Third, cultural boundaries between Norrland and the rest of Sweden can be drawn by mapping patterns in material cultural heritage, such as settlements and artifacts, and immaterial ones, such as language varieties. The concept of kulturgräns (‘cultural border’) by the ethnologist Sigurd Erixon (Edlund & Frängsmyr Reference Edlund and Frängsmyr1995), has been developed in, for instance, the project Kulturgräns norr (Edlund & Greggas Reference Edlund and Greggas2000). Fourth, the concept of Norrland and speakers’ attitudes towards this region are discursively constructed. Eriksson (Reference Eriksson2010) has analyzed popular representations of the Swedish north in contemporary news articles about Norrland retrieved from the database Presstexts. The study is based on articles from the daily newspaper Dagens Nyheter from the years 2000–2004 retrieved in two searches, one with the keywords Jämtland, Gävleborg, Norrbotten, Västerbotten, and Västernorrland and one with the keyword Norrland. Using critical discourse analysis tools, she shows that the texts presuppose that Norrland is a region in need of support, with migration from the area and a surplus of men, without critically revising the claims made. Similarly, its counties are constructed as sparsely populated areas without acknowledging variations within them (Eriksson Reference Eriksson2010:52). Her analysis (Eriksson Reference Eriksson2010:57) of whether the articles with the keyword Norrland further specify the specific areas involved (e.g. counties, municipalities, or cities) shows that only seventeen out of thirty-eight articles provide this information. When the areas are specified, the information typically comes at the end of texts referencing Norrland. She suggests that in this way, the areas are represented as familiar but abstract spaces with well-known connotations. Furthermore, specified areas primarily involve the northernmost counties (Norrbotten and Västerbotten) and use specific aspects of them to represent the whole county. Symptomatically, the province of Dalarna (a mountainous province with traditional rural areas that is a part of Svealand) is more often labeled Norrland than the Norrland coastal province of Gävleborg (Eriksson Reference Eriksson2010:57). Eriksson (Reference Eriksson2010:57) thus argues that journalistic practices of framing the area reinforce stereotypes. Through them, certain aspects of the region become universalized, and Norrland becomes a place that lacks variation. Contemporary representations of Norrland are related to the broader and much discussed dichotomy between center and periphery (Eriksson Reference Eriksson2010:25), and historical issues related to societal questions in the comparatively poor north that were much debated when the northern part of Sweden became industrialized during the late nineteenth and early twentieth centuries (Edlund & Frängsmyr Reference Edlund and Frängsmyr1995:293). Because the natural resources in Norrland were essential to industrialization and hence played an important part in the development of modern Sweden, Norrland was then framed as ‘Framtidslandet’ (‘The Land of the Future’, Sörlin Reference Sörlin2023).
To conclude, place names are not only expressions denoting specific places. Besides their referential function, they have social and cultural functions (Ainiala & Östman Reference Ainiala and Östman2017:3). They vary in style according to situation, reflecting speakers’ conceptions of and attitudes towards such places. Speakers also have feelings attached to certain places. Helleland (Reference Helleland, Helleland, Ore and Wikström2012:109) reflects upon place identity, arguing that:
names are abstractions of the places they refer to, substituting physical features with a wide range of expressions … place names constitute links between name users and the named objects, as well as with the past.
In this sense, place names also have a symbolic function.
3. Cognitive categorization processes
Early work on categorization processes includes Rosch’s (Reference Rosch1973) work on natural categories, and Medin and Schaffer’s (Reference Medin and Schaffer1978) work on exemplar-based reasoning. Rosch demonstrated that category judgments are based on comparisons between new instances and perceptually salient natural prototypes of each category, and Medin and Shaffer (Reference Medin and Schaffer1978) showed that comparisons with exemplars (i.e. remembered examples of a category) and the contexts of these memories are also involved when people form categories.
It is well established that linguistic labels impact categorization (Althaus & Plunkett Reference Althaus and Plunkett2016:770). For instance, one study by Lupyan et al. (Reference Lupyan, Rakison and McClelland2007) has shown that linguistic labels can facilitate category learning. In the study, participants were first taught to categorize ‘alien’ as ‘those to be approached’ or ‘those to be avoided’ (Lupyan et al. Reference Lupyan, Rakison and McClelland2007:1007), and then given accuracy feedback that was either accompanied by nonsense labels of the categories presented or by no linguistic label at all. The study showed that even though all participants had been given the same accuracy feedback and the labels were essentially redundant, those given labels for the categories they were shown were better at learning the categories than those who had not. Yet other studies have shown that linguistic labels can impede the identification of visual categories (e.g. Robinson & Sloutsky Reference Robinson and Sloutsky2007) and that labels can induce a focus on commonalities between the entities that are part of a given category. One example of the latter finding is Althaus and Plunkett’s (Reference Althaus and Plunkett2016) study of the eye movement of 12-month-olds while these infants were familiarized with the categories of leaves and shells. The study showed that labels influence how categories are processed. Infants presented with labels for the objects they were shown noticed their commonalities. In contrast, those not shown such labels focused on the variability between the objects displayed.
The level of categorization that is most basic to human experience (henceforth basic-level category) includes objects and organisms such as dogs, cats, cars, buses, chairs, and sofas (Rosch Reference Rosch, Rosch and Bloom Lloyd1978:30), which may be included in larger, more inclusive and hence more abstract superordinate categories (e.g. dogs and cats may be categorized as kinds of animals, cars and buses as kinds of vehicles, and chairs and sofas as kinds of furniture), or subdivided into smaller, less inclusive, more specific subordinate categories (e.g. dogs may be subdivided into the categories of Alsatians and Poodles, cars into the categories of Volvo and Porsche, and chairs into armchairs and office chairs). However, categorization processes are not neutral. Rosch (Reference Rosch, Rosch and Bloom Lloyd1978) argued that unexpected precision in the way objects are categorized (e.g. referring to one’s pencils as one’s Mongol number 3 pencils) can have a snobbish or satirical effect, and Cruse (Reference Cruse1977) argued that lower degrees of specificity de-emphasize features that are omitted while higher degrees of specificity emphasize the features that are added by a more specific category.
The parts of geographical partonomies are all basic-level categories. Still, the labels speakers use (e.g. Sweden, Svealand, and Stockholm) about the same geographical area (e.g. Stockholm) may differ in their specificity. In one case, the categories employed may be fine-grained and specific (e.g. refer to towns, or parts of towns), and in another, broad and imprecise (e.g. refer to regions, countries, or continents).
Based on the results of empirical investigations such as those above, we take the nouns that speakers use for geographical places to reflect speakers’ categorization of these places and mediate how they categorize these places to others. Next, we describe the methods and materials employed in our analysis of categorization reflected in speakers’ uses of the phrases och Norrland, och Svealand, och Götaland and Norrland och, Svealand och, and Götaland och.
4. Materials and method
The och region subsets were excerpted on 18 February 2023 from the GT, NT, and SMT subcorpora of the Swedish Korp corpus (Borin et al. Reference Borin, Forsberg and Roxendal2012). The subcorpora were chosen because they include diverse types of texts written by different speakers of Swedish and are intended for different audiences. GT is the most official subcorpus with texts of Swedish agencies and authorities by representatives of the public, NT the most public one with texts in standard, written Swedish by skilled writers (journalists) for large readerships, and SMT the most informal one with texts written by people expressing their views on topics online. Given these differences, GT provides information on categorization by speakers of Swedish in official contexts, NT on categorization by speakers in public contexts, and SMT on categorization by speakers typically representing themselves.
At the time of excerpting the data, GT comprised twenty-five subcorpora and around 128.8 million sentences, NT fifty-nine subcorpora and 54.9 million sentences, and SMT sixty-nine subcorpora and 843.5 million sentences. The region och subsets were excerpted from the same subcorpora on 10 June 2024. GT then comprised thirty-five subcorpora and 175.8 million sentences. NT still comprised fifty-nine corpora, but 55.4 million sentences, and SMT sixty-nine 69 corpora and 862 million sentences.
4.1 Data collection and preparation of data
We excerpted 1,000 random instances of each phrase with more than 1,000 instances from each subcorpus, and all instances of phrases with less than 1,000 instances in the subcorpus. These processes resulted in nine subsets of och Norrland, och Svealand, and och Götaland (henceforth och region subsets) comprising a total of 5,818 instances, and nine subsets of Norrland och, Svealand och, and Götaland och (region och subsets) comprising 9,000 instances. The frequency differences reflect the fact that there were less than 1,000 instances in the och Svealand and och Götaland subsets.
Next, we removed duplicates, near-duplicates (i.e. nearly identical instances from the same source text, e.g. SVT NYHETER 2014), and instances that did not make sense graphically or content-wise (e.g. ‘10.8 14.0 16.3 13.4 20.3 … och Norrland’ [GT och Norrland]) from the subsets. Furthermore, instances with syntactic forms that do not involve coordination of phrases within a clause (i.e. with any of the nouns Norrland, Svealand, or Götaland) were excluded. They are instances where och (‘and’) coordinates two or more main clauses. One example is (4), in which two main clauses are coordinated with the main clause of Norrland har enorma tillgångar (‘Norrland has enormous resources’).
Taken together, 14% of the och region subsets and 20% of the region och subsets were excluded in the preparation of the data described above.
4.2 Analysis of data
Next, we identified the geographical categories coordinated with Norrland, Svealand, or Götaland in the remaining instances by analyzing the heads of the phrases coordinated with each of the och Norrland, och Svealand, och Götaland, Norrland och, Svealand och, and Götaland och instances. The analyses were done individually by each researcher followed by meetings in which the analyses were compared and discrepancies discussed.
Some of the och Norrland (22%), och Svealand (3%), and och Götaland (11%) instances were coordinated with phrases for several geographical locations. Because they were occasionally very complex, but typically involved coordination with phrases for the same type of geographical location,Footnote 4 they were categorized in accordance with the first phrase.
4.2.1 Main categories
The following main categories were identified.
Continent: For instance, Afrika (‘Africa’), Sydamerika (‘South America’), and Europa (‘Europe’). In (5), Asia and Norrland are coordinated.
Country: For example, phrases such as Sverige (‘Sweden’) and riket (‘the country’) or landet (‘the country’) implying that Sweden is the country, but also phrases for other countries (e.g. Kanada ‘Canada’ or Turkiet ‘Turkey’) coordinated with Norrland, Svealand, or Götaland.
province: Phrases that designate the contemporary administrative units of län (‘county’) or the historical but still vital units of landskap (‘province’). They were sorted into one category partly because some names (e.g. Gotland) are used for both län and landskap, and partly because the relationship between län and landskap is geographically intricate. In some cases, a län is a smaller part of one or several landskap, for example Kronobergs län (the southwestern part of landskapet Småland) and Stockholms län (which is part of both landskapet Uppland and landskapet Södermanland geographically). In yet others, län consists of more than one landskap: Norrbottens län and Västerbottens län consist of the landskap Norrbotten and Västerbotten respectively, and the landskap of Lappland which is divided between the two.
Region: Norrland, Svealand, and Götaland.
Areas: Phrases for somewhat vaguely delimited areas such as Mälardalen (around the lake Mälaren in Svealand), Bergslagen (the elongated geographical area in Svealand traditionally associated with mining and metallurgic industry), and Norrlandskusten (‘the coastal area of Norrland’).
places: Cities, towns, and municipalities (e.g. Stockholm, Örebro, and Vetlanda).
Islands and lakes (e.g. Öland and Vänern).
Other geographical entities: Phrases for geographical categories that do not fall into any of the above geographical categories. They include phrases for islands, lakes, or seas outside of Sweden (e.g. Phuket, Medelhavet ‘the Mediterranean’), directions (e.g. söderut ‘south’, uppåt ‘up’), and areas such as skogsbygd (‘forest area’) and glesbygd (‘sparsely populated area’), which do not have an exact location but may be in, or spread over, one or more of the regions. This category also includes historical names of geographical areas in or outside contemporary Sweden.
4.2.2 Subcategories
Some of the Norrland, Svealand, and Götaland instances are coordinated with single nouns (e.g. Gotland and Vetlanda) and others with attributes modifying and hence delimiting the area denoted by the noun. These more delimited areas make up more specific geographical categories and were thus labeled as such. In (9), the description of Asia is narrowed down to South Asia and therefore categorized as continent-specific.
In the same way, the adjective södra (‘south’) specifies which parts of Sweden are denoted by södra Sverige (‘south of Sweden’) and the noun phrase östra delarna av (‘the eastern parts of’) which parts of the regions of Svealand and Götaland are denoted by the phrase östra delarna av Svealand och Götaland (‘the eastern parts of Svealand and Götaland’). Instances including the phrases södra Sverige and östra delarna av Svealand och Götaland are thus categorized as country-specific and region-specific, respectively. The categories of area-specific, province-specific, and place-specific are based on the same logic.
Table 1 summarizes the identified geographical categories sorted in a continuum from the largest to the smallest categories, with region in the middle.
Table 1. Continuum of geographical categories combined with Norrland, Svealand, or Götaland
4.2.3 Instances excluded from the study
In the analysis, two categories were identified that were not relevant for the focus and hence excluded from the study. First, because our focus was on coordination with Norrland, Svealand, or Götaland, and not with phrases for more delimited areas of these regions, we removed och Norrland, och Svealand, and och Götaland instances that do not denote the whole region. One example is (10) in which the attributive phrase östra (‘eastern’) modifies and hence delimits not only the area denoted by the coordinated noun (Götaland) but also, in an elliptic way, the Svealand instance retrieved from the subcorpus.
For the present study, we decided that attributive phrases denoting adjacent areas, such as västra Svealand och Götaland, modify both coordinated nouns, whereas those denoting areas further apart, e.g. södra Götaland and Norrland, do not. Instances such as these were found in the region och subset. They were the most common in the Svealand och instances (16%), the second most common in the Götaland och instances (8%), and the least common in the Norrland och instances (2%), suggesting that Svealand och and Götaland och instances are more prone to being preceded by an attributive phrase delimiting the area denoted by the noun than Norrland och.
Second, because our focus was on geographical categories, phrases for non-geographical categories (e.g. snö ‘snow’, vinter ‘winter’, sommar ‘summer’, fiske ‘fishing’, and bönder ‘farmers’) denoting weather, seasons, leisure activities, and people were also excluded. These instances were primarily found in the Norrland instances (207 instances), which may be compared with thirty-five Svealand instances and fourteen Götaland instances, suggesting that there are differences between Norrland, Svealand, and Götaland also regarding this type of data.
5. Results
Figures 1–3 show the proportions of phrases for geographical categories coordinated with the remaining och Norrland, och Svealand, and och Götaland instances in the subsets excerpted from GT, MT, and SMT; Figures 4–6 show those in the Norrland och, Svealand och, and Götaland och instances. The categories include nouns for the regions of Norrland, Svealand, and Götaland, phrases for areas within the regions (i.e. region-specific areas, provinces, places, place-specific areas, areas within regions, and lakes and islands in Sweden), phrases for larger geographical areas (i.e. continents, countries, and continent-specific and country-specific areas), and phrases for other areas (e.g. geographically vague areas in Sweden, seas, islands, and areas outside Sweden).
Figure 1. Och region in governmental texts.
Figure 2. Och region in news texts.
Figure 3. Och region in social media texts.
The proportions of phrases for geographical categories coordinated with the och region subsets (Figures 1–3) are very similar. In all three subsets, Svealand (93% in GT, 90% in NT and SMT) and Götaland (87% in GT, 89% in NT, 68% in SMT) are coordinated with phrases for the three main regions considerably more often than Norrland is (14% in GT, 28% in NT, 15% in SMT) (RQ1). Moreover, Norrland is coordinated with phrases for areas within regions considerably more often (67% in GT, 56% in NT, 66% in SMT) than Svealand (7% in GT, 9% in GT and SMT) and Götaland (6% in GT, 12% in NT, 7% in SMT) (RQ2). Furthermore, the results show that Norrland (14% in GT, 13% in NT and SMT) is coordinated with phrases for areas that are larger than regions more frequently compared to Svealand (0% in GT, 1% in NT and SMT) and Götaland (1% in GT and NT, 2% in SMT).
The proportions of phrases for geographical categories in the region och subsets (Figures 4–6) similarly show that Svealand (37% in GT and NT, 68% in SMT) and Götaland (68% in GT, 51% in NT, 76% in SMT) are coordinated with phrases for the regions more often than Norrland (2% in GT, 15% in NT, 6% in SMT) (RQ1). Moreover, Norrland is coordinated with phrases for areas within regions more often than Svealand in SMT (66% vs. 23%) and with Götaland in all three subsets (51% vs. 26% in GT, 53% vs. 45% in NT, 66% vs. 20% in SMT) (RQ2). However, the proportion of Svealand och instances in GT and NT (57% in both subsets) coordinated with phrases for areas within regions is higher than both the Norrland och instances (51% in GT, 53% in NT) and Götaland och instances (25% in GT, 45% in NT). Although a very small proportion of Norrland och instances (2% in GT, 15% in NT, 6% in SMT) are coordinated with phrases for regions and primarily with phrases for areas within regions, coordination with Norrland och involves a large proportion of instances coordinated with phrases for other categories apart from these two (47% in GT, 32% in NT, 28% in SMT). Moreover, coordination with Svealand och in GT and NT is substantially different from coordination with och Svealand and Svealand och in SMT.
Figure 4. Region och in governmental texts.
Figure 5. Region och in news texts.
Figure 6. Region och in social media texts.
Taken together, the proportions of phrases in Figures 1–6 confirm the hypothesis that Svealand and Götaland are more often coordinated with nouns for the three main regions than Norrland, and those in four out of six subsets confirm the hypothesis that Norrland is coordinated with phrases for geographical places within the regions of Sweden more frequently compared with Svealand and Götaland. The results highlight asymmetries in how Norrland, Svealand, and Götaland and areas within these regions are categorized, but also show that Svealand is commonly followed by phrases for areas within regions in GT and NT.
Sections 5.1–5.4 discuss the usage patterns of coordination with phrases for regions, areas within regions, areas larger than regions, and other areas in the och region and region och subsets.
5.1 Coordination with phrases for regions
Not only do the proportions of phrases for regions coordinated with Norrland, Svealand, or Götaland display differences between the regions, but so do the types of phrases coordinated with each of the regions and the order in which they occur.
Norrland is typically coordinated with Svealand, with Svealand commonly being the first of those coordinated, as in example (11).
The proportion of Norrland instances coordinated with Svealand is the largest in NT (83% of both the och Norrland and Norrland och instances), somewhat smaller in GT (53% of och Norrland instances and 56% of Norrland och instances), and the smallest in SMT (35% of och Norrland and 38% of Norrland och instances). The och Norrland instances in all three subsets are coordinated with Götaland (12% in GT, 10% in NT, 5% in SMT) and those in SMT (13%) with Norrland. All these och Norrland instances coordinated with Norrland emphasize differences between two different referents of Norrland. As in (12), which reflects opposition to labeling Norrland as one homogenous unit, they reflect attitudes at odds with the asymmetry in categorization reflected in sentences (1)–(3) and the usage patterns of Norrland, Svealand, and Götaland observed in the present study (see Figures 1–6).
A relatively large proportion of the Norrland instances in GT and SMT are coordinated with both Götaland and Svealand in the order Götaland, Svealand, and Norrland (26% of instances in GT, 7% in NT, 38% in SMT), and the och Norrland instances in GT and SMT in the order Svealand, Götaland, and Norrland (9% in GT, 1% and 9% in SMT).
Most och Svealand instances (86% of instances in GT, 93% in NT, 89% in SMT) are coordinated with Götaland, with Götaland commonly being the first of the coordinated nouns as in (13).
Considerably fewer och Svealand instances (13% in GT, 7% in NT, 10% in SMT) are coordinated with Norrland or with combinations of all three regions.
The Svealand och instances are also commonly coordinated with Götaland (58% in GT, 59% in NT, 56% in SMT), but also with Norrland (33% in GT, 31% in NT, 21% in SMT) or with both Norrland and Götaland (6% of instances in GT, 8% in NT, 9% in SMT are part of the combination Norrland, Svealand, and Götaland, and 3% of them in GT, 2% in NT, and 14% in SMT are part of the combination Götaland, Svealand, and Norrland).
Götaland is typically coordinated with Svealand (84% of och Götaland instances in GT, 95% in NT, 88% in SMT, and 99% of Götaland och instances in GT, 92% in NT, 87% in SMT), with Götaland commonly being the first of the coordinated words.
To sum up, Norrland, Svealand, and Götaland differ not only in their tendencies to be coordinated with phrases for regions but also in terms of what regions they are coordinated with and in what order they are mentioned. Typically, the southernmost region is mentioned first, the northernmost last, and Götaland and Svealand coordinated.
Some of the differences observed are related to genre. Most examples of region and region coordination in NT concern weather reports. Because Sweden is an elongated country, the weather is usually reported in, on the one hand, the unit of Götaland and Svealand (the southern half of the country) and, on the other, in the unit Norrland (the northern half). Similarly, Norrland can be discerned as an economic unit, an agricultural unit, or a natural type in many examples in GT. The examples from NT and GT reflect Norrland as an established unit from many perspectives compared with other established units. Concerning at least weather, agriculture, and nature types, it makes sense to treat Svealand and Götaland as one unit rather than Norrland and Götaland, which is reflected in the many examples of Svealand and Götaland, the few examples of coordination between Norrland and Götaland, and the somewhat more frequent examples of coordination between the neighboring regions Norrland and Svealand.
5.2 Coordination with phrases for areas within regions
Svealand and Götaland instances are commonly coordinated with phrases for region-specific areas (e.g. södra Norrland ‘the south of Norrland’). They comprise 22/28 och Svealand instances in GT (78%), 52/66 instances in NT (79%), and 14/40 (35%) in SMT, and 216/265 Svealand och instances in GT (82%), 222/268 in NT (83%), and 82/134 in SMT (61%). The och Svealand instances are coordinated with phrases for parts of Götaland, and the Svealand och instances with phrases for parts of Norrland or Götaland. In (14), Svealand is coordinated with a phrase for southern Norrland.
The och Götaland instances coordinated with phrases for region-specific areas comprise 12/18 instances in GT (67%), 21/30 instances in NT (70%), and 9/37 (24%) in SMT, and the Götaland och instances 18/30 of instances in GT (60%), 25/35 in NT (71%), and 19/36 in SMT (53%). All these instances but one (coordinated with a phrase for a more specific part of Götaland) are coordinated with phrases for parts of Svealand. They are typically from, or related to, weather forecasts and involve topics such as temperatures, cloudiness, precipitation, wind, seasons, traffic situations, water yield, and wildfires in the areas, but there are also topics about forestry, employment, electricity prices in the areas, or the spread of animals such as wild boars, mites, or spruce bark within them. One example is (15), which deals with an expected increase in the proportion of spruce forests in Götaland and the southern part of Svealand.
Considerably fewer of the Svealand and Götaland instances coordinated with phrases for areas within regions involve phrases for provinces (e.g. Skåne), province-specific categories (e.g. Gävleborgs kustland ‘the Gävleborg coast’), vague areas (e.g. Norrlandskusten ‘the Norrland coast’), or places (e.g. Norrköping). The most common combinations include Götaland instances in all three subsets (3/18 och Götaland instances in GT [17%], 6/30 in NT [20%], 20/37 in SMT [54%], and 5/30 Götaland och instances in GT [17%], 9/35 in NT [26%], 12/36 in SMT [33%]) and och Svealand instances in SMT (17/40 [(42%]) and NT (10/66 [15%]), and Svealand och instances in SMT (25/134 [19%]). The Svealand instances are primarily coordinated with provinces in Götaland or Norrland (13/17 instances), but the Götaland instances with provinces within Götaland (15/20 of och Götaland instances in SMT, 4/6 in NT, and 8/12 Götaland och instances in SMT) and particularly with Skåne, Gotland, or Småland.
The Götaland instances specify the location of the province within this region (i.e. describe a part–whole relation) rather than compare Götaland to a province denoted by the coordinated word.
Relatively many Götaland instances in GT are coordinated with phrases for vaguely delimited areas within regions (e.g. Mälardalen in Svealand). They make up 3/18 of och Götaland instances (17%) and 7/30 Götaland och (23%) instances. Less than 14% of Svealand instances and Götaland instances are coordinated with phrases for area-specific categories.
Unlike the Svealand and Götaland instances, the Norrland instances are rarely coordinated with phrases for region-specific instances. They comprise 86/517 och Norrland instances in GT (17%), 24/183 instances in NT (13%), and 4/473 (1%) in SMT, and 19/219 Norrland och instances in GT (9%), 40/164 instances in NT (24%), and 5/89 in SMT (6%). Instead, most Norrland instances are coordinated with phrases for provinces, and particularly with the neighboring province of Dalarna in northern Svealand, the island of Gotland, and the southernmost province of Skåne showing that the asymmetry reflected in sentence (3) is common. Instances coordinated with provinces include 286/517 och Norrland instances in GT (55%), 74/183 in NT (40%), and 267/473 in SMT (56%), and 122/219 Norrland och instances in GT (56%), 60/164 in NT (37%), and 40/89 in SMT (45%).
Moreover, the och Norrland instances in NT (66/183 instances [37%]) and SMT (162/473 [34%]) and the Norrland och instances in NT (34/164 instances [21%]) and SMT (33/89 [37%]) are commonly used with phrases for places such as Malmö in sentence (1), Göteborg, Sundsvall, Falun-Mora in sentence (2), and Stockholm and Göteborg in sentence (3). This type of coordination is unusual in instances coordinated with Svealand or Götaland (0–14% of instances), further highlighting the asymmetry in categorization reflected in sentences (1)–(2).
Norrland instances in GT are commonly used with phrases for more vaguely defined areas within regions, e.g. Bergslagen and Mälardalen in Svealand and Västkusten (the Swedish ‘West Coast’) in Götaland. They make up 90/517 of och Norrland instances (17%) and 43/219 Norrland och instances (20%) in this subset. Coordination with these phrases typically builds on similarity and properties shared with Norrland. For instance, Norrland and Bergslagen share properties related to social issues, the economy, and natural resources.
A small proportion of the rest of Norrland instances coordinated with phrases for places within regions denote areas (≤ 12%), places (≤ 9%), or province-specific (≤ 5%), area-specific (≤ 1%), or place-specific (≤ 3%) categories.
In conclusion, Svealand and Götaland are more commonly coordinated with region-specific phrases than Norrland. Instances such as these resemble region and region coordination in the sense that the region is involved. The noun for the region is part of the coordinated phrase and its function is to provide information on which part of the region is relevant. Coordination with Norrland is more heterogeneous. It primarily involves phrases for provinces but also phrases for areas, places, or more specific parts of them. The differences align with the previously observed asymmetry in categorization involving Norrland, on the one hand, and Svealand and Götaland, on the one hand.
5.3 Coordination with phrases for areas larger than regions
Most instances coordinated with phrases for areas larger than regions include country-specific areas. They include both och Svealand and och Götaland instances in NT and Svealand och and Götaland och instances in GT, as well as all six Svealand och instances and all three Götaland och instances in NT. The Svealand and Götaland instances are primarily coordinated with phrases for the south of Sweden (e.g. sydsverige and södra Sverige), sometimes including both these regions in their references and sometimes suggesting that the south of Sweden is south of Svealand. In (17), one of the seasons is coming to Svealand and to parts of southern Sweden.
The Norrland instances are primarily coordinated with phrases for mid-Sweden (e.g. Mellansverige and Mellersta Sverige), indicating that this part of the country is south of Norrland.
Considering that Sweden’s geographical center is in the south of Norrland, instances such as these suggest that the north, south, and middle of Sweden are defined by population distribution (its demographic center) rather than geography.
Most och Götaland and Svealand och instances in SMT (3/4 instances of each subset) are also primarily coordinated with country-specific areas, but the rest of the instances are coordinated with phrases for countries. The latter group includes och Norrland (15/28 instances [54%]), Norrland och (59/95 [62%]), and och Svealand instances (9/12 [75%]). The relatively large proportion of Norrland instances coordinated with countries align with the somewhat sweeping categorizations of Norrland reflected in instances (1)–(3). Sentence (19) shows that asymmetry in categorization may be grounded in the identity of speakers.
5.4 Coordination with phrases about other geographical units
Instances combined with phrases for directions (e.g. norrut ‘north’, uppåt ‘up’, söderut ‘south’) are primarily found in the Svealand och subsets and about the weather or related phenomena in all three subsets. Most of them (12/17 instances in GT, 18/22 in NT, 28/35 in SMT) refer to a northward direction as in (20), again highlighting less specific and more sweeping information on areas further north.
Instances coordinated with phrases for vague areas such as glesbygd (‘sparsely populated area’) landsbygd (‘rural area’), skogslän (‘forest county’), inlandet (‘the inland area’) tend to be found in the Norrland instances in GT (30/30 och Norrland instances for other geographical areas in this subset, and 103/110 Norrland och instances), the och Norrland instances in SMT (36/38 instances) and the Norrland och instances in NT (13/20 instances). This type of coordination typically highlights properties that the areas (are perceived to) have in common with Norrland.
6. Concluding discussion
Results show that the Norrland, Svealand, and Götaland regions are far from treated as three equal entities of the same category. Svealand and Götaland are more commonly coordinated with nouns for regions than Norrland in all subsets (RQ1), and Norrland is more commonly coordinated with phrases for geographical places within the regions of Sweden compared with Svealand and Götaland in most subsets (RQ2). This suggests that the asymmetry in categorization reflected in sentences (1)–(3) is widespread in all three types of discourse investigated here, ranging from texts in the most private and presumably least (politically) correct discourse SMT to the public discourse NT to official GT.
We may conclude that categorization strategies involving Norrland are notably different from those involving Svealand and Götaland. Not only are towns, provinces, and other areas within Norrland labeled as Norrland but Norrland is also compared with even larger geographical categories comprising countries and continents more commonly than Svealand and Götaland.
The results align with those obtained by Eriksson (Reference Eriksson2010) in her analysis of representations of Norrland in contemporary news texts. Representations of Norrland tend to come with less detail than those of other regions. At the same time, they are accompanied by certain specific connotations. The patterns are seen not only in SMT but also in GT and NT. Considering the similarities in the proportions of phrases for regions, areas within regions, areas that are larger than regions, and other geographical categories in all three och region subsets (Figures 1–3) and SMT in the region och subsets, it appears safe to say that they are highly representative of coordination with Norrland, Svealand, or Götaland. Moreover, considering the similarities between the subsets, they do not appear due to how the subcorpora were compiled.
However, the region och subsets in GT and NT (Figures 5 and 6) are partly different due to the large proportions of region-specific coordination with Svealand in official contexts about the weather or the spread of animals and vegetation in parts of other regions. Presumably, the high frequency of the Svealand instances coordinated with these phrases is related to the topics they address and the size and location of this region. Given that most instances deal with the weather and the spread of animals or vegetation, which may be similar in adjacent regions, there may be numerous reasons for thinking and talking about Svealand and parts of adjacent regions this way. Still, the Svealand and Götaland instances coordinated with phrases for areas within regions are qualitatively different from those coordinated with Norrland. The Svealand och instances include phrases for areas within the other two regions (e.g. södra Norrland ‘south Norrland’), the Götaland instances include phrases for region-specific areas within Götaland specifying a part of this very region, and the Norrland instances include phrases for provinces, areas, or places in Norrland, again accentuating the asymmetry in categorization strategies. As in (1), where Norrland is used for the place from which a representative comes, these Norrland instances commonly reflect sorting places within Norrland into the considerably larger category of Norrland. In conclusion, categorization strategies involving Norrland are notably different from those involving Svealand and Götaland. Not only are towns, provinces, and other areas within Norrland labeled as Norrland, but Norrland is also compared with even larger geographical categories comprising countries and continents more commonly than Svealand and Götaland. In contrast, the Svealand instances coordinated with phrases for region-specific areas do not necessarily result from asymmetric categorization. Given that the noun for the other two regions is typically part of the coordinated phrase, speakers appear to have had the regions in mind when using these phrases but then qualified what specific parts of them are relevant.
There may be several reasons for the differences in categorization reflected in the present study. Some might be related to Sweden’s geography, population density, and categorization processes grounded in speakers’ body–world knowledge (Lakoff Reference Lakoff1987). Considering the length and elongated shape of Sweden and the fact that most Swedes live in either Svealand or Götaland, a less detailed categorization of geographical entities within Norrland might be close at hand for most speakers of Swedish even though Norrland, Svealand, and Götaland are all regions. Put differently, they might be because areas in the north are further away and hence less familiar to most speakers of Swedish than areas in the south. The relevance of the identity of the speakers is similarly suggested by the order in which the regions and their parts are mentioned – for example from the south to the north or in connection with region-specific coordination from Svealand in the middle to part of Götaland in the south or Norrland in the north.
Yet other differences are related to genre. For instance, there were differences between subsets related to the fact that different genres serve distinct functions, have different target audiences, and focus on different perspectives of the regions, and possibly also to differences between the denotations of Norrland, Svealand, and Götaland in the subcorpora. In GT, Norrland is either used for the five counties of Norrland, the forest-rich counties in Sweden (including some counties in Svealand), or an area used in international comparisons (national region). No matter if each of these official uses is precise, the fact that the areas denoted by Norrland vary in a standardized way in official contexts such as these contributes to the conceptualization of Norrland as a unit with low specificity/high degree of generality.
In contrast, instances in SMT carry a distinct personal flavor. They are often connected with local issues, and here the concept of Norrland is more intricately connected with the provinces of Norrland. In SMT, Norrland appears synonymous with the north, as shown by example (22), taken from a context discussing people’s understandings of which landscape fits which region. Here, the subjective perspective is highlighted.
Genre differences such as these may thus have contributed to the differences between the subsets.
The asymmetries observed in this data have implications. Given that category labels induce focus on commonalities (Althaus & Plunkett Reference Althaus and Plunkett2016), uses of the noun for the region Norrland for its parts (e.g. its towns and provinces) highlight commonalities between the parts denoted (no matter what they are like, they are all like ‘Norrland’) and hide attributes specific to each more specific place. As observed by Eriksson (Reference Eriksson2010:57), they involve using more general aspects of the region for specific counties, villages, and towns, whereby what is universal represents what is specific, and what is specific to these places is overlooked. In that way, uses such as these may reinforce stereotypes (everything in Norrland is one way, whereas the parts of Svealand and Götaland differ in substantial ways).
The present study was based on a mix of quantitative and qualitative analysis of no less than 18 subsets, each comprising up to 1,000 Norrland, Svealand, or Götaland instances. Still, it only scratches the surface of how these nouns are used. Qualitative semantic analyses of this data type are potentially fruitful complements to the present study. Considering the subjective quality of SMT, in-depth investigations of SMT might be particularly fruitful. Furthermore, semantic analyses of the phrases for the non-geographical categories (e.g. snö and Norrland ‘snow and Norrland’) removed from the study potentially show what semantic fields are associated with the three regions and what connotations the nouns for the regions have. Second, although the results of this study are representative of coordination with Norrland, Svealand, or Götaland in GT, NT, and SMT, there is more to learn about the regions from other constructions, including these nouns (e.g. those with prepositions signaling movement towards these areas, or with adverbs of place).
Finally, it may be fruitful to investigate how the asymmetric categorization of place observed here relates to the attitudes of speakers and how it is perceived among those living in places referred to as a region, and the region, in turn, portrayed as a homogenous unit with low specificity/high degree of generality. Recent debates in social and established media show that it has brought about a great deal of frustration that may only sometimes be understood by those categorizing the Swedish regions in asymmetric ways. Yet another question concerns whether the asymmetry observed here is a Swedish phenomenon or whether it may be observed in language about areas in other Nordic countries and the rest of the world.
Acknowledgments
We express our gratitude to three anonymous reviewers for their constructive feedback, which has greatly improved the quality of this paper.
Open access
