The Immigration Issue and the vox Populi: Letters to the Editor in six Scandinavian Newspapers 1970–2016

ABSTRACT Letters to the editor are argued to be a staple of the public sphere and, by providing regular citizens with a platform to voice their concerns to a large audience, a central democratic function of a liberal press. However, the actual weight given by the press to citizens’ deliberation on issues and the contribution of these letters in terms of content is still little researched, and historical and comparative studies are largely absent. In this article we offer a case study of letters on the immigration issue in six Scandinavian newspapers (N = 1065). Charting their volume and content using content analysis and comparing them to the more elite-dominated columns and regular news items (N = 3264), we identify major historical and national variations in the salience of the issue and the weight given the vox populi in the press. Constructing a discursive space of immigration letters using multiple correspondence analysis and subsequent clustering into seven categories of letters, we identify a historical movement towards more problematizing and cultural discourse, strongest in the Danish newspapers. We also identify Sweden as a particularly interesting case, with fewer letters and less difference between letters and other newspaper content, for which we suggest some hypotheses.


Introduction
Migratory movements have over the last fifty years appeared as an increasingly important issue, politically and socio-culturally, all over the so-called Western world. An especially complex and multifaced phenomenon, it provides major dilemmas and challenges including the fundamental issues of who to admit and exclude, integration into the welfare state, and how to get along in times with different conceptions of the good (Benhabib 1996). This makes immigration a good test of how modern democracies are able to debate difficult public issues (Benson 2013;Gripsrud 2019). In public deliberation of such issues, not only the quality of arguments but also the range of voices and perspectives are important (Habermas 1962). In such debates, the liberal press is expected to play a particularly important role as a platform for diverse public debate and encourage the participation of citizens (McNair 2000).
Many studies have been done on the regular news coverage of immigration in Europe, which typically emphasize problems and negative consequences of immigration (for an overview, see Eberl et al. 2018). While the predominance of elite voices over the voices of immigrants and citizens is another common finding (e.g., Georgiou and Zaborowski 2017), little is still known about the place the press has provided for regular citizens' voices, and what these have brought to the deliberation on the immigration issue.
Here the institution of letters to the editor is central. As a place where citizens can directly voice their concerns on public issues, it has been argued to be "essential to the effective operation of the democratic system" (Hynds 1991, 124). At the same time, letters' real deliberative role is highly debated, not least due to concerns about the representativeness of the contributors and the amount of real debate taking place (Wahl-Jorgensen 2001).
In this article, we present a case study of over thousand letters to the editors on this issue in six Scandinavian newspapers from 1970 to 2016. Through studies of their changing volume, and content analysis of their framing and subjects, we chart -from the birth of modern immigration into these societies to the present situation -how six Scandinavian newspapers have carried out their democratic function by providing a platform for citizens' direct voices on the issue. Did the letters, both in their numbers and content generally reflect the general news coverage and the more elite-and expert-dominated columns, or did they contribute something different to the public discussion? Has this changed through this long period of immigration, and how has this varied in the three countries? By its comparative design and focus on the full history of modern immigration, the case of Scandinavia offers a rare view of the changing and varying roles of the institution of letters in the press for debating complex and thorny public issues.

Letters to the Editor and the Immigration Issue
The literature on letters to the editor has traditionally focused on their deliberative function and the representativeness of the views expressed there. Such letters have been argued to be a staple of the public sphere (Habermas 1962), a forum for public discussion of community concerns by ordinary citizens, "transcripts of the town square … a debating society that never adjourns" (Nader and Gold 1988, 54). The actual role of letters in such deliberation, however, has been much debated.
The first caveat is that the actual dialogue taking place in letters to the editor appears limited (Wahl-Jorgensen 1999), a trait which has also been found in analyses of comment sections in net newspapers. People present opinions but engage very little in deliberation (Andersen 2020). The second caveat regards the representativity of letters, both in the case of their authors and their content. Although Singletary (1976, 537) proposed that readers view letters to the editor as "effective, influential expressions of opinion", much research has cautioned against taking letters as a "public opinion thermometer" (Grey and Brown 1970;Kleis Nielsen 2010;Pounds 2006;Sigelman and Walkosz 1992). First, the people who write them appear to be demographically and politically unrepresentative of the population. In 1979, Singletary and Cowling (1979, 165) reviewed past research and found "the letter writer to be middle-aged or older, male, well educated, well occupied, conservative", and later studies suggest this is a very persistent bias (c.f. Grey and Brown 1970;Kleis Nielsen 2010;Reader, Stempel, and Daniel 2004;Cooper, Knotts, and Haspel 2009;McCluskey and Hmielowski 2011). Second, while all journalists are gatekeepers for what kind of events, subjects and voices appear in the news (White 1950), editors appear to play a particular crucial role in the case of letters, which might further contribute to this disparity. 1 In a series of studies, Wahl-Jorgensen found that editors tend to have a low opinion of writers of letters, viewing them as non-representative of the public and even "crazy" (2002a); that editors "privilege individual expression over the expression of activist groups" and "prefer the emotionally charged stories of individuals" (2001, 303); select the letters according to the rules of relevance, brevity, entertainment value and authority (2002b); and insist on a modicum of propriety from the writers (2004). Moreover, editors prioritize letters that are local in both origin and content (2002c). Other studies similarly highlight the active role of editors (see e.g., Gregory andHutchins 2004, Kleis Nielsen 2010). While most studies emphasize letters to the editor as a biased expression of public opinion, the literature is not conclusive. Some, like Sigelman and Walkosz (1992, 945), while recognizing that letters will often not be an accurate representation of public opinion, emphasize that they "can, under certain conditions, provide an accurate gauge of public thinking on controversial issues". Others, like Hayek, Mayrl, andRussman (2020, 1127) finds that "letters closely reflect the readers' political positions" (see e.g., Buell [1975] and Hill [1981] for similar conclusions). Their representativity, it seems, varies by the context and the issue at hand. Interestingly, letters might function as public opinion without accurately representing it. Readership research have found letters not only to "stimulate debate and hold the attention of readers" (Romanow et al. 1981, 57), and to be "widely read and highly valued" (Raeymaeckers 2005, 200), but also to be viewed by both citizens and politicians as generally reliable indicators of public opinion (Singletary 1976;Herbst 1998).
The often-heated debate on our subject -immigration -brings to our attention another side of letters, namely their role as a platform for venting frustrations and expressing disagreement and negative viewpoints (Lander 1972;Buell 1975;Wahl-Jorgensen 1999;Cooper, Knotts, and Haspel 2009;Kleis Nielsen 2010;Richardson and Franklin 2004;McCluskey and Hmielowski 2011). As argued by Grey and Brown (1970, 454), "[o]ne of the functions of the letters to the editor in a democratic society is that of catharsis. A letter column gives the irate, the antagonist, the displeased, a chance to speak out and to be heard". In the context of immigration, one should therefore expect that letters are more negative than other coverage. The controversies surrounding the issue might, however, also limit such tendencies. Fear of negative reactions might make some people more reluctant to voice their views in public, and debate editors less likely to publish their letters, if they feel such views are outside the sphere of legitimate controversy (Hallin 1986;Wahl-Jorgensen 2004).
Letters to the editor, like other content in newspapers, are complex outcomes of many levels and types of influences (Reese and Shoemaker 2016), some which might be contradictory. This, and the fact that most of the cited research on letters regards countries with very different media systems and other issues for debate and was done in time periods which overlap with only smaller parts of the five decades of our study, cause considerable uncertainty about the transferability of their results to our case. That many of these studies have been conducted before the digital transformation of the infrastructure of the public sphere following the popular breakthrough of the Internet (Rasmussen 2016) is also worth emphasizing, even if this period is only relevant for the final two of our five decades. 2 For many reasons, then, our study of the role of letters to the editor in the public debate on the immigration issue in Scandinavia must necessarily be very explorative.

The Immigration Issue in the Scandinavian Press
The influx of labor migrants and refugees into the Scandinavian states over the last five decades has put immigration high on the political agenda in all three countries (Brochmann and Hagelund 2012;Green-Pedersen and Krogstrup 2008;Gripsrud 2019). While modern immigration started very late in a European context, two developments turned immigration into a central political issue in Scandinavia. The first was international events that brought large inflows of refugees and labor migrants to the previous culturally homogenous region, such as political upheaval (Chile 1973;Iran 1979;Turkey 1980), wars and humanitarian crises (the Vietnam War 1955-75; the Yugoslav wars 1991-2001; the Middle East wars 2001-) and the expansion of the European Union (2004). The second was the increasing problematization of the lack of integration of immigrants into the workforce and the "cultural turn" in the immigration debate in the 2000s (Brochmann and Hagelund 2012). Issues and events such as violent attacks by militant Islamic groups (e.g., the shooting of the Norwegian publisher of Satanic Verses in 1989; 9/11 2001; the Muhammad cartoons controversy 3 ), oppression and attacks on women (e.g., the "honorary" murder of Fadime Şahindal in Sweden in 2002) focused attention on Islam and immigrants' attitudes and customs. The contentiousness of the latter part of the issue is also related to the universalist and generous character of the Scandinavian welfare states, founded on high work participation, egalitarian wage structures, a willingness to pay substantial taxes, and high trust in core institutions. If a large proportion of citizens fall out and fail to take part, this threatens both the financial basis and the legitimacy of the welfare state (Hagelund 2003).
We know quite a lot about the general news coverage of the immigration in Scandinavia, if mostly limited to studies of single countries and the last two decades (but see Eide, Alghasi, and Simonsen 2020;Eide and Simonsen 2007;Figenschou and Beyer 2014;Horsti 2008;Gripsrud 2018;Hagelund 2003;Madsen 2000;Togeby and Gaasholt 1995;Strömbäck, Andersson, and Nedlund 2017;Yilmaz 2016). A common finding is that media representations of immigrants tend to be focused on problematic and negative issues and that elite voices dominate. Longer trends suggest that immigration news has become more politicized and focused on immigrant culture. In this regard, newspaper coverage in Scandinavia appears to share central characteristics with other European countries (Eberl et al. 2018), although studies suggest a somewhat less negative and more humanitarian-oriented discourse here (Hovden, Mjelde, and Gripsrud 2018). The latter is in line with repeated findings that Scandinavian citizens have the most positive attitude towards immigration in Europe (Heath et al. 2016). There is some variation: immigration in the news, like in public opinion as measured in surveys and voting patterns, has been least problem-focused in the Swedish debate, and most so in Denmark, with Norway somewhere in the middle (Hovden and Mjelde 2019a;Mjelde and Hovden 2019).
Given that letters to the editor have earlier been found to closely relate to the ongoing debate in the newspapers and prioritize local issues (Gregory and Hutchins 2004; Kleis Nielsen 2010), we should expect the immigration letters to generally follow the ebb and flow (not only in volume, but also in their framing and subjects) of the regular press coverage on the issue, and exhibit some of the same national differences as described above. Regarding the letters' function as a steam valve for venting grievances about immigration, the expectations are that they will likely be more negative than other coverage, but as noted, this can be subdued by self-censorship and editorial selection, and this might vary by country. Anti-immigration parties have been increasingly electorally successful in all three countries since the 1970s (see e.g., Widfeldt 2018). Ravndal (2018) suggests that these have performed a "safety valve" function, releasing pressure from dissatisfied activists. This finding fits with the emergence in all three countries of alternative right-wing online-only media outlets, whose raison d´être is that mainstream media allegedly suppress the vox populi, particularly on the immigration issue (Holt 2020;Ihlebaek and Nygård 2021). This latter argument has been particularly common on the (far) right in Sweden, birthing the term of an "opinion corridor" to describe the limited range of socially acceptable political views on immigration (see e.g., Edwards 2019). If true, this might have contributed fewer (and less critical) letters mailed to the mainstream press and a lower acceptance of critical letters. Our data can provide no direct insight into such processes. But through charting the volume of letters, their framing, and subjects, and comparing them to regular news coverage and the more elite-dominated columns, we can suggest something about the role letters to the editors have played in the deliberation on the issue in Scandinavian immigration debate. Important here is the similarity of Denmark, Sweden and Norway on key systemic dimensions: they are small, unitary and culturally homogenous social democracies with similar media systems and welfare states (Knutsen 2017;Hallin andMancini 2004, Skogerbø et al. 2021), which has led to arguments that they present "as close to an experimental situation as is possible in social sciences" (Pettersen and Østby, 2013, 76). Together with a relatively similar history of immigration, Scandinavia is a good case for studying comparatively and over time how the institution of letters to the editors has handled a major public issue. We focus on three questions: RQ1: What prominence has been given in the Scandinavian press to letters to the editor on the immigration issue, both in total and compared to other forms of coverage? RQ2: How does the content of these letters, in terms of framing and subjects, vary by country and change over time?
RQ3: How did their content differ from columns and regular news coverage?

The Data
Our data are 1065 letters to the editor about immigration that were printed in six leading Scandinavian newspapers from 1970 to 2016: the broadsheets Aftenposten (NO), Dagens Nyheter (SE), and Jyllandsposten (DK), and the tabloids VG (NO), Aftonbladet (SE), and Ekstrabladet (DK). For comparison with the debate in other parts of the newspaper, we included news articles (N = 2639) and regular columns (N = 625), in total 4329 items. To create a representative sample of articles from each year, we collected all immigrationrelated items from every day during four constructed weeks per year. Trained national teams of assistants carried out two rounds of sampling based on a reading of the full newspaper, and later, coding. Following the initial sample of 29,000 newspaper texts on immigration, a stricter subsample was extracted, including only items where debate on the immigration issue was clearly present, and shorter items were removed. 4 Of the resulting sample, every second article was coded. The corpus of texts is not, as most other studies of the mediation of immigration, primarily concerned with the representation of immigrations in the press, but with immigration debate in forms of subjects, framing and arguments. For such reasons, for example, sports stories, where immigrants are often featured prominently as athletes, are rare in the corpus, but sometimes turn up because the story in some way includes some argument or frame related to the debate on immigration. Due to the complex nature of the immigration issue, there was no simple way to identify such stories aside from having coders meticulously read every possibly relevant story before deciding on its inclusion using the sampling guidelines. 5 The most central variables in our analysis are country, newspaper, year, genre, subjects, and framing, of which the three first are self-explanatory. From a more detailed initial coding, genres were categorized into (1) letters to the editor, (2) columns (editorials, opinion pieces and commentary in column format from the newspaper staff and from outside, and op-eds) 6 , and (3) news (regular news articles). Their identification was based on the placement in dedicated sections and typical genre traits like bylines and length. Subjects concern the content in the texts, and typically involve specific agents, social institutions, and parts of the debate. The list was developed iteratively through a reading of literature and test coding of a random and stratified sample of newspaper items. In the following analysis, twenty categories are used, some examples being the use of welfare services, crime, working life and illegal immigration. 7 An item could contain multiple subjects. Letters and news articles averaged two to three subjects, and columns three to four.
Framing concerns, in our study, how immigrants and the immigration issue are presented in the newspaper texts. Loosely inspired by Goffman (1974), who in turn was inspired by work by scholars like Gregory Bateson and Alfred Schütz, the framing tradition in news studies focuses on the interpretive, symbolic context in which news stories are discussed, and how this forms central organizing ideas for making sense of relevant events (Gamson and Modigliani 1989). Framing does, however, not have a single definition which is agreed upon by most media scholars. While some have lamented this fact and made attempts to unify this "shattered paradigm" (Entman 1993), others have seen this variation as necessary for fruitful research (D'Angelo 2002). In our work, we follow the general approach to framing and the specific frames used by Benson (2013) in his research on immigration news, where framing suggests answers to the question: "What kind of problem (or positive phenomenon) is being attributed to immigration or immigrants?" (Benson 2013, 6). Some subjects are highly correlated with specific framings (e.g., civil rights and victim frames, or crime and threat frames), but most subjects are more mixed in this regard. In our coding we, like Benson, identified three main frames (threat, hero, or victim) generally linked to more immigration-critical and -positive position-takings, the difference between immigrants as a threat and as victims being especially well-founded in literature (Mancini et al. 2021). These three frames were divided into ten subframes for important distinctions, e.g., seeing immigrants as victims of racism or humanitarian crises, and being a threat to public order or to the fiscal sustainability of the welfare state. Most of these subframes also follow Benson's study, but some were changed to better suit the Scandinavian corpus. 8 Conceptually, our framing analysis appears closest to what D'Angelo (2002) has called the "critical" paradigm of news framing research, with frames as "persistent patterns of cognition, interpretation, and presentation, of selection, emphasis, and exclusion" (Gitlin 1980), which promote particular problems, benefits and moral evaluations (Entman 1993). And as is common in this strand of framing research, we are interested in not only how frames are an outcome of the routine nature of news-making (Tuchman 1973), but also formed by hegemonic structures (Hall 1978). This said, frame building in letters, columns and in general news on immigration are obviously the outcome not just of different authors (citizens, elites and experts, journalists), but also different processes and influences. Consider, for example, the importance of source relations and news-gathering resources for framing in general news coverage, versus the importance of the selective role of the debate editor, and the short length and genre requirements for letters to the editor. For such reasons, a comparison of framing in different newspaper genres must be done cautiously.

The Rising Salience of the Immigration Issue
Our first research question regards the place newspapers have historically given to letters on the immigration issue. Figure 1 shows that the volume of newspaper coverage rose markedly from the mid-eighties, but peaked earlier and at a higher level in Sweden and Denmark. 9 This, together with the high number of letters to the editor in Denmark, supports arguments that immigration became a heated public issue much earlier in these two countries than in Norway, and that this happened more dramatically in Denmark (cf. Green-Pedersen and Krogstrup 2008;Brochmann and Hagelund 2012). The volume of letters in all countries generally follows the year-by-year ebb and flow of other news and debate (versus news r = .72, versus columns r = .77), supporting earlier findings in literature of this correlation, but also suggests some national variation. 10 The volume of news and letters are for example more strongly correlated in Denmark, probably reflecting a more perfervid and permeating debate in its public sphere.
Letters in Sweden form an interesting contrast to the other two countries, as they take up a much smaller proportion of the national corpus both in total and relative to columns. After the millennium, there have been two letters for every column article in Norway and Denmark, but in Sweden it has been the other way round. And while Norway saw an increase in both letters and columns towards the end of the period, the number of columns increased markedly more than letters in Sweden. This dominance of columns over letters in Sweden shows that direct opinions of elite voices have here been given much more room than regular citizens in the public deliberation in the mainstream press.

A Shifting Space of Immigration Discourse
Our second research question regards the main subjects and framings in the letters to the editors on the immigration issue in Scandinavia. How has this changed, and has the content of the letters been different in the three countries? A simple inspection of threat and victim frames (Figure 2) in the six newspapers shows that the letters in the Danish newspapers in the later decades more often contained some kind of threat frame, while victim framing has been persistently higher in the Swedish newspapers.
While no sampling of newspapers can be fully representative of the national newspaper discourse (not least because newspapers have evolved, quasi-organically, to cater to different audiences and electorates) 11 , the similarities of national trends in the countries' leading tabloid and broadsheet suggest that these newspapers are symptomatic of the more general differences of the national press discourses. Similar national parallels are found for many subjects. Racism has for example been a more common subject in letters in Sweden, and religion likewise in Denmark.
We are, however, reluctant to base our discussion of the letters on such counts of isolated characteristics, as they ignore the polysemy and interrelations inside the texts (Kracauer 1952). Despite their often short length, the majority of the letters concerning immigration have several subjects and one in four invoke several subframes. 12 For such reasons we will use the methodology of correspondence analysis (MCA) to construct a space of immigration letters (Figure 3). 13 The analysis establishes a specific profile for each letter to the editor based on the presence (or absence) of twenty subjects and ten frames, and through chi 2 -metrics define the letters as overall similar or different to each other, regardless of other associations (including year, country and publication). We then identify the major divides (the principal axes) in this space and changes in the letters over time. Second, using cluster analysis on this space, we make a typology of seven groups of letters which is then studied for changes over time in the three countries. The methodological approach in this way mixes traditional and modern approaches to content analysis. It tries to account for the totality of the text (if only its registered properties) rather than studying single, disconnected variables, while through the use of traditional manual coding alleviates some of the problems of identifying substantial subjects and framings in complex texts through purely lexical, semi-automatic methods (for a critique, see Brookes and McEnery 2018).
The analysis identifies two major oppositions which explains two-thirds of the substantial variation in the table. 14 The first axis (the vertical dimension in Figure 3) opposes letters dealing with political regulation of immigration and their integration in working life and the welfare state, with positive framing of immigrants as "good workers" and as humanitarian victims, against letters where the culture and religion of immigrants, multiculturalism and also the oppression of women are combined with negatively laden subjects (crime) and framing of immigrants as a threat to public order and social cohesion. The second axis (the horizontal dimension) opposes letters where the subject of racism and framing of immigrants as victims of racism and discrimination is central, to more threat-based discourses, both cultural (focus on religion and culture of immigrants, national identity, and the threat to social cohesion), the fiscal strain on the welfare state and public order. 15 From the newspapers' placement in the space of axis 1 and 2, we can first identify a general historical movement, shared by most newspapers, from letters concerned with immigrants' integration into working life and welfare services, towards issues of cultural integration and racism. Second, we see again how letters in the Danish newspapers have been more concerned with problematic consequences of immigration, and that this tendency has been both persistent and increasing vis-à-vis the other countries. Third, letters in tabloids and broadsheet newspapers in a country tend to have similar positions, suggesting again the importance of national events and news agendas for what is discussed here. To get a better idea of what kind of letters are opposed in this discursive space, we have grouped those similar in subjects and frames into seven clusters by the statistical similarity of their profiles (Tables 1 and 2). 16 (1) Racism, the first group of letters, is also the most complex, mixing accusations and rejections of racism and discussion on what constitutes racism. It is closely related to the second group, (2) Media coverage, which consists of letters that comment on how the media handle the immigration issue (e.g., providing or withholding the ethnicity of perpetrators, or protesting portrayals of issues and persons as racist or bigoted). The next group, (3) Admittance and eviction, is also quite specific, mainly dealing with the entrance and deportation of specific immigrants (individuals, families, or groups), and often with sympathetic framing. (4) Welfare burdens are letters that are mainly negative, typically concerned with welfare abuse or the financial strain placed on the welfare state by the number of immigrants arriving. (5) Good immigrants, in contrast, is largely positive to immigrants and immigration, emphasizing their positive contribution to society. Also mostly negative is the group (6) Troublesome religion, focusing on the lack of cultural integration by immigrants, the subordination of immigrant women in Muslim cultures, the reign of familial loyalty over the rules of the state, etc. A final negative group of letters, (7) Dangerous immigrants, thematizes immigrants as a threat to public order (e.g., crime, terrorism). Some excerpts from exemplary letters 17 for the clusters are given in Table 2.
Regarding the historical development and national differences in the distribution of the seven groups of letters (Table 3), the Swedish and Norwegian letters clearly appear most similar. Swedish letters, though, more often fall in categories that focus on the beneficial sides to immigration and less often on the problematic sides of religion than the Norwegian letters. While "racism" letters are numerous both in Sweden and Norway in the whole period, they are much less common in Denmark. Instead, the Danish letters much more oftenat least since the eightiesemphasize the problematic sides of immigration, such as the financial burden of immigration, the problematic sides of religion and immigrants as threats through participation in crime and terrorism. The patterns found are quite similar to those found in regular news coverage of the issue in the same newspapers in the period (Hovden 2020;Hovden and Mjelde 2019), which similarly identifies a rising concern with the integration of immigrants, especially in regard to their use of welfare services and the cultural integration, and that Denmark offers the most problematizing discourse, and Sweden the least. So not only by volume but also by their content, the letters to the editor appears closely connected Table 2. Excerpts from exemplary letters to the editors in each of the seven clusters.
-I am not a race hater! (#1 Racism: AP 4. July 88) I am not a racist, I am not a racist. Will you who repeatedly declare that many Norwegians are race haters try to understand this? We are not against foreignersquite the contrary! In the hearts of the Norwegian people there is a strong desire to help people in need. It is the injustice politicians do to their own people we oppose. It is when we see what immigrants receive, in contrast to what many Norwegians in need do not get, that we react (…) It is not their [the immigrants] fault! It is an unfair policy that is to blame. Some blame might also be placed on the press. (…) Norway should help people in need, it is our duty, but some of these people have been Norwegian citizens since birth. (…) Are only 'Swedish' murderers named? (#2 Media coverage: DN 6. Dec. 02) DN recently reported on the torture-like murder of a woman in Lindesberg (…) The readers almost never get to know anything about the perpetrator. (…) This time DN made an exception. It named the murderer (…). Can it be so simple as the murderer this time had a Swedish-sounding name? Or will DN from here on publish the names of all perpetrators, at least murderers? As a reader, I want to know which people commit violent crimes. (…) But I suspect that there will be no change. DN is wholly convinced that negative reporting about immigrants increase xenophobia. Let the Portuguese deserters stay (#3 Admittance/deportation: AB 5. July 71) Three Portuguese, the deserters Germano Barros, Jose Rokha and Maria Rohka, will be deported from Sweden to in the parliament pandering to voters want to force the government to grant residency to the Palestinian Church occupants! The "spontaneous" action finally succeededaided by the extremely skilled lobbyist effort by the terrorist-friendly, naïve politicians and the senior clergy in Copenhagen (…) Only when we can afford to treat our ill, elderly and mentally ill can we also solve other people's problemsand not a second before. Without them, no Norway (#5: Good immigrants: AP 6. July 09) Lots of people say bad things to immigrants, but how would Norway be without them? How many Norwegians do you see behind the counter at the vegetable shop? I have only seen immigrants, and they are mostly from the Middle East. And most of the housekeepers in private homes come from the Balkans and Poland. There are also lots of Indians and Chinese and probably others from other countries, who have a different educational background than we do. I think we should thank the immigrants for all the work they do! The veil hinders Danish contact (#6 Troublesome religion: JP, 4. April 05) [Past Sunday] I spoke to a couple of sweet young Muslim women, who looked nice and dressed nicely like other young Danish women. Next to them was a "black-headed gull" (…) To me, these "black-headed gulls" signal little education, lack of intelligence and submissiveness, whether that is actually the case or not. (…) If Muslim women want to be treated liked other Danish women and taken seriously, they must get rid of the veil. Or they will never be integrated. It is not just a veil but to us Danes a provocative expression of an invasive and aggressive religion (…) Denmark is the Danes' country, and Danes decide and set the agenda. If immigrants are too unintelligent to understand this, they can get out of here. Politicians live in a fantasy world (#7 Dangerous immigrants: VG, 20. June 93) I read in VG this summer about an Iranian who attacked and hurt a Norwegian girl badly. The same Iranian has previously threatened to kill a Norwegian social worker. But he has nonetheless been allowed to stay in the country. The list of Norwegians who have been attacked and abused by immigrants is getting very long. Our politicians nonetheless ask themselves why xenophobia is increasing in this country. Do they really understand so little of the human mind? Or do they live in a fantasy world?
to the overarching and changing conversation in the press about how to handle immigration. There are however interesting national variations, which leads us to our final research question.

Letters, Columns, and News Articles: Differences in Framing and Subjects
We have so far considered the volume and content of letters, both in a historical and national comparative perspective. Our final question regards letterś specific contribution to the immigration debate in the press in Scandinavia. Did the letters emphasize other themes and subjects than the other parts of the newspapers' coverage? And are there national differences here? Regarding both subjects and framing, the debate genres -letters and columnsfluctuate more through these five decades than news articles do. This appears to speak to not only the routine nature of news gathering (Tuchman 1973), but also the cyclical nature of immigration itself: A new group appears, and their admittance, the finances, their integration and so forth must be handled by politicians and the civil service, again and again. But letters also differ from both news and columns (Figure 4 and Table 4). 18 As expected from the literature, threat frames are more likely to be found in letters than in columns and news articles, while the opposite is true for victim and hero frames. Sweden is a particularly interesting exception, with generally less difference in framing between the three genre categories. Table 3. Distribution of the seven clusters of immigration letters by country and decade. When it comes to subjects, racism, religion and multiculturalism are more common in the debate genres than regular news, echoing claims of an increasingly culturalized debate in Scandinavia (Yilmaz 2016). Letters, however, are typically less concerned with such subjects than columns are, suggesting a divide between popular and elite interest in this part of the immigration discourse. This divide is much smaller in Denmark, perhaps reflecting that such cultural subjects earlier became central in Danish mainstream political debate (ibid.). This divide also varies in its national emphasis: in Sweden, columnists are more often concerned than the vox populi with racism, Norwegian columnists more often with multiculturalism and Danish columnists are, unlike their Scandinavian neighbors, less, not more, often concerned with religion than the authors of the letters are. In Norway and Denmark the vox populi are also notably less concerned with integration than the columnists, which might speak to a more persistent popular view of immigrants as permanent "others", and if so, echoes findings of lower educated groups in these countries as much more negative to immigration than higher-educated groups (see e.g., Bohman 2018). Columnists are also more often concerned with politicians' and parties' handling of the immigration issue than both letters and regular news, emphasizing their elite-oriented coverage of the issue.

Discussion
The purpose of this study was to examine how the immigration issue has been expressed in letters in Scandinavian press since it appeared in its modern form in the 1970s. The analysis shows that letters very much bear the imprint of the national debate on immigration in each country. They tend to follow the ebb and flow of other content in the newspapers, and have, like regular news articles and columns, increasingly emphasized cultural Margins following logistic regression with control for sesquidecade (interaction between country and genre). Statistically significant differences between letters and the other genre categories are marked (*).
and problematic sides of immigration, especially since the mid-eighties. There are relatively small differences between letters in the leading tabloids and broadsheets in this regard. These national similarities, also across press genres, are likely due to several harmonizing mechanisms. Writers of letters are likely engaged by the same (mediated) national events and debates as journalists are, and journalists are more likely to publish letters relevant to the news agenda (Wahl-Jorgensen 2004). Without intimate knowledge of editorial practices and the authors of letters' motivations we are unable to shed light on the relative importance of such factors.
What we have showed is that the role of letters has been quite different in the press discourse on immigration in the three countries. Swedish letters much more often use victim frames than the two other countries' letters, and Danish letters much more often use threat frames. Norway, here, and in most other regards, appears as a middle case. The Swedish letters' lesser focus on the problems of immigration is not an unexpected finding, as it echoes national differences found in regular news coverage on the issue (see also Hovden and Mjelde 2019a), research on Scandinavian attitudes to immigration and Islam (Bohman 2018;Lundby et al. 2017), and in party-political cleavages, where the much earlier politicization of the immigration issue in Denmark appears particularly important to understanding the Scandinavian differences (Green-Pedersen and Krogstrup 2008). 19 The Swedish case is especially interesting. There are markedly fewer letters on the immigration issue in the Swedish newspapers than in the other two countries, not just in absolute numbers but also relative to the columns. The Swedish letters are also much more similar to other press coverage in their content and framing. Danish and Norwegian newspapers have, in contrast, devoted much more space to letters throughout the five decades, letters have here been more numerous (both in total numbers and relative vis-à-vis the columns), and they have emphasized the threats and problems of immigration more strongly, also in comparison with the columns. This support arguments that the debate in Sweden has been more dominated by experts and elites (c.f. Gripsrud 2019), and suggest that the letters in the Danish and Norwegian newspapers have contributed to a less inhibited public debate, where the popular classes' concerns (often negatively framed) of immigration have had a more natural place. But why so large national differences, given that we here are studying very similar countries, with similar press systems and with a quite similar history of immigration?
We can suggest four hypotheses. The repression hypothesis (popularized in Sweden as the "opinion corridor") is that this is an effect of Swedish newspaper editors having policed the debate more strictly, with anti-immigrant voices having been kept out of the debate to some extent, with the press thus acting as a cordon sanitaire for public debate on the issue. 20 There are, however, additional and alternative explanations. The vox publica hypothesis is that the difference simply reflects that Swedes are more tolerant of immigrants, and less disagreement here between elites and popular classes on the issue (c.f. Bohman 2018). The elite hypothesis is that letters, rather than being repressed or representative of the public mainly reflects the discourse of dominant agents, in particular politicians, which functions as "prime definers" and agenda-setters (c.f. Hall 1978). If so, the fact that Sweden not until 2010 had a significant far-right party that could oppose other parties in public debate might then be a good reason for both the low numbers of letters in Sweden and their low negativity. A final, public culture hypothesis would point to the longue durée of the shaping of the national public spheres, and to suggestions that Swedish public culture is generally more elite-and expert-oriented (c.f. Gripsrud 2019). 21 If so, our findings might reflect more fundamental differences in the way the national public spheres and media system works, and to a different role and importance of the institution of letters to the editor. If true, similar national patterns should be observable across completely different debated public issues.
Whatever the explanation for the national differences, the importance of letters as a vox publicaboth in regard to their weight, their relative volume and in their capacity to bring in other voices and perspectives to the deliberation of this thorny public issueappears greatest when the issue is most politicized, and public opinion most divided. In this way, letters to the editor appears to have played a very important historic role in the public debate of this issue in Scandinavia.

Notes
1. The op-ed section is perhaps the most exclusive forum. Ciofalo and Traverso (1994, 51) find that "professional journalists, public figures and propagandists dominate the pages, and that editors firmly control the agenda". 2. We will argue, however, that even given the changes of the mediated public sphere following digitalisation, letters to the editors are likely to continue to be important. While social media and comment sections in the newspapers have provided other opportunies for citizens to directly voice their concerns in public, letters to the editors central placement in newspapers makes them more likely to be read by a general audience, and, very importantly, much more likely to be read by politicians and other elite audiences. 3. For a discussion of this interesting case, see Eide, Kunelius, and Phillips (2008). 4. News/feature articles below 200 words and debate/opinion items below 50 words were removed. 5. For a more detailed discussion of the methodology, see Hovden and Mjelde (2019b). 6. While it would have been preferable to usa a more nuanced coding of columns, this proved difficult in the analysis due to their small numbers when split over countries and decades. 7. Subjects: Use of welfare programs, education, working life, integration policy, cultural and social customs, attitudes to immigrants, religion, national security, crime, public economy, immigration as a political issue, arrival and return of legal immigrants, illegal immigration and human trafficking, multiculturalism, national identity of Scandinavians, immigration debate, the role of media, oppression of women, civil and political rights, parallel societies. 8. Main frames and subframes: VICTIM: Global economy(*), humanitarian, war(+), racism and discrimination,. HERO: Cultural diversity(*), integration, good worker. THREAT: Jobs(*), public order, fiscal, social cohesion. + = not included in Bensons study. * = included in the MCA as supplementary categories due to few registrations. 9. The increase is much larger than what can be explained by changes in newspaper formats and layout. While the total number of articles in Norwegian press increased by 37 percent from 1983 to 2015, the number of articles on immigration increased by 367 percent. Source: Retriever. 10. Pearson correlation by year: Letters vs. news articles = Swe .40, Nor .49, Den .67. Letters vs. columns = Swe .56, Nor .84, Den .55. 11. As shown elsewhere (Hovden 2020), the Danish social liberal newspaper Politiken is more similar to the Norwegian and Swedish newspapers in coverage of the immigration issue. 12. 85% of the letters are coded for more than one of the subjects, 35% have at least four subjects. Only 12% invoke two or more of the three main frames, but 25% invoke at least two of the eleven subframes.
13. While the use of MCA for the analysis of texts can be traced back several decades (c.f. Lebart and Salem 1994), it is still rarely applied to analysis of news (for an early example, see Bastin and Bouchet-Valat 2014). For a short introduction the logic of MCA, see Le Roux and Rouanet (2010). 14. Eigenvalues of the first five axes and modified explained variance: .083 (37%), .077 (29%), .066 (16%), .056 (8%) and .050 (4%). The analysis is weighted by country and sesquicade to adjust for the different volume of letters over time and in three countries. 15. Later axes appear to mostly bring nuances to the oppositions already discussed. The third axis opposes letters emphasizing positive sides of immigratione.g. immigrants as being well integrated and good workersand discussing issues of multiculturalism and people's attitudes to immigrants, versus letters that emphasize immigrants as victims or threats to national and civic security. The fourth axis opposes letters concerned with the immigration debate more explicitly (where the theme of racism is again central) to more cultural themes, in particular the role of religion and the oppression of women. Note that for the calculation of the clusters, the first ten axes are used, not just the first two. 16. Hierarchical agglomerative clustering (Ward) of the letters on the first ten axes from the former MCA analysis. 17. This means that the letters are selected by their statistical centrality in the clusters. 18. See detailed tables in the online appendix, which also provides additional statistical information on the MCA and the cluster analysis. 19. Relevant here is also that of the three countries, Denmark is closest to the liberal media system, with fewer local newspapers, less political variation and higher ownership concentration (Skogerbø et al. 2021). 20. It cannot simply be an effect of the rise of alternative media in Sweden as an alternative vent for negative sentiments, as this Swedish difference is a persistent trait already from the eightiesdecades before alternative immigration-critical online blogs and news sites appeared. 21. If true, it appears somewhat puzzling that Swedish newspaper journalists in the Worlds of Journalism survey (Hanitzsch et al. 2019) are much more likely than their Scandinavian colleagues to say that to "Let people express their views" is important in their job (Swe 91%, Nor 61%, Den 55%). 22. LOWESS is a generalization of moving average regression to display smoothed general trends in a scatterplot.

Disclosure Statement
No potential conflict of interest was reported by the author(s).