The rise and fall of ‘propaganda’ as a positive concept: a digital reading of Swedish parliamentary records, 1867–2019

ABSTRACT Based on digital readings of all records from the Swedish parliament 1867–2019, we examine how the concept ‘propaganda’ was used in the debates. To track the concept, we have extracted word window co-occurrences, bigrams, and keywords. Research on the history of propaganda in liberal democracies has emphasized that the meaning of the concept was open-ended before WWI. By 1945, it had been contaminated by authoritarian propaganda, and its negative connotations were cemented at least by the 1960s. Our analysis, however, shows that ‘propaganda’ was used mainly in a negative sense from 1867 to 2019. Nevertheless, it was also possible to use ‘propaganda’ in a positive and neutral sense between the 1910s and 1980s. We suggest that a period of de-ideologization in Sweden post-WWII made it possible to use ‘propaganda’ as long as the issues were seen as non-controversial. The radicalization in the late-1960s meant that authorities and previously non-controversial issues became contested. To suggest one-directional ‘propaganda’ in order to implement what politicians had decided was in people’s best interest became difficult in this context. In this new communication setting, ‘information’ was a more flexible term in contexts where ‘propaganda’ had previously been used in a neutral or positive sense.


Introduction
In May 1946, only one year after the collapse of Joseph Goebbels' propaganda apparatus, the Swedish Social-Democrat Emil Olovson -debating improved traffic safety -took the speaker's podium in the Swedish Riksdag: 'The propaganda is still being carried out, and I suppose it will have to continue for some time to come, if a better condition is to be achieved at all'. 1 While there were different opinions about the methods needed to improve traffic safety, no one reacted against the idea that 'propaganda' was a key strategy.
In today's political discourse, 'propaganda' is loaded with negative connotations of lies and manipulation. In discussions concerning the state and the future of democracy, national security, and an open society, 'propaganda' is seen as a threat. 2 Research on the history of propaganda in Western and liberal democracies, including Sweden, has emphasized a conceptual change during the first half of the twentieth century. Before 1914, the meaning of the concept was more open-ended. By 1945, it had been contaminated by the horrors of authoritarian propaganda, and its negative connotations were completely cemented at least by the 1960s. 3 Nevertheless, this cannot be the whole story, at least not everywhere, because as late as 1992, a Swedish parliamentarian asked for 'propaganda work at schools, catering establishments, and fairs, to promote increased fish consumption'. 4 How could 'propaganda' be presented as a political solution in 1992 if authoritarian regimes -as research has argued -had contaminated the concept already in the 1940s?
In this article, we present a conceptual history of 'propaganda' in a particular political discourse: Swedish parliamentary debates. These debates constitute an interesting case since they represent an arena where opposing political actors stand against each other. In the parliamentary records, conflicts over political concepts can be followed over time, as well as the shared language and concepts that unite opposing sides. The analysis here is based on the minutes from the Swedish parliament (Kammarens protokoll), beginning with the establishment of the bicameral Riksdag in 1867 until 2019. We ask the following research questions: How was 'propaganda' used, and what did the concept refer to in the Swedish parliamentary debates between 1867 and 2019? How could 'propaganda' continue to be unproblematic well into the welfare state years? And what was it that eventually made it less useful as a neutral and positive concept?
The article explores these questions by drawing on previous research regarding the history of 'propaganda' in Western and liberal democracies, as well as from digital approaches to conceptual history. The analysis is divided into three parts. The first two present statistical results on the distribution of 'propaganda' in the minutes from the Swedish parliament. In the third part, we discuss and contextualize the empirical results and present possible explanations and interpretations.

Previous research on the history of 'propaganda'
There are mainly two, partly conflicting, historical accounts of how the concept of propaganda has changed over time. Both versions share the same starting point: when the Catholic church picked up the word in the early seventeenth century, in the context of 'the propagation of our faith to the entire world'. 5 During the following three hundred years, the concept spread to various social and political spheres. Here, however, the interpretations start to differ.
The first account usually emphasizes the open-ended meaning of the concept up until the early twentieth century. 6 Then, regional and global conflicts, in particular WWI, created a need to promote one's own positions and demoralize the enemy, something that later evoked scepticism against such campaigns. All this served to put 'propaganda' in a bad light, a conception that was cemented during WWII. Hence, according to this understanding, after the two world wars, 'propaganda' was heavily burdened with negative connotations. 7 In his history of the public relations industry and public information in Sweden, Larsåke Larsson reproduces this perspective. Larsson does not even treat propaganda and public communication efforts before the 1940s as part of the history of public relations, despite the many overlaps of concepts and practices. 'Propaganda' -in line with linear accounts of the concept's historical development -is treated as a separate concept and activity, something that public relations practitioners left behind as they became professional. 8 Nevertheless, as pointed out by Paul Jonathan Meller, not much empirical research has focused on the longitudinal development of the concept. At least in a British context, there seems to be an etymological gap of three centuries, between the early seventeenth century and the early twentieth century. 9 Meller also notes the influence of early propaganda and public relations theorists' writing in the 1920s. The American Edward Bernays' narrative in Propaganda (1928), for instance, where he explained that 'propaganda' was used in a neutral sense up until WWI, has often been reproduced by later historians. 10 Other researchers have presented a more complex analysis. Wolfgang Schieder and Christof Dipper have shown, based on a variety of pamphlets, newspapers, private correspondence, published speeches, and parliamentary records, that the modern meaning of 'propaganda' has its roots in the aftermath of the French revolution, as counterrevolutionary groups started to label revolutionary activities as 'propaganda', meaning something destructive and terrorizing. 11 In the nineteenth century, some radical groups used it to refer to their own communication, but often reluctantly. Overall, 'propaganda' was still used to dismiss deviant actors. 12 The real change came in the late nineteenth century when business interests started to use the concept in the context of advertising. This meant that some of the negative connotations were pushed to the background in favour of the technical and more neutral aspects of propaganda, referring to the methods of spreading messages. After this conceptual clean-up, it was possible to use 'propaganda' as a positive term in a wider political context. 13 After WWI, the negative connotations started to re-emerge, but it was still possible to use the concept in a positive sense. Soon, however, it became associated with the methods of authoritarian regimes. Schieder and Dipper conclude: 'The history of the concept of propaganda is therefore to be interpreted less in the sense of a continuous perversion of its original meaning, but rather as the unsuccessful attempt to underlay the term with a permanent positive meaning'. 14 Meller makes similar conclusions about the British case, based on newspaper discussions and governmental documents between 1854 and 1902. 'Propaganda' was used in a variety of social contexts, but most often in a negative way. 15 Interestingly, a more positive usage of 'propaganda' seemed to be possible despite WWI. During the interwar period, the advertising sector in countries like Germany, the UK, and the US tried to establish a kind of social propaganda concept, which gained temporary success. During the 1920s and 1930s, 'political and commercial advertising became understood as forms of propaganda and social engineering'. 16 Elin Gardeström -using various sources from, for example, parliamentary debates, the advertising industry, and the cooperative movement -has made a similar observation regarding Sweden in the 1930s. 17 During WWII and the following decades, 'propaganda' in a positive sense was still possible in countries like the UK and Sweden. 18 Hence, while the dominant historiographical narrative of 'propaganda' emphasizes a decreasing trend of using the concept in a neutral and positive way after WWI, other researchers have observed an opposite trend -also related to the discourse of parliamentary politics. In fact, as has been shown in this section of the article, some researchers have indicated that it was even possible to use 'propaganda' in this neutral way during the first decades after WWII.
Our article contributes to the historiography of 'propaganda' with an empirically grounded study of the concept over a long period of time, from 1867 to 2019, in one coherent political source: the records of the Swedish parliament. Previous research has mostly focused on more limited periods, but digital methods make it possible to study the meanings of the propaganda concept in a massive dataset both before WWI and after WWII.
Doing conceptual history in the digital age: theory, methods, and sources 'Propaganda' could be seen as a key (or basic) concept. In the tradition of conceptual history, key concepts are those that are both indispensable and contested in political discourse. They escape simple definitions because the struggle to define them is an important part of politics. 19 The project undertaken by Reinhart Koselleck and his colleagues was to track and register conceptual changes as the old world of the eighteenth century transformed into the (German-speaking) modernity of the nineteenth and twentieth century. New words became key concepts, old ones were replaced, and old concepts acquired new meanings. These conceptual changes, Koselleck argued, were fundamental to the formation of modernity. A new conceptual landscape meant that the horizon of experiences and expectations was transformed, that new identities became possible, and that a new field of political action emerged. Nevertheless, redefined concepts would also point backward and carry old meanings into modern settings. Significant to these processes was a democratization of language use, where a wider range of speakers applied concepts in new domains. This made it important to analyse the use of concepts in political and other contexts. 20 The conceptual analysis presented in Geschichtliche Grundbegriffe: Historisches Lexikon zur politisch-sozialen Sprache in Deutschland was the result of the joint efforts of a team of researchers manually tracing the history of individual concepts in a wide variety of sources. 21 Efforts have been made to utilize digital text databases and computational methods to continue -and perhaps challenge -what Koselleck started. The computational approaches are usually based on the analysis of word co-occurrences and the idea that you 'shall know a word by the company it keeps'. 22 Peter de Bolla and his team at the Cambridge Concept Lab have examined different word co-occurrence metrics in order to capture the way 'concepts are constructed through "constellations" or "bundles" of individual words'. 23 Michael Gavin et al. have proposed an alternative method based on vector semantics, but it is founded on similar principles: 'concepts are clusters of terms that co-occur throughout a corpus'. 24 The model makes it possible to distinguish between different clusters, indicating different contexts of use. In large-scale computational approaches, where hundreds of thousands of texts are processed, the specifics of the language used within distinct genres or social groups might be lost. What Gavin et al. suggest, however, is a 'conceptual history through computationally assisted close reading', where individual works are read against the patterns emerging from the large-scale analysis. 25 This is an approach that we embrace in this article.
The Swedish parliamentary records have been digitized by the Riksdag Library and are available at riksdagstryck.kb.se and data.riksdagen.se. The records contain both speeches and various notes of activities in the chambers (bicameral Riksdag, 1867-1970, and the unicameral Riksdag, 1971-), such as voting, indexes, and reports of attendance. In this article, we have used all the records from 1867 to 2019 -in total some 17,000 records of varying length, compiling a corpus of 504 million tokens. The quality of the optical character recognition is fairly good, and no additional cleaning has been conducted during the work for this article. For analytical purposes, an automatized part-of-speech tag process was performed on the corpora, which means that all tokens are assigned with a word class and a lemmatized root. 26 The computational methods used to track the propaganda concept are bigram extraction, word window co-occurrence, and keyword extraction. Bigrams were extracted with the tool AntConc to examine the words that modify 'propaganda' ('political propaganda', 'religious propaganda', et cetera). Words co-occurring with 'propaganda' were extracted within a window of ten words (for nouns) and one word (for adjectives) before and after our target word. Keywords were extracted with AntConc based on the keyness measure log-likelihood. Keyness in this context is a measure of the frequencies of words that cooccur with a target word ('propaganda') in a fixed window, compared to the frequencies of all words in the complete corpora. Words defined as keywords are those that are more likely to co-occur with the selected target word. Last, close reading was used along with the computational methods to contextualize and verify their output. The methods are combined in order to study how 'propaganda' was used and what the concept referred to. Statistical covariations do not prove causality, but one advantage of the chosen methods is that they are transparent and that statistical results can be traced back to the sources, making it possible to examine specific contexts by close reading. The statistical findings in the analysis are hence discussed and interpreted in relation to both concrete examples from the empirical material and to previous research on the political climate in postwar Sweden.

'Propaganda' as a negative label
From the establishment of the bicameral Riksdag in 1867 to the breakthrough of democracy in Sweden around 1920, the use of 'propaganda' became gradually more frequent. Figure 1 shows the normalized frequency of the lemmatized noun 'propaganda' in all records and displays how 'propaganda' slowly increases in the second half of the nineteenth century and decreases after its peak in the 1930s, only to return to low frequencies in recent decades. In total numbers, the lemma 'propaganda' occurs about 10,000 times in the entire corpus, but only 600 times up until the 1920s.
Divided into four periods, Table 1 presents lists of the top 10 most frequent bigrams that refer to specific kinds of propaganda -'Catholic propaganda', 'socialist propaganda', foreign propaganda', et cetera (more general bigrams such as 'through propaganda', and 'such propaganda' are excluded) -from 1867 to the 1930s. The result presented in Table 2, showing the most frequent bigrams from the 1940s to the 2010s, is very similar. Besides condemnations of 'anti-semitic propaganda' and 'racist propaganda', the concept was used to characterize communication from adversaries as manipulative ('political propaganda', 'party propaganda', 'Russian propaganda'), or to highlight the label 'propaganda' as it is used to repress legitimate communication ('homosexual propaganda' in a homophobic discourse).
It is difficult to quantify the positive and negative usages of 'propaganda' by digital means alone. Sentiments are embedded in political and linguistic contexts, and the analysis of word frequencies needs to be supplemented by a close reading of individual cases. To examine how 'propaganda' was used and what it referred to, we have performed a reductive digital reading of the total corpus, combined with a manual examination of more ambiguous bigrams and their contexts in the parliamentary records. Examples of such bigrams include 'political propaganda', 'religious propaganda', and 'party political propaganda'.
In the nineteenth century, 'propaganda' was usually of a religious kind. 'Catholic' and 'papist' propaganda were seen as a threat to the protestant faith and society in general. Some, like the priest Albert Sjöholm, affiliated with the Protectionist Party, feared the establishment of a Catholic party: 'What will happen to the state, if the Catholic propaganda is allowed to continue in its stealthy way, that is another question, for it could very easily happen that we too in our country could get a clerical party, which, as I believe, would not be a blessing'. 27 Similar, Frithiof Grafström, also a priest, feared an infiltration of Catholic priests in the Swedish protestant state church: 'A new Klosterlasse could then appear and spread papist propaganda unhindered within the Swedish Church'. 28 Religious propaganda was also associated with the notion of anti-military propaganda, something that was 'against everything called defence', according to the MP and military captain Ernst Liljedahl. 29 From the early twentieth century and onwards, propaganda became less of a religious matter and more associated with threatening political ideologies such as anarchism, socialism, and pacifism. One MP called for tougher laws to restrict the 'attack of fanatical anti-military propaganda'. 30 Anarchists represented another threat, leading to 'the most serious crimes': 'anarchist propaganda is extremely dangerous to society'. 31 If we focus specifically on co-occurring adjectives, we also find that 'propaganda' was associated with aggressive communication acts. Co-occurring adjectives modify 'propaganda' as 'impertinent' (fräck), 'powerful' (kraftfull), 'reckless' (hänsynslös), and 'violent' (våldsam). Hence, 'propaganda' was mainly used to characterize messages as threatening and unwanted. What Schieder and Dipper identified as a pattern in the early nineteenth century in a German context is seen in the Swedish parliament one hundred years later: 'propaganda' was primarily a label used to dismiss communication from groups seen as political adversaries or even a danger to society. 32  (1) Catholic propaganda (5) Anti-military propaganda (31) Political propaganda (66) Communist propaganda (1) Political propaganda (2) Political propaganda (16) Anti-religious propaganda (56) Christian propaganda (1) Religious propaganda (2) Propaganda of the deed (12) Revolutionary propaganda (51) Papist propaganda (1) Socialist propaganda (2) Religious propaganda (11) Religious propaganda (25) Anti-Christian propaganda (1) Anti-defence propaganda (8) Communist propaganda (19) Church propaganda (1) Socialist propaganda (6) Anti-state propaganda (13) Separatist propaganda (1) Anarchist propaganda (6) Anti-military propaganda (13) Anticlerical propaganda (3) Anti-Christian propaganda (9) Bolshevik propaganda (2) Anti-defence propaganda (9) Gender conscious propaganda (1) Pacifist propaganda (8) (41) Political propaganda (42) Foreign propaganda (25) Anti-semitic propaganda (31) Racist propaganda (17) Anti-semitic propaganda (28) Anti-semitic propaganda (23) Social-democrat propaganda (18) Anti-semitic propaganda (15) Racist propaganda (15) Party propaganda (23) Party political propaganda (17) Neo Nazi propaganda (8) Social-democrat propaganda (10) Race baiting propaganda (21) Bourgeois propaganda (12) Moderate propaganda (7) Party political propaganda (8) Communist propaganda (17) Party propaganda (12) Party political propaganda (6) State propaganda (7) Liberal party propaganda (13) Communist propaganda (11) Neo Nazi propaganda (6) Russian propaganda (6) Phm propaganda (12) Religious propaganda (7) Anti-Jewish propaganda (5) Religious propaganda (5) State propaganda (12) Group propaganda (7) Anti-democratic propaganda (5) Homosexual propaganda (4) Right-wing propaganda (11) Right-wing propaganda (7) Bourgeois propaganda (5) Putin's propaganda (3) Cases of propaganda used as a non-negative concept -highlighted by Clark, Auerbach and Castronovo, and others -were rare in the Swedish parliament before WWI. 33 One example where 'religious propaganda' was used in a positive sense is found in 1929. Ernst Klefbeck -a Social Democrat and a priest -explained that 'it has not occurred to me to say that I do not promote religious propaganda. I do that every single Sunday'. 34 However, this case was an exception.
Theoretically, it would be possible to characterize political communication in general as 'political propaganda', even when messages are seen as legitimate by the speakers. Statements expressing such views can be found in the protocols. Viktor Herou, representing the communist party, explained in 1925 that workers on strike made 'political propaganda for an idea that ignites'. To restrict such propaganda would not prevent 'noble citizens from fighting for what they believe is right'. 35 However, more common is that 'political propaganda' represents something false or misleading, such as one-sided claims or exaggerated statements. One example is when a Social-Democrat in 1984 explained that the moderate party 'spreads political propaganda and tries to scare young people with something that only exists in the imagination of moderates'. 36 These negative aspects of 'political propaganda' were often implicit rather than outspoken.
That the label signified something negative is evident from the many casesoccurring from the late 1800s to the 2000s -where MPs reacted to 'political propaganda' being spread in tax-funded facilities or institutions, or by state representatives who were expected to be neutral. Already in 1886, the priest Erik Jakob Ekman opposed those who refused subsidies to the worker institute in Stockholm because it was said to spread 'anti-Christian propaganda' and 'political propaganda'. 37 In another example, the communist representative Oscar Öhman stated in 1929 that 'conscripts are exposed to intensive political propaganda from commanding officers'. 38 A related issue was political propaganda in schools. Liberal MP Sven Bengtsson stated in 1928 that 'I believe that no religious or political propaganda should occur at our folk high schools'. 39 A speaker from the social democrats asked in 1967 about actions 'to prevent political propaganda from teachers in schools'. 40 Other usage of the label followed the same pattern. MPs were concerned about political propaganda at retirement homes, in public service broadcasting, at post offices, and in public theatres. 41 The political actors did not always agree on what to characterize as 'political propaganda', but still used the concept in much the same way. It was mainly used as a label to classify messages as inappropriate in contexts where neutrality was expected. Similar critique was often expressed in terms of 'party political propaganda'. For example, an MP from the Centre Party state in 1973 that 'party political propaganda in public governmental service facilities is common in oneparty states, but Sweden should be spared from it'. 42 To summarize, the overall conceptualization of 'propaganda', from 1867 to 2019, was negative, in various ways. This result is in line with previous research in conceptual history, at least related to the UK and Germany. The main function of 'propaganda' was and is to label communication as negative and unwanted. 43 The rise and fall of positive 'propaganda' While our results reveal that 'propaganda' was used mainly in a negative sense, positive and neutral examples can be found. One way to trace such usage is to examine 'propaganda' as a collocate to a more neutral index word. Here we have examined 'road traffic safety' as a collocate to 'propaganda' and two alternative keywords: 'upplysning', and 'information'. 'Upplysning' was used as a synonym for 'information', referring to details, facts, and messages. It also means enlightenment and education. Figure 2 shows that 'propaganda' and 'upplysning' dominated in the 1940s, 1950s, and 1960s. In the late 1960s and in the 1970s, 'information' started to co-occur more often, but it was not until the mid-1970s that the concept took over. Until then, the promotion of 'road traffic safety' was essentially a matter of 'propaganda' and 'upplysning', not 'information'. In fact, the last time 'propaganda' was used in this context was in the mid-1980s.
The promotion of road traffic safety was not the only issue that Swedish parliamentarians discussed in terms of 'propaganda' in a neutral or positive sense. To trace such issues, we examined keywords co-occurring with 'propaganda' in the corpus. Based on words more frequent within a word window around 'propaganda' (10 words before and after), compared to the frequencies of these same words in all protocols (for each decade), we manually identified instances where propaganda was suggested as a neutral or positive solution to the issues debated. Tables 3 and 4 show the manually identified issues. The lists of issues are not complete; what is shown are issues related to words more likely to be used together with 'propaganda', not all the words used in that context.
To mention 'propaganda' in non-negative terms was rare before the 1910s. Based on the method of keyword extraction, the only occurrence identified in the parliamentary debates from 1867 to 1909 shows up in a quote from a Norwegian report on governmental factory inspectors in Switzerland in 1889. These inspectors 'made propaganda for safety measures', which led to improved worker safety. 44 In the 1910s and 1920s, Figure 2. Raw frequencies for 'propaganda', 'upplysning', and 'information' as they co-occur in a window of 10 words before and after 'road traffic safety' (trafiksäkerhet) for each year. Raw frequencies are used due to the relatively low co-occurrence frequencies.
'propaganda' was used as a positive label on a few occasions, primarily related to food and health. It was in the 1930s that the number of topics started to increase. Food and health continued to be important issues requiring 'propaganda', but also various topics related to the economy and industrial and agricultural production, world peace, and history. These issues confirm Gardeström's observation that the concept of propaganda in Sweden became associated with political advertising for the good of society in the 1930s. 45 Table 4, showing topics debated from 1940 to 1989, indicates that similar issues remained matters of propaganda in the welfare state. However, we also see a greater Table 3. Lists of issues to which propaganda was suggested as a neutral or positive solution, from 1867 to 1939. 1867-1899 1900-1909 1910-1919 1920-1929 1930-1939 Table 4. Lists of issues to which 'propaganda' was suggested as a neutral or positive solution, from the 1940s to the 1980s. After the 1980s, no neutral or positive issues were found in relation to 'propaganda'. 1940-1949 1950-1959 1960-1969 1970-1979 1980-1989  variety of issues on the lists. Marriage, insurance, fire safety, children's literature and environmental protection -the list of issues to handle with 'propaganda' grew longer and more diverse after WWII. This was a period when politicians believed that 'propaganda for a better upbringing' would reduce juvenile delinquency, when 'propaganda' was suggested as a method to educate consumers about shoes, and when 'propaganda' targeting employers and employees was seen as a way to get former inmates reintegrated into the workforce. 46 Issues related to alcohol, narcotics and tobacco remained on the list for several decades. As Johan Edman has pointed out based on the Swedish parliamentary records, narcotics in particular, but also alcohol, have been topics characterized by strong political consensus: 'Members of Parliament (MP) from the far left to the far right, agreed that the drug problem was the most serious contemporary problem, that it was a culturally alien problem that required extraordinary solutions, and that it was a problem whose severity could not be questioned. The Swedish parliamentary material is full of these consensual foundation bolts'. 47 Such a consensus made 'propaganda' uncontroversial since most politicians agreed upon both the problem and the solution. The concept of 'propaganda' was used without hesitation, it seems, as long as it referred to a good cause, a non-controversial topic, or an issue many politicians agreed upon. The usefulness of the concept peaked in the 1960s, and it became less favourable in the following years. As shown in Table 4, in the 1980s, anti-drinking efforts and road traffic safety were the only areas left where 'propaganda' could be suggested as a solution. 'Propaganda' as a possible political solution became significant in the Swedish parliament during the interwar period and stayed significant long after 1945. This is noteworthy since much previous research on the concept regards WWI, and certainly WWII, as the turning points for when 'Propaganda' became impossible as a positive or neutral concept in liberal democracies. 48 The lists of issues in Table 4 hint at another turning point: the lists get shorter after the 1960s.

Discussion
The results of the computational analysis are limited by the corpus and the chosen methods. They indicate trends, but the methods do not provide us with explanations. In the following, we present some potential interpretations by situating the results in a political context in order to discuss how it could be that 'propaganda' continued to be unproblematic well into the welfare state years. We will also discuss potential explanations as to why the use of the concept in a neutral sense started to drop in the 1970s.
Compared to many other countries, Sweden had no great postwar debates about collaborations, guilt, and responsibilities related to WWII in the late 1940s and 1950s. There was growing support for welfare state reforms, and the cold-war climate created a national unity around peace and democratic values. Opposing voices did exist, but they were marginalized in the public debate. 49 A coalition between the Social Democrats and the Agrarian party ruled the country from 1951 to 1957. This was, however, not the first time Sweden had had a coalition government. During WWII, for example, all parties except the Communist Party were part of a coalition government. After the war, the economy was booming and there were no major conflicts on the labour market. 50 In previous research, the 1950s is characterized as a time of political harmony, compromises, and consensus. 51 A leading figure in the public debate, Herbert Tingsten, editor in chief at Dagens Nyheter, described already in 1951 what he saw as a 'de-ideologization' in the Swedish society. Political differences still existed, but the parties had an ability to cooperate without ideological differences getting in the way. 52 To use 'propaganda' to make people change their behaviour, according to an agenda most politicians agreed upon, might not have been seen as controversial in such a context. As pointed out by Kjellgren, the organization of the various communication efforts contributed to such a consensus. Interest groups, unions, professional organizations, et cetera, were often represented in the committees and boards responsible for the campaigns, facilitating consensus on an organizational level. 53 This state of political consensus started to change in the second half of the 1960s, when the personal became political and new issues were put on the political agenda. 54 In their account of the political impact of the radical movements of the late 1960s in Sweden, Bjereld and Demker emphasize the questioning of established norms and traditional authorities such as teachers, priests, and parents. Individual voices and choices mattered more than traditional values and power structures. 55 Similar trends, although embedded in different national contexts, were also visible in other countries at that time. 56 Hence, using 'propaganda' to implement what politicians and experts had decided was in people's best interest became more difficult in such a context, and the neoliberal turn in the 1980s strengthened this tendency. 57 How authorities should communicate with the citizens became a contested issue in itself. A governmental report on so-called public information from 1969, for instance, declared that informing the citizens was a one-directional process involving the spread of facts, rules, and recommendations from the state to the public in order to educate the citizens. 58 This perspective was questioned when the report was debated in public in the early 1970s. 59 One of the issues in the debate was the notion that the report did not take two-way communication seriously. The dominance of top-down communication was linked to various social problems and conflicts, which Åke W. Edfeldt, professor in pedagogy, articulated in an interview: 'The lack of communication is the core issue in all societies that have grown to see a widened gap between individuals and the authorities. That is what lies behind revolts, wild strikes and other unrest in society'. 60 An editorial in Dagens Nyheter from 1971, to give another example, stated that: 'public information that only goes in one direction -top down -is nowadays considered totally unsatisfactory -[. . .] information should be a "two-way process". If the communication is bad, it will probably generate elm issues'. 61 The 'elm issues' was a reference to a conflict in Stockholm in 1971, where a political decision to cut down trees to make room for a new metro line and a shopping mall was met with public protests and violence. The municipal commissioner most associated with the decision, the Social Democrat Hjalmar Mehr, explained himself that an 'information problem' was the cause of the conflict. The decision itself was not the problem, but that protesters were ignorant as a consequence of lacking information. As pointed out by Helldén, Mehr reproduced the traditional view of public communication: politicians should make the decisions and inform citizens after the fact. 62 This view clashed with communication ideals represented by the new social and political movements at the time -some of which organized the public protests against the decision to cut down the trees. The new movements emphasized direct democracy, participation, debate and dialogue. 63 Similar ideas came up when the elm issue was debated in the Swedish parliament in 1971. A representative from the Communist party stated that 'democratic processes must include proper information to those who are concerned by the decisions, and decisions made on the basis of extensive debate'. 64 Many others argued for public communication beyond the one-directional model in the early 1970s. A group of communication scholars led by Kjell Nowak summarized the main critique in 1970: 'Communication must mean a process where both parties can give information and have an impact on equal terms'. 65 This contrasting notion of communication emphasized transparency, dialogue, public participation, and active citizens. If these were the goals, the linear transmission model of communication fundamental to 'propaganda' was no longer adequate. The Swedish voices were not isolated, of course. Similar ideas were developed elsewhere. Hans Magnus Enzenberger argued in 1970 for a 'mass participation in a social and socialized productive process'. 66 In 1975, James Carey started to examine communication as culture, centred on the 'ceremony which draws persons together in fellowship and commonality'. 67 James Grunig, who presented his influential models for organizational communication in the mid-1980s, developed his critique of the dissemination paradigm already in the 1970s. 68 Related ideas were also reflected in the Swedish parliamentary debates: I find it very positive that the Finance Committee in its report sees public information as a communication between authorities -state and municipal -and citizens, not one way in the direction from above and downwards but also in the other direction: from the bottom up. It is important that the Committee states that public information is needed to enable citizens to express their ideas and interests in the democratic process and thereby influence the development of the society. 69 Both Nowak and the Social-Democrat in discussing the Finance Committee in 1971, quoted above, used the word 'information' when they described a symmetrical model of communication where citizens were active as participants. Internationally, 'information' became a central concept in engineering in the 1950s, and soon spread to the social sciences, politics, and popular discourse. 70 In Sweden, it was used in the parliament already in the early twentieth century, as a synonym to 'upplysning', but often in the plural. 'Information' and 'informations' (informationer) referred to specific intelligence about something, from someone. Those in need of information were often the parliamentarians themselves, who needed it in order to make informed decisions. 71 When the concept re-entered politics in the mid-twentieth century, it was borrowed from technical and scientific domains and had a wider use, mostly in its singular form. Norbert Wiener's theory of cybernetics popularized the idea that virtually all aspects of man and machine could be treated as problems of communication and information. Wiener explained in 1950: 'To live effectively is to live with adequate information'. 72 This broad and abstract understanding of 'information' made it a useful concept adaptable to two-way communication models. 'Propaganda', on the other hand, was not as flexible. It was hard to combine 'propaganda' with a view on communication characterized by two-way flows, and this made it less useful as a concept in the new political context. 'Upplysning' might have had similar limitations. It was often used together with 'propaganda' to describe campaigns and educational efforts. 'Several things need to happen if we want to achieve road traffic safety -information [upplysning] and propaganda, teaching in schools, removing dangerous obstacles on the roads, better vehicles and clear and well-thought-out traffic regulations'. 73 'Upplysning' did not carry the negative connotations that 'propaganda' did, but the history of the concept connected it to a similar communication model: those with knowledge were supposed to enlighten the ignorant. 74 The frequencies for the different keywords in Figure 3 show that the use of both 'propaganda' and 'upplysning' decreased when 'information' became more frequent. We can also see that words associated with two-way flows of communication -'dialogue' (dialog) and 'conversation' (samtal) -became more prominent from the 1970s and 1980s onwards. The political vocabulary of communication shifted, from one centred on the spread of 'propaganda' and 'upplysning' from the state to the citizens, to one focused on information and participation (Figure 4).
The shift in ideal communication models, from the 1960s to the 1980s, can also be illustrated by two governmental reports. In the previously mentioned report from 1969, the commission of inquiry stated that 'mass communication' had become more important than 'personal contact' in the communication between authorities and citizens. 75 In a follow-up report fifteen years later, a new commission of inquiry took a more balanced stance for both 'indirect communication' (meaning mass communication) and 'direct communication' (interpersonal communication) between bureaucracy and citizens. 76 This is not to say that public information from now on was a two-way flow, giving citizens status as equal communicators in relation to established institutions. Most public information continued to be mass communication but was labelled 'information' and not 'propaganda' or 'upplysning'. Here, 'rhetorical redescription' is a concept close at hand, describing attempts to reframe negatively perceived actions or states of affairs as something positive -or vice versa. 77 'Information' may have been used to dress up the practice of one-directional propaganda and 'upplysning' in a more neutral fashion. As a technical term rather than a contested political concept, it was well suited for a rhetorical redescription of politics as usual. What was left of 'propaganda' was mainly the negative connotations associated with it.

Conclusion
This article presents new findings and new interpretations of the conceptual history of propaganda in the political discourse of parliamentary debates. Firstly, in the Swedish parliament, 'propaganda' was mainly used in a negative sense between 1867 and 2019. There is no sign of a widespread notion of 'propaganda' as a positive concept before 1914, in contrast to what previous research has often claimed. Secondly, it became possible to use 'propaganda' in a positive and neutral sense between the 1910s and 1980s. This is a period when it is usually assumed that 'propaganda' became less useful as a neutral concept. The results confirm Schieder and Dipper's observation that the modern concept of 'propaganda' started as a negative characterization of messages spread by adversaries. 78 It was possible as a neutral political concept only after it was redefined in a commercial context, but this transformation was never completely successful. In the German context examined by Schieder and Dipper, 'propaganda' was impossible as a neutral concept after the fall of the Nazi regime. Nevertheless, the situation in postwar Sweden was very different.
This calls for an explanation: if 'propaganda' after WWII was heavily associated with Goebbels and totalitarian methods -why did Swedish parliamentarians use it in discussions about the promotion of fruit juice, traffic safety, maternity care, fish consumption, and antismoking efforts up until the early 1990s? There were obviously other forces at play that both supported the use of the concept and eventually made it problematic. A few contextual factors should here be considered: first the political consensus of the 1950s as a fertile ground for the promotion of non-controversial issues in a pre-war 'propaganda' fashion; then the emerging radicalization of society in the late 1960s and the critique of the state and authorities regulating citizens' lives. These developments, we argue, helped to make 'propaganda' useful in a neutral sense long into the postwar era, until the concept eventually became burdened by exclusively negative connotations. The process of changing communication ideals, from one-way to two-way communication, could be another explanation for this conceptual change, making the dissemination model associated with 'propaganda' and 'upplysning' out of date. In this process, the popularization of alternative communication concepts may have been an important driving force. 'Information' was a common buzzword in the 1970s -but it was still rare in the 1950s, when everyone was familiar with 'propaganda' (positive and negative). Initially, 'information-(s)' meant intelligence distributed within the parliament. The spread of 'information' in its singular form made it easier to use the concept in a variety of contexts. It could refer to anything and nothing, but usually carried positive connotations: nobody was against 'information' per se. Related communication terms such as 'dialouge' and 'conversation', which became more common in the 1970s and 1980s, are further indications that communication ideals were changing. A concept like 'information' seems to have provided a space for political action that 'propaganda' was not able to. Thus when 'information' was available it became less tempting to suggest 'propaganda' as a political solution.