Comunicación y Sociedad Universidad de Navarra | Facultad de Comunicación
Username: Password:
Ok Sign inForgot your password?

Calidad Revistas Científicas Españolas
Author / Norbert KERSTING Professor. University of Münster, Germany.
Author / Abel REIBERG Researcher. Freie Universität. Berlin, Germany.
Author / Phillip HOCKS Researcher. University of Münster, Germany.
More authors:  1 2 3
Article / Discourse Quality in Times of Populism: An Analysis of German Parliamentary Debates on Immigration Policy
Contents /

1. Introduction

It seems that populism is on the rise around the globe. In Europe, it is the right-wing populists that have won supporters e.g. in France, Italy, Great Britain, Netherlands, Belgium, etc.  For decades, Germany was an exception to this trend. However, in recent years, particularly in the course of the European migrant crisis, populists were able to increase their influence. The right-wing party Alternative für Deutschland (AfD) is close to winning mandates in national elections and established parties are also reacting on the populist trend.

Populism is considered a serious threat to democracy because it reduces complexity to the extreme, by ignoring the plurality of society and denying the existence of diverse interest groups. (See for the notoriously vague definitions Engesser, Ernst, Esser, and Büchel (2016)). In this article, we examine one of the pillars of democracy, political discourse, in order to see if the erosive effects of populism manifest. We analyze discourse quality by calculating the cognitive complexity of contributions to the discourse in the central arena of the democratic system, the parliament. The focus of the analysis is on debates on immigration policy, in which a spread of populism and a decline of discourse quality seems most likely. The period studied is the 18th legislative period, from its beginning (October 2013) until the time that the data was gathered (December 2016).

Our approach to analyze deliberative quality heavily relies on existing work (Kersting, 2005, 2012; Kies, 2010). Brundidge, Reid, Choi, and Muddiman (2014) analyzed the quality of discourse in U.S. political Blogs using the concept of Cognitive Complexity. Following their approach Beste and Wyss (2014) investigated the relationship of Cognitive Complexity (CC) and discourse quality thoroughly on a theoretical and methodological level. Following the approaches of various authors, they consider the LIWC dictionary, developed by Pennebaker and Francis (1996), a valuable tool for calculating CC. Using the LIWC, Wyss, Beste, and Bächtiger (2015) studied the development of discourse quality in the Swiss parliament. Mainly applying the methods developed by said authors, we examine the quality of discourse in the German parliament.

The article is structured as follows. Corresponding to the title, we firstly point to the reasons for the widely-shared opinion that we are living in “times of populism”. We then elaborate “discourse quality” as an important indicator for the state of democratic political systems. Following this, we explain why the performed “analysis of parliamentary debates on immigration policy” can be considered a suitable procedure, when looking for the erosive effects of populism on political discourse. The details of the analysis are explained in a methodological section. After giving a detailed overview on the discourse quality in debates on immigration policy in the German Parliament in the result section, we conclude, that (against our expectations) discourse quality stayed rather stable. Even in the years when populism was fueled by the events of the European migration crisis, the discourse quality in the central arena of the political system did not show a decline in the form of a clear negative trend.


2. Times of Populism

“We are seemingly living in populist times” - This impression, articulated by Moffit (2016) in the beginning of his recent book, is widely shared by political scientists. Looking at the large and growing body of literature it becomes clear, that the interest in populism does not just resemble a scientific fashion, but relates to the increasing presence of a political phenomenon. Around the globe populist movement are gaining momentum and populist leader are taking office. As becomes apparent when looking at the different examples, populism ideology or populist rhetoric is shown by actors on both sides of the political spectrum. While American “neo-populism” (Roberts, 1995) is very visible also in left-wing parties and movements, it is the right-wing parties that are responsible for the “populist zeitgeist” (Mudde, 2004) in Europe. Here parties like the front national, Partij voor de Vrijheid or the UK Independence Party were able to increase their share of votes almost continuously in the past decades.

For a long period, Germany was an exception to this trend. In Germany, the political party system since World War II consisted of the conservative parties (the Christian Democratic Party CDU and its Bavarian sister party Christian Social Union CSU), the liberal party (FDP) and the social democratic party (SPD) on the left. In the 1980s, the left ecological oriented green party (Büdnis 90/Die Grüne) emerged and with the German unification, the left party (Die Linke) entered the Bundestag. Immigration policy controversies were mostly resolved in the mid-nineties with the Dublin treaty and new according German legislation. In the 2010s however, a new right-wing party emerged. Firstly, focusing on anti-EU the Alternative für Deutschland (AfD) concentrated on immigration policies. In mid-2015 the ruling coalition of CDU, CSU and SPD und Chancellor Merkel, suspended the Dublin regulations. More than 800.000 refugees entered Europe mostly via the so-called “Balkan route”. On the one hand, the German population welcomed the political refugees, on the other hand the AfD and parts of the Bavarian CSU highly criticized this new policy. During the New Year’s Eve incident of sexual harassment in cologne caused new anti-refugees sentiments within the German population.

Posing itself as an alternative to the established “elites”, the AfD won mandates for the parliaments in 13 of the 16 German states. Based on its success in elections, the party established itself also as an actor on public political discourse. It seems that with Germany yet another European country is experiencing the rise of right-wing populism.

  1. According to him, the term “people” as used by populists signifies a positively “imagined community”, while the label of the “corrupt elite” is often attached to specific political opponents of populists. With the distinction of   “the” people and “the” elite, populists deny the plurality of a given population or electorate and often present themselves as the representatives of the “volonteé general” shared by a homogeneous community. The “thin ideology” of populism can be attached to various “thick” ideologies like communism or nationalism etc. (Mudde, 2004). Seen as a mode of communication, it seems the more possible that populism may diffuse between political actors. Bartolini (2011) e.g. compares populism to a “virus” that infects different party systems. This metaphor underlines the understanding of populism as a subject of diffusion among political actors and as a threat to democratic political systems. It is the majority of scholars that worked on populism (though with relevant exceptions) who perceive populism as a danger to democratic political systems (Gidron & Bonikowski, 2013). The anti-pluralistic core of populism seems incompatible to the established norms of democracies. While the erosive effects of populism may result in the disablement of democratic institutions and the official abolishment of democratic rights, it seems more likely, that the erosive effect will firstly show in discursive practice. Since the unwritten rules to democratic discursive practice are constantly reestablished in contested fashion, a populist turn may firstly appear here. In the following, we will explore whether an erosion of democracy is becoming apparent in the German political system, focusing on the quality of political discourse.


3. Quality of Discourse

The focus on political discourse is preferred over a focus on political institutions because of two problems of institutional analysis: First, institutions do not change easily and therefore only undergo alterations slowly. Changes on a short-term scale like reactions to certain events will not be mirrored within the characteristics of the institutional setting. Second, changes might not even occur at all. As Wyss et al. (2015) show with reference to the example of Switzerland, political institutions might be immune to transformation even if the political culture has changed. Because institutions are rather resilient to changes, a focus on informal practices can be fruitful to detect political change. One central informal practice of politics is the way politicians conduct their speeches. In every session of the parliament, politicians share their thoughts and ideas through speeches. Democracies are highly dependent on this exchange of arguments between actors. Discourse ethics is considered a necessary condition to agree on and introduce legitimate policies in modern societies (Chambers, 1995). Scholars of deliberative studies agree on the idea, that decision-making “should be “talk-centric” rather than “voting-centric”, i.e., outcomes should be determined by reasons rather than numbers” (Spröndli, 2003, p. 1). Which means, that the process of decision-making rejects pre-existing preferences but rather considers political actors as “listen to each other, reasonably justify their positions, show mutual respect, and are willing to re-evaluate and eventually revise their initial preferences through a process of discourse about competing validity claims” (Steenbergen, Bächtiger, Spörndli, & Steiner, 2003, p. 21). Thus, it is assumed, that a better discourse, or in other words, a higher discourse quality, enhances the decisions it leads to. Fishkin (1995) argues that preferences will be better informed and Cohen (1989) that decisions will be more legitimate. A high discourse quality distances the decision-making process from a simple bargain or other trade-offs.

When accepting the important role of deliberation for democracies, the question remains how to measure good deliberation. We consider two approaches, which follow a similar logic but differ by implementation. Steenbergen et al. (2003) have developed one of the most renowned approaches: The Discourse Quality Index (DQI). The DQI roots on Habermas’ discourse ethics and considers concepts emphasized in Habermas’ work: participation, justification, respect and constructive politics. The results indicate to which extend a discourse complies with Habermas’ principles of ethnic discourse (Steenbergen et al., 2003).

Since the coding needs to be done manually, applying the DQI to a larger sample is costly. Therefore, we opted for an alternative approach based on measuring Cognitive Complexity (CC). This approach is rather narrow, but increases the usability for large datasets. CC was originated in psychology and is used “to analyze the sophistication of individual’s belief systems and information processing” (Beste & Wyss, 2014, p. 1). It is a two-dimensional concept, which measures the differentiation (1) and integration (2) of a speaker or writer (Owens & Wedeking, 2011). First, differentiation captures the quantity of aspects an individual considers in his/her speech, the extent of perspective. For example, the issue of migration might consist of an economic (labor market integration), social (adaption of social system), moral and security dimension. Second, integration measures the understanding of depth of a certain issue. To what extend is the writer or speaker able to link relationships, and subordinations among the different dimensions (Wyss et al., 2015, p. 6).

The concept of deliberation, CC and populism stand for their own, but overlap greatly in certain areas. On the one hand, deliberative theory deals with openness, reflexivity, rationality and respect (Kersting, 2005). CC focuses on complex reasoning and the existence of a wide range of interests and attitudes (Beste & Wyss, 2014). On the other hand, populism as a “thin”, less elaborate ideology focusses on a “pure people”, with an idealized conception of the community, their sovereignty and the postulation of a “corrupt elite” and a “dangerous Other” (Mudde, 2004; Taggart, 2004). In this regard, populism lacks the respect for the “Other”, which is an important category in the theory of deliberative quality.

Discourse Quality and CC agree on the enhanced results through good deliberate processes. In addition, CC and deliberative theory focus argumentation. Not only the number of arguments is considered important in both approaches but also interdependencies have to be born in mind. As Beste and Wyss (2014) have shown in their study empirically, the results of CC and DQI regarding the qualities of a debate overlap sufficiently. Thus, CC can be used as a measure to examine the quality of a discourse. The reason why we applied CC as our concept from a practical point of view will be presented in the methodology part. In the following section, we delve on the reasons for the focusing on parliamentary minutes as a source of data.


4. Parliamentary Debates

Parliamentary discourses provide researchers with the possibility for a dynamic perspective on political change. The nature of speeches varies heavily over time, issue, speaker etc. Parliamentarians are well-interconnected individuals who react in their discourse contributions to events on an almost daily basis. Parliamentary debates are in general considered a stage for government and opposition (Brennan & Hamlin, 1993) and the parliament is seen in a crucial role within the democracy (Wyss et al., 2015). The parliamentary sphere remains the place where politicians meet, argue, dispute and decide on policies and laws. It is, despite new possibilities within the media landscape, still the arena where they can display their beliefs and policy preferences. Thus, the parliamentary speeches can be considered a valuable source.

From all speeches held in the parliament during the 18th legislature, we selected those speeches covering immigration policy for our analysis. Immigration policy is not only a field of particularly intense political discourse, but also a field where participation in political discourse has proven particularly rewarding for right-wing populist. The anti-pluralist position of right-wing populists seems to fit well to an anti-immigration position. In the reasoning of parties like the AfD, immigrants are rhetorically excluded from the “imagined homogeneous community” referred to with the term people. But differently, right-wing populists like the members of the AfD are not only distancing themselves from the perceived “corrupt elite”, but also migrants as “parts of a foreign people”. Often this goes hand in hand, for example, when the AfD is criticizing “Merkels Migration Misery”.

With such reasoning in immigration policy debates, right-wing political parties have gained momentum in Europe. In Germany, the AfD, which had established itself as an EU-critical party, soon shifted its focus on immigration policy. It then saw a massive growth in members, when the discourse on immigration policy dominated the political arenas, at the time of the so called “European migration crisis”. With the national elections coming up it seems like for the first time since the founding of the green party, another party may establish itself on a national level. This success of the AfD logically resembles a threat to competing parties. To this threat, the already established political parties may react by adopting strategies or positions, that for the AfD have proven successful. This makes most sense in the field that the AfD is winning its votes – immigration policy. Therefor in debates on immigration policy the spreading of populism and an erosion of discourse quality may seem most likely. Our research tries to analyze whether the political discourse in the German Parliament became less complex and if the quality of discourse declined. Or as a counter reaction to the growing populism, the Parliament pertained a higher quality of deliberation. How the analysis of debates was performed is explained in the following section.


5. Methods

As explained above, the quality of political discourse was analyzed focusing on parliamentary debates. The parliament cannot only be considered a key forum for political discourse in modern democracies, it also resembles an accessible source of data. In most countries, the parliamentary debates are protocoled in parliamentary minutes. The Bundestag, which may be considered the more relevant of two chambers of German parliament, is no exception in this. The minutes of the German Bundestag are published on the website of the Bundestag. The minutes were processed and annotated in the course of a previous research project[1]. Of these, only the speeches on immigration policy were relevant to our study. This view follows a most likely case design: As explained above, right-wing populist parties often focus on issues of integration and immigration policy. Furthermore, an appliance of a populist approach may turn out particularly promising in such debates. Accordingly, a diffusion of populist reasoning may be particularly likely in debates on immigration policy. If a general decline of quality of political discourse appeared in the last years, it is likely to become visible here. Following this logic, the selection of speeches on immigration policy resembles the first part of our analysis.

This selection was based on a significantly high share of specific keywords in a given speech. Similar to the procedure presented by Blätte and Wüst (2017) the keywords themselves were gathered in a two-step process. First, a number of seed words in a canonical form were defined. With these seed words, a first set of speeches on immigration policy was merged to a sub-corpus. In the second step, this sub-corpus was compared to the rest of the parliamentary speeches in order to identify the typical wording in immigration policy related speeches. As such, words that were significantly often part of the sub-corpus compared to the rest of the corpus were considered. With the derived dictionary of words typical for immigration policy debates, speeches where gathered as a (second) sub-corpus for the actual analysis. Following their selection, the speeches were analyzed in regards to their discourse quality. For this, cognitive complexity was considered a proxy of discourse quality.

As outlined in the theoretical approach CC is a concept which measures sophistication and quality of human expression on a two-dimensional scale. Historically the measurement and coding has been done manually. Thanks to computerized procedures, such time-consuming efforts are not necessary anymore. Despite advantages in reliability and saving resources, computer-based approaches bear risks. One might argue, that it seems near to impossible for a computer to identify the differentiation and integration of a speech or text, both aspects of the CC. Different approaches have been developed to solve the issue of manual coding procedures and computing the CC. Several of these techniques rely on the Linguistic Inquiry and Word Count dictionary (LIWC). The LIWC is a dictionary, which groups words and relies on the count of their prevalence. It has been applied numerous times in different fields of social science research and is considered as a reliable and valuable tool among experts (Beste & Wyss, 2014). Beste and Wyss identified four approaches by different studies to compute CC with the LIWC. All these studies differ[2] in the formula and LIWC categories they use. As their study has shown, the approach of Owens and Wedeking (2011) proved to be the most fruitful[3] when computing the quality of discourses. The approach contains the following criteria of words: causation, six-letter, insight, discrepancy, inhibition, tentative, inclusive, exclusive, negations and certainty words (Owens / Wedeking 2011, p. 1055–1057). The score for CC is calculated[4] using the formula:


To illustrate the procedure, the following example is presented: The category of causation[5] contains words such as because or hence and counts them within each unit of observation. The underlying assumption is to measure the distinctiveness and separation of ideas and arguments (Owens & Wedeking, 2011). Evaluating all existing approaches and especially the results by Beste and Wyss, we decided to apply the formula by Owens and Wedeking to calculate the CC scores. The calculation was done in R, while the required data on the prevalence of word categories in the text corpus was calculated with the LIWC software. 

As mentioned above, the text corpus already included data on the speaker, among that, the name, function and party affiliation. Additional data on the speaker, regarding age, gender, state of origin and type of mandate was gathered from the website of the Bundestag. Furthermore, data indicating the vote share of the AfD was derived from the “politbarometer”[6] poll and included in the analysis, as will become visible in the results section.


6. Empirical Results

After the introduction of a theoretical concept regarding the quality of discourse and the outline of measurement approaches using CC, we will now set forth preliminary results. Therefore, we analyzed all contributions to debates on immigration policy in the German parliament. Before taking a closer look at the prevalence among parties and other possible influencing factors, we would like to present the distribution along the CC values. As can be seen in Graph 1, the prevalence of the CC values follows approximately a Gaussian bell curve. Based on the above presented algorithm, we obtained values ranging from 26.36 to 54.60 with an average score of 38.6.



 Graph 1. Histogram of DQ



Following this, the parliamentarians’ statements were classified according to their quality. Every statement is considered of having a good discourse quality if it surpasses the threshold of 42 scale points. Statements underneath the threshold value of 37 are considered as poor quality. The remaining statements are classified neutral. This allocation yields in 145 good, 669 neutral and 280 poor statements. 

Before turning to the development over time and the influence of the populist-party AfD, a few descriptive insights will be presented to enable the reader with a better understanding of the underlying structure of the dataset. From here on, absolute values of CC are replaced by centered values of the variable. This enables the reader to better understand the relations of changes and differences, since the variable is now scaled as a share of the standard deviation. On the individual level, examining each speaker, the characteristics render the following distributions:

As could be expected, there is no significant difference of discourse quality between male and female. The mean value of both categories is almost identically, -0.01 and 0.01 for women and men respectively. So technically women have a slightly worse discourse quality but only by a margin of two percent of the standard deviation. The variance among the male speaker is more broad, especially due to a higher number of extreme outliers. Nonetheless, no significant differences in discourse quality exists between male and female members of the German parliament.



 Graph 2. Boxplot of  DQ by Age




Graph 3. Boxplot of DQ by Gender




A similar result is obtained when focusing on the discourse quality among different age groups. As illustrated in Graph3, despite the youngest age group up to 29 years the prevalence seems to be the alike. The average qualities of statements are located close to zero. Also the variance is similar for all five oldest age groups, 75% of all statements lie approximately within two standard deviations from the overall mean. Only members of parliament younger than 30 have a positive average quality score of 0.62. But these findings are skewed, since this age group is by far the smallest with only six statements by five different speakers.


Graph 4. Boxplot of DQ by type of mandate




Another interesting influencing factor might be the type of mandate. The German electoral system provides two possibilities: A member of parliaments can get elected either via a direct ballot, winning the majority vote in one’s personal electoral constituency, or via the party’s list of nominated candidates as part of the proportional representation. Again, a large difference does not occur between these two types of mandates. Graph 4 illustrates, that parliamentarians who won their seat through the majority vote have slightly higher discourse quality than candidates from the party list. The average difference between both electoral types amounts to 0.28 standard deviations. Keeping in the mind the variation of the variable, this difference can be neglected.



Graph 5. Map of Germany by DQ



Last, we will take a closer look at the discourse quality compared to the state of origin of the speaker. One might expect, that parliamentarians of some regions might be more inclined to lower discourse qualities based on a higher prevalence of populism in their local political sphere. As Graph 5 shows, there are regional differences. However, these differences are limited in scope. Members of the German Bundestag from Bremen have the lowest discourse quality above all with an average value of -1.37. They are followed by Hamburg (-0.63) and Thüringen (-0.53). By far the best discourse quality average is measured in the case of representatives from Sachsen-Anhalt with 0.49. Other states with high averages are Bayern, Mecklenburg-Vorpommern and Schleswig-Holstein. All other federal states fall somewhere in between this range, as displayed in Graph 5. Yet underlying causation and influencing factors remain. First, these descriptive results are to a certain extend skewed. Bremen, the state with the lowest average value of discourse quality by far, has only two statements within the dataset, both from Charlotte Motschmann (member of the conservatives). Other states contributed more statements to the debate, but Saarland for example also had two different speakers. Thus, the average values of those countries with little participation in the debate are heavily influenced by performances of a few of individuals.

Drawing back on our theoretical approach, an interesting approach might be to further investigate the relationship between the spread of populism and the discourse quality in the German Bundestag. So far, we rather approached the issue from a federal-holistic point of view. But one might argue, that not developments on the federal but on the state level cause politicians to adjust their speeches. Due to limited resources, we were not able to further investigate this relationship further in the Länder.

So far, the results for the individual level of discourse quality in debates on immigration policies at the German Bundestag did not render many valuable insights. Neither gender, nor age or the type of mandate helped in explaining differences in the discourse quality. The origin state of a speaker seems to account for some differences. Members of parliament from Sachsen-Anhalt, Nordrhein-Westfalen, Bayern Sachsen, Schleswig-Holstein and Mecklenburg-Vorpommern showed a lower quality of discourse, Parliamentarians from Thüringen, Saarland, Bremen and Hamburg showed a higher one. This may be explained by provincial elections in some of the Länder, nevertheless the strong relationship between the east-west cleavage did not appear.

Next, we will focus on the aggregative level of party membership. Here the discourse quality scores, measured through CC, are aggregated based on the fraction membership in parliament. During the 18th legislative period the green party (Bündnis 90/Die Grüne), the socialist party (Die Linke) were opposition parties. The social democrats (Sozialdemokratische Partei Deutschlands) and the alliance of conservative parties (Christlich Demokratische Union and Christlich Soziale-Union) were in a government coalition.



Graph 6. Parliamentarian statements classified by parties



As introduced in Graph 1 we classified statements according to their quality on a three-dimensional scale. This measure is meant to simply highlight the differences between the parties. Graph 6 displays the distribution of statements by party over the entire sample period. The high share of poor statements in the contributions by the green party is the most evident difference found. Almost every second statement by delegates of the green party falls in the category of poor quality. Second in this regard is the socialist party with a share of 30% statements of poor quality, whereas the social democrats and the conservatives have 25% and 18% respectively. Looking at statements with good quality, the rank order simply reverses. Here, the conservatives lead with 16%, followed by the social democrats with 14%, the socialist party with 10% and the green party with 4%. This leads to the conclusion, that indeed there exist party differences in discourse quality in the German parliament. Here opposition parties seemed to have a lower deliberative quality and showed a lower complexity in their statements. It remains to be seen whether these differences were existent also in the long term or developed over time through influencing factors.



Graph 7. History of DQ of Conservatives



Graph 8. History of DQ of Social Democrats



Graph7 to Graph10 display the history of discourse quality in the immigration debate over the 18th legislative period. As could be expected after the just described results, the discourse quality of the green party is significantly worse than that of the other parties. Nonetheless, comparing each party with itself, it becomes obvious, that all parties tend to remain stable. There are differences over time, but no overall trend can be identified. Particularly the socialists have a large variance on discourse quality per month, ranging from 1.63 to -1.33. Whereas the green party average negative values most of the time, the conservatives scores almost always positive cc-values. Surprisingly, the CC values of all parties underwent major drops in 2016, but not at the same time. In February 2016, the conservatives worsened their discourse quality by one standard deviation compared to January. The same applies for the green party in January and the socialists in March. Only the social democrats remained stable in the beginning of the year and even improved their discourse quality. Their speeches lost quality later in May. The Socialists and the green party are the two parties who experienced a second fall of discourse quality in 2016, the socialists in July and the green in October.



Graph 9. History of DQ of Greens



Graph 10. History of DQ of Socialists



Especially the almost simultaneous drop in the beginning of 2016 was expected. The discourse on immigration policy at that point in time was mainly covering the events in Cologne on New Year’s Eve. A high number of Incidents of sexual harassments by immigrants in that night had caused a shift in media coverage and more negative reports on refugees. It may be argued that these incidents may have caused a strengthening of populists and a decline of discourse quality. Doubtless, the populist party AfD very much focused on the incidents and significantly gained support. In the following, we will look at the role the AfD regarding the discourse quality in the German parliament.







Graph 11. History of discourse quality and AfD poll results



Graph 11 outlines the relationship between the overall discourse quality in the German parliament in debates on immigration policy and the estimated voters support for the AfD. At least from mid-2015 on a clear upwards trend for AfD is evident. In less than two years, the party gained almost ten percent in the polls. It becomes clear, that with the exception of the first quarter in 2016 (shortly after the cologne incidents) there seems to be no clear relationship between this upward trend of the AfD and the quality of discourse. The short- term decline of the discourse quality in the first term of 2016 may be caused by the more negative press coverage after the incidents in cologne and the growing support for the AfD. However, the average level of discourse quality in the Parliament remains stable and a negative trend cannot be identified.

Evaluating the results from Graph11 we can therefore conclude, that there is no direct influence by the AfD on the quality of discourse on immigration policy in the German parliament. However, it is worth noting that the AfD yet remains a player outside the parliamentary arena. This is likely to change after the national elections in September 2017. Thus, the question, if and how the AfD is able to influence the quality of discourse, is worth further investigation then.

To summarize the above presented findings, we conclude that the discourse quality varies slightly over time but clear long-term trends are not evident. In fact, the quality of discourse stays rather stable. In contradiction to the theoretical derived expectations, there is only a short period in 2016 where low discourse quality is related to AfDs strength. There is no correlation between discourse quality and the age and gender of the speakers and only minor differences regarding the MPs state of origin. Finally, the largest difference for the discourse quality lies within the parties. The quality of the green party’s and the socialist party’s statements have a lower quality than the two other parties have. Interestingly the two parties with lower quality also resemble the opposition in the parliament. From that, it can be hypothesized that the government, or at least a multi-party government, scores higher in CC values compared to the opposition.


7. Conclusion

In this article, we examined discourse quality in times of populism. Therefore, we analyzed debates on immigration policies in the German Parliament using a quantitative, dictionary based approach to measure Cognitive Complexity as a proxy for discourse quality. As the results show, the discourse quality differs greatly between the parliamentary groups with the statements of opposition parties (green party and the socialist party) having a lower complexity then those of the ruling parties (social democrats and the conservatives). Opposition parties seem to argue less complex.

A clear influence between the rise of the AfD and discourse quality in parliament could not be detected. Hence, we concluded, that populism did not enter the German parliament yet. After early 2016 the negative media coverage on refugees and the rise of the AfD had only short-term effects. In the long-term Parliamentarians resisted the temptation of populism.

Mainly due to limited resources, we were only able to monitor the 18th legislative period. This implies that aspects like the possible decrease in discourse quality between government and opposition could not be investigated further. An approach across different configurations of the parliament would be necessary to identify whether the found differences in parties’ discourse quality is due to the role of government and opposition or due to partisan characteristics. In addition, we assumed as part of the case selection, that debates on immigration are the most likely to be influenced by populism. We regard this assumption legitimate, however a test based on debates concerning different policy fields would be desirable.

Furthermore, we assumed that populism and poor discourse quality go hand in hand. This assumption needs to be further investigated, ideally with a direct approach of measuring populism (Engesser et al., 2016). Last, as we concluded in the results section, the AfD yet is an outside player of the parliamentarian arena. Therefore, the situation is different as for example in Switzerland, where the populist party SVP entered the parliament, which resulted in a significant drop of discourse quality (Wysset al., 2015). Further research should focus on those settings in which the AfD is part of the parliamentary arena. This is either possible on the state level with currently thirteen parliaments with AfD membership or likely after the next election on federal level in September 2017, when the AfD will likely win mandates for the Bundestag. 

To conclude, in times of populism discourse quality in the German Bundestag debates seems to be stable so far. Due to German history, the political elites and candidates in the national parliament resist populist and nationalistic rhetoric. It will be interesting to see, if this trend continues with right-wing populists as members of the parliament.






Bartolini, S. (2011). Political Parties, Ideology and Populism in the Post-Crisis Europe. Poros Conference 7-10 July 2011. Retrieved from

Beste, S. & Wyss, D. (2014). Cognitive complexity as a proxy for high quality deliberation? Retrieved from

Blätte, A. & Wüst, A. M. (2017). Der migrationsspezifische Einfluss auf parlamentarisches Handeln: Ein Hypothesentest auf der Grundlage von Redebeiträgen der Abgeordneten des Deutschen Bundestags 1996–2013. Politische Vierteljahresschrif, 58(2), 205-233.

Brannan, G. & Hamlin, A. (1993). Rationalizing parliamentary system. Discussion Paper Series in Economics and Econometrics 9313, Economic Division, School of Social Sciences, University of Southampton.

Brundidge, J., Reid, S. A., Choi, S. & Muddiman, A. (2014). The “Deliberative Digital Divide:” Opinion Leadership and Integrative Complexity in the U.S. Political Blogosphere. Political Psychology 35, 741–755.

Chambers, S. (2009). Rhetoric and the Public Sphere: Has Deliberative Democracy Abandoned Mass Democracy?. Political Theory 37(3), 323-350.

Cohen, J. (1989). Deliberation and Democratic Legitimacy. In Matravers, D., Pike, J. (2003), Debates in Contemporary Political Philosophy. London: Routledge in association with the Open University.

Engesser, S., Ernst, N., Esser, F. & Büchel, F. (2016). Populism and social media: how politicians spread a fragmented ideology. Information, Communication & Society 8 (20), 1109-1126.

Fishkin, J. S. (1995). The Voice of the People: Public Opinion and Democracy. New Haven: Yale University Press.

Gidron, N. & Bonikowski, B. (2013). Varieties of Populism: Literature Review and Research Agenda. Weatherhead Working Paper Series, No. 13-0004. Retrieved from

Hawkins, K. A. (2009). Is Chávez Populist?: Measuring Populist Discourse in Comparative Perspective. Comparative Political Studies 42(8), 1040-1067.

Kaman, M. (2016, November 14). Schröder lobt Gabriels Populismus – nur was bringt’s?. Welt. Retrieved from

Kazin, M. (1995). The Populist Persuasion: An American History. New York: Basic Books.

Kersting, N. (2005): The quality of political discourse: Can E-Discussion be deliberative?. Paper at annual conference of the British Political Science Association Leeds 5-7 April 2005.

Kersting, N. (Ed.) (2012): Electronic democracy. Opladen: BB publisher. IPSA series: The World of Political Sience.

Kies, R. (2010): Promises and Limits of web-deliberation. New York: Palgrave.

Moffitt, B. (2017): Global Rise of Populism. Performance, political style, and representation. Stanford: Stanford University Press.

Mudde, C. (2004). The Populist Zeitgeist. Government and Opposition, 39, 541–563.

Owens, R. J., & Wedeking, J. P. (2011). Justices and Legal Clarity: Analyzing the Complexity of U.S. Supreme Court Opinions. Law & Soc'y Rev 45, 1027–1061.

Pennebaker, J. W. & Francis, M. E. (1996). Cognitive, emotional, and language processes in disclosure. Cognition and Emotion 10(6), 601-626.

Roberts, K. (1995). Neoliberalism and the Transformation of Populism in Latin America: The Peruvian Case. World Politics 48(1), 82-116.

Spörndli, M. (2003). Discourse Quality and Political Decisions: An Empirical Analysis of Debates in the German Conference Committee. Discussion Paper No. SP IV 2003-101, Wissenschaftszentrum Berlin für Sozialforschung.

Steenbergen, M. R., Bächtiger, A., Spörndli, M. & Steiner, J. (2003). Measuring Political Deliberation. A Discourse Quality Index. Comparative European Politics 1, 21–48.

Taggart, P. (2004). Populism and representative politics in contemporary Europe. Journal of Political Ideologies 9(3), 269-288.

Wyss, D., Beste, S. & Bächtiger, A. (2015). A Decline in the Quality of Debate? The Evolution of Cognitive Complexity in Swiss Parliamentary Debates on Immigration (1968–2014). Swiss Political Science Review 21, 636–653.






[1]The so called Polmine project was conducted by the Institute for Political Science at the University of Duisburg-Essen in collaboration with the Institute for German Language in Mannheim (IDS) and resulted in the Polmine Text Corpus. The corpus can be analyzed using the so-called corpus workbench (CWB) of the IDS as well as the Polmine R package. Specific versions of the corpus as well as the mentioned software are publicly available.

[2]According to the study by Beste and Wyss, the correlation coefficients (polychoric) range between 0 and 0.93 between the different approaches. Indicating that the choice of formula influences the results heavily.

[3]Polychoric correlation coefficient of 0.56 of the Owens and Wilkens (2011) approach to the manually coded values (Beste & Wyss, 2014, p. 9).

[4]Values for the categories have been standardized.

[5]For further information regarding other categories, please see Owens and Wedeking (2011).

[6]See for more information.

© Communication & Society - School of Communication - University of Navarra | Contact Us | Legal Notice | Site Map