!
!
A disproportionate degree of Euroscepticism in the Dutch media?
A comparison between the Dutch news media coverage of the European Union, real-life trends and public opinion
!
!
!
!
!
!
!
!
!
!
Graduate School of Communication Political Communication Master Thesis
!
Thesis supervisor: Dr. D. C. Trilling!
Student: Daan Brinkman (10592172)!
!
June 27, 2014!
Abstract
An interesting question in communication science deals with the intensity of news reporting about certain issues. In this study a computer-assisted content analysis of 97.310 Dutch newspaper articles about the European Union for the period 1999-2014 is conducted in order to investigate whether the tone of this coverage is as Eurosceptic as assumed. Besides, the extent to which the newspapers reflect real-world trends and the public opinion is measured. Results indicate that the overall sentiment of the Dutch coverage of the EU is slightly positive instead of negative and Eurosceptic. Furthermore, it turned out that the analyzed newspapers do not reflect most real-world trends in their articles about the European Union.
Introduction
„Up yours Delors. At midday tomorrow Sun readers are urged to tell the French fool where to stuff his ECU”.
This was the headline of British newspaper The Sun on November 1st, 1990, as a response to the then French President of the European Commission’s supposed attempts to force European federalism upon the UK. British tabloids are known for their Eurosceptic news coverage, however, they are certainly not alone in their use of a negative tone when covering the EU. An intriguing question in communication science deals with effects of this intensity of news reporting about certain issues.
Soroka (2012) suggest that mass media produce content that is systematically more negative than reality. In his study this is the case for unemployment, inflation and interest rates, however, it seems unlikely that this selection mechanism is only unique to these subjects. Iyengar (1985) tested this hypothesis for the issues of energy and - again - inflation and unemployment. Indeed, the (television) news coverage turned out to be more negative than real-life trends are and influenced the public concern in this way. It follows that, if media overrepresent negative trends, people will tend to have a view of the topic that is somewhat more negative than is warranted. Applying this to trust in the European Union, a study of de Vreese (2007) showed that framing EU politics in a negative way can fuel public cynicism an scepticism.
This again proves the impact of news coverage on public opinion, in this case about the European Union. That is why de Vreese et al. (2007) state that Euroscepticism is - at least partly - „a product of the diet of information one consumes about European affairs”. Others even say that the media is the one to blame for the democratic deficit of the European Union (Statham, 2006). Given the fact of decreasing trust in the European Union (Eurobarometer, 2013) and low election turnout rates (European Parliament, 2009), these allegations - if true - suggest a major negative role of the media in the European democratic system. This paper seeks to investigate for the case of Dutch newspapers whether it is true that the tone of the media coverage of the European Union is Eurosceptic, to what extent they cover real-world trends and events concerning the European Union and whether the media reflect the public opinion at that moment. Besides, it will consider whether media coverage derives its sentiment about Europe to national trends or European trends.
Based on these questions the following research question is formulated:
!
„To what extent reflect Dutch newspapers real-world developments and public opinion in their coverage of the European Union?”
Theoretical framework
As a starting point, it is necessary to state that negative news is more attractive than positive news - which is sometimes even seen as ’non-news’ - and therefore has a bigger impact. Soroka (2006) for example shows this in his study on asymmetric responses to economic information, in which he found that public responses to negative economic media content are much greater than are public responses to positive economic media content. In another study Kleinnijenhuis et al. (2006) focused on political news instead of economic news. This study showed as well that negative news on television and in newspapers has a much bigger impact on trust in party leaders and even on voting intentions than positive news. So from this point of view it is understandable that there is plenty of negative news coverage about the European Union, which is proven by several studies (Hawkins, 2012, de Vreese, 2007, de Vreese et al., 2007). The first hypothesis is therefore:
!
H1: The sentiment of Dutch newspapers’ coverage of the European Union is overall negative
!
The second element that this study seeks to investigate is whether the tone of media coverage reflects real world developments or whether this sentiment is more negative as well. The relevance of this comparison is nicely expressed by Soroka (pp.30, 2003): „A central issue in the study of media effects is being able to distinguish between effects that are truly media driven and effects that are simply the product of world indicators. If the media are only reflecting real-world circumstances, what may appear as a media effect is more appropriately viewed as real-real-world effect. In short, „media effects” lie somewhere in the gap that exists between media content and reality.”
A theoretical basis for the presence of more negative news coverage than real-life trends would suggest can be found in the framing concept. The process of framing involves a communication source presenting and defining an issue (de Vreese, 2005), in a way that resonates with existing underlying schemas among their audience (Shoemaker & Reese, 1996). This does not mean that most journalists try to manipulate their audiences, but it can be useful to decrease the complexity of a certain topic in their outlets (Scheufele & Tewksbury, 2007). As an example, Schuck & de Vreese (2006) examined the impact of risk and opportunity frames on public support for the enlargement of the European Union in 2004. The use of these kind of frames might be useful for understanding what the precise advantages and disadvantages of EU enlargement are. On the other hand, results of this study show that participants in the opportunity frame condition showed significantly higher levels of support compared to participants in the risk condition. So, journalists
should bear in mind that they are influencing the public opinion of citizens towards the concerned topics.
In this paper the focus will be just on the use of negative or positive frames in covering the European Union by the Dutch media and not about conflict framing and the like. In a study quite similar in approach to this one, Hollanders & Vliegenthart (2011) investigated the empirical relationship between the real economy, consumer confidence and economic news coverage in national newspapers in the Netherlands during the period 1990 - 2008. They found that media-attention for economic developments causes consumer confidence, with more negative news decreasing consumer confidence. Doms & Morin (2004) showed as well that there are periods when reporting on the economy has not been consistent with actual economic events. As a consequence, there are times during which consumer sentiment is driven away from what economic fundamentals would suggest. In the case of economic news such negative consumer sentiment and lower confidence adds to a negative spiral of declining demand, contracting production, soaring unemployment and even lower consumer confidence. Based on these findings and the earlier discussed studies on news coverage versus real-life trends of Iyengar (1985) and Soroka (2012), it is assumed that the Dutch newspapers’ coverage of the European Union is more negative than the real-world trends. Therefore hypothesis 2 is formulated:
!
H2: Dutch newspapers tend to cover the European Union in a different way than real-world trends would suggest
!
In order to test this hypothesis the real-world trends have to be defined. First of all, economic indicators are often used to describe and compare developments in the world. As seen in the studies of Iyengar (1985) and Soroka (2012), unemployment rate is an example of such an economic indicator. In their research on media coverage versus real-world trends, Goidel & Langley (1995) include - besides unemployment - Gross Domestic Product (GDP) and inflation as well as economic indicators. These three indicators are often defined as the three key economic indicators. However, since this paper is not only focused on economic indicators it is chosen to incorporate just GDP and unemployment rate as real-world indicators. Therefore, the following subhypotheses are formulated.
!
H2a: Dutch newspapers tend to cover the European Union in a different way than GDP rates in the EU would suggest
!
H2b: Dutch newspapers tend to cover the European Union in a different way than u n e m p l o y m e n t rates in the EU would suggest
Besides economic indicators, there are several other ways to derive the real-world status of (countries of) the European Union. The Organization for Economic Co-operation and Development (OECD) launched the Your Better Life Index (BLI) in May 2011. This tool includes 11 dimensions of well-being, which forms an interesting overall index of the real-world status of welfare in countries. However, since the data is only available since 2011 this index is not suitable for the discovering long term trends. It is decided to zoom in on two key elements of the index: health and education. Both indicators are often described by expected (life and education) years. The following sub-hypotheses related to health and education are formulated
H2c: Dutch newspapers tend to cover the European Union in a different way than educational expectancy rates in the EU would suggest
!
H2d: Dutch newspapers tend to cover the European Union in a different way than life expectancy rates in the EU would suggest
!
The study of Doms & Morin (2004) showed that more negative media coverage of the economy than real-world trends would suggest, leads to a negative consumer sentiment and lower confidence and in the end can contribute to a lower demand, contracting production, soaring unemployment and even lower consumer confidence. When applying these findings to the European Union context this would mean that more negative news causes lower confidence in the EU. So, the more negative the evaluations of the EU a citizen is exposed to, the more likely one is to cast a vote for an Eurosceptic party, as van Spanje & de Vreese (2012) showed, or will not vote at all. On the contrary, it can as well be said that the more visible and the more positive the tone towards a given party is in the news media, the more voters are inclined to vote for this party (Hopmann, 2010). Besides to actual voting behavior and turnout, this applies as well to the trust citizens have in the European Union. To figure out whether the Dutch newspapers influence the Dutch public opinion about the EU or whether it just reflects the attitude of the Dutch population towards the European Union, hypothesis 3 is formulated:
H3: Dutch newspapers tend to cover the European Union in a different way than public opinion polls about the EU would suggest
!
In the next section the methods that will be used to analyze how negative the Dutch news media coverage of the European Union is related to the real-world trends and public opinion in Europe will be described.
!
!
!
Methodology Media coverage
To asses the news coverage of the European Union by Dutch newspapers, a computer-assisted content analysis is conducted on the 4 main Dutch daily (paid) newspapers: NRC Handelsblad, De Volkskrant, De Telegraaf and the financial newspaper het Financieele Dagblad. The relevant articles are selected by using LexisNexis’ database of legal and journalistic documents. There are differences in the available time-periods of the newspapers (NRC Handelsblad since 1990, de Volkskrant since 1995, de Telegraaf since 1999 and het Financieele dagblad since 1994). Since it is hard to find all data for the real-world trends in the European Union from 1990 on and given the fact that de Telegraaf articles are only available since 1999, it is decided that the analyzed time-period is from January 1st, 1999, until May 21, 2014, which is the day before the European elections in the Netherlands.
The news articles are selected based on the criteria that they contain the keywords ‚European Union’, or ‚European Election’, or ‚European Commission’ or ‚European Parliament’ or ‚European Council’ (in Dutch: ‚Europese Unie’ of ‚Europese verkiezingen’ of ‚Europese Commissie’ of ‚Europees Parlement’ of ‚Europese Raad’), which leads to 97.310articles to analyze in total. The programming language Python is used for analyzing these articles by an automated content analysis.
!
Tone of coverage of the European Union
The extent to which Dutch newspapers cover the European Union in a sceptical or negative way is measured by filtering the coverage by keywords that belong to a negative sentiment. There are several examples of such a sentiment analysis of news coverage. Young & Soroka (2012) for example developed a Lexicoder Sentiment Dictionary (LSD). Negative words they use are ‚disappointment’, ‚neglect’, anxiety’, aggression’ and ‚sadness’ for example, while a positive sentiment contain words such as ‚benevolence’, ‚respect’ and ‚intelligence’.
Since this paper focuses solely on the Dutch news coverage of the European Union a sentiment lexicon for the Dutch language is needed. Atteveldt et al. (2008) provided a baseline for Dutch sentiment analysis research by applying sentiment analysis techniques to the Dutch language, using the Dutch thesaurus of Brouwers (1989). Furthermore, Jijkoun & Hofmann (2009) created a Dutch lexicon, which contains 2,836 negative polarity and 1,628 positive polarity words. Their evaluation based on a manual comparative assessment of the words used, showed that the method
they used achieved an accuracy of 0.82 at the top 3,000 negative words, and 0.62 at the top 3,000 positive words.
De Smedt & Daelemans (2012) developed - as part of the Pattern natural language processing software package for Python - an open source subjectivity lexicon for Dutch adjectives. They extracted adjectives from online Dutch book reviews and manually annotated them (by 7 annotators) for polarity (positive vs. negative), subjectivity (objective vs subjective) and intensity strength (e.g. good vs incredibly good). This resulted in a lexicon of 1,044 adjectives, containing 48% positive, 36% negative and 16% neutral assessments. The reliability in ratings given by the different annotators is measured by using Fleiss’ kappa. The Fleiss’ kappa for polarity is quite high (k=0.63). Since this lexicon is ready to the last gaiter button to be applied to the programming language Python, de Smedt & Daeleman’s (2012) lexicon will be used in this study for measuring polarity.
!
Real-world trends
As state before, this study focuses on the following 4 key indicators to measure real-life developments: GDP Growth Rate, Unemployment Rate, Educational Expectancy Rate and the Life Expectancy Rate. While the first two often used indicators provide information about the economic situation, the others give a (limited) insight into countries’ educational and health level. The data for all indicators is available on Eurostat (2014), the statistical agency of the European Commission.
!
Public opinion
The public opinion of EU citizens over time is monitored by the EU Barometer. Again this public opinion can be divided in several important categories, such as opinion about the Euro, about the enlargement, about health and about the European Parliament. These categories consist of different measurements of public opinion such as the extent to which they feel informed about it, the way they support it and their awareness. Since the focus here is on trust in the European Union in general the average trust in the different European institutions will be the measurement of citizens’ public opinion about the EU in this study. Eurobarometer (2013) measures the trust of European citizens in nine different institutions: European Parliament, European Commission, Council of Ministers, EU Court of Justice, European Ombudsman, European Central Bank, EU Court of Auditors, Committee of Regions and Social & Economic Committee. For some unclear reasons Eurobarometer measures only 5 of them since 2008. The average of the trust in these institutions serves as indicator of the trust in the European Union in this study.
Results
In the first part of this section the real-world trends and the public opinion in the EU over the past 15 years will be analyzed. When taking into account the real-world trends since 1999 it is important to stress out that the focus will not only be on the trends of the then 15 EU member states, since real-world developments might be based on the impact of enlargement of the European Union. Therefore not only the trends of the 15 member states of the 90’s will be taken into consideration, but the rates of all 28 member states since 2013. For the case of the public opinion, one is dependent of the Eurobarometer. The European Commission conducts this survey only among member states. Therefore it is not possible to analyze the public opinion of all 28 member states from 1999 on. However, concerning the public opinion the Dutch public opinion is the most important for the purpose of this research.
In the second part the results of the content-analysis of Dutch newspapers’ coverage of the European Union will be presented.
!
GDP Growth Rate
When looking at the average GDP growth rate in the 28 member states since 1999, only the years 2009 (-4,5) and 2012 (-0,4) have negative figures, which means the GDP grew in all other 13 years. The graph in figure 1 illustrates this trend. These results apply to the Netherlands as well, the country saw a decrease in its GDP of -3,7 in 2009 and -1,2 in 2012. However, the Netherlands had a negative figure in the year 2013 as well: -0,8. (Eurostat Real GDP Growth Rate Volume, 2014).
!
Unemployment Rate
When looking at the unemployment rates in the European Union over time (Eurostat Unemployment rate by sex and age groups, 2014), it can be noticed that the unemployment decreased from 1999 on from around 9.4% to 6.7% in the beginning of 2008, with a small increase between 2001 and 2004. After 2008 the rates increased again due to the Eurocrisis to 10.9% in 2013. For the most recent years the same trends are visible in the individual case of the Netherlands, one can see an increase from 3% in 2008 to 7,3% in the beginning of 2014. However, the Netherlands saw a bigger increase in unemployment rates between 2001 and 2005. Figure 2 illustrates these trends.
!
!
!
!
Figure 1. GDP Growth Rate
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
Figure 1. Unemployment Rate!
!
!
Educational Expectancy Rate
Concerning the educational level in the European Union one can consider the rate of educational expectancy. This rate corresponds to the expected years of education one has over a lifetime. Since early school leavers face a higher risk of social exclusion and poverty the world in general and the EU in particular strives to reduce the number of young people in this situation and thus encourage a high educational expectancy rate. The school expectancy data of Eurostat (2014) in figure 3 shows that the expectancy rate in the European Union overall increased with nearly one more educational year in the last 15 years (from 16.8 to 17.6). 2004 and 2006 are the only years in which there was a small decrease in expectancy. The Netherlands in particular saw an overall increase from 17.3 to 19.1 in 2012. 2002 was the only year the educational expectancy rate in the Netherlands dropped down a little bit.
!
Figure 3. Educational Expectancy Rate
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
Life Expectancy Rate
The World Health Organization monitors the life expectancy of each country and region over time, Eurostat uses the figures of this organization to report the expected age rate of its member states. Given the fact of for example medical developments and lower infant mortality rates, it is not strange that life expectancy rates did increase every year. When looking at the European Union, one finds out that in 2002 the expected age was 77,2 and in 2012 (the most recent figures) 79,6 years (Eurostat, Life expectancy by age and sex, 2014). It has to be said that for the years 1999, 2000 and 2001 only figures for 17 Euro area member states are available. From 2002 on the life expectancy rates of the other 11 countries are incorporated. Since it makes a huge difference whether one looks at these 17 or 28 countries in terms of life expectancy, it is chosen to look at the figures for the whole EU from 2002 on. For the specific case of the Netherlands the figures since 1999 are available and show that the expected ages at birth raised from 77,4 in 1999 to and 80,5 in 2012. The only thing that strikes here is that the rate decreased from 80,6 in 2011 to 80,5 in 2012. Figure 4 shows the trends in life expectancy for both the EU and the Netherlands between
!
Figure 4. Life Expectancy Rate
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
Public Opinion
Data of the Eurobarometer (2013) shows that the amount of trust in EU institutions fluctuated between 1999 (61% - 68% tend to trust) and 2009 (40% - 47%) with a peak in 2009 (64% - 74%). However, after 2009 - the first year of real Eurocrisis - the trust decreased tremendously, resulting in a recent 40% - 47% of trust in EU institution in 2013. Figure 5 illustrates these developments.
!
Figure 5. Eurobarometer - Trust in the European Union
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
News coverageAfter filtering out the invalid results, a total of 97.310 news articles have been analyzed. ‚Het Financieele Dagblad’ (33,2%) and ‚NRC’ (31,9%) are responsible for the most news articles about the European institutions or elections. ‚De Volkrant’ (21,4%) and ‚de Telegraaf’ (13,6%) paid a little less attention to theses topics. When taking into account the different Dutch newspapers, it might be interesting to take a look at the differences in mean sentiment between these titles. The most striking result to emerge from these data is that the right-wing, conservative newspaper ‚de Telegraaf’ scored the highest mean sentiment (M=,0387 SD=,1054), followed by financial
newspaper ‚het Financieele Dagblad’ (M=,0365 SD=,0987). These newspapers were the only newspapers that know an ultimate minimum sentiment of -1,0. ‚De Volkskrant’ (M=,0266 SD=, 0989) ‚NRC’ (M=,0239 SD=,0958) were the the most negative newspapers. However, the results show that all newspapers together had a relatively positive mean sentiment (M=,0307 SD=,0990). The highest sentiment was 0,9 and was reached by both ‚het Financieele Dagblad’ and ‚de Volkskrant’. Finally, a linear regression analysis shows that only between ‚de Volkskrant’ and ‚de Telegraaf’ happens to be a significant, negative effect (p=< .01, β=-.025, t=-2,848). When ‚de Volkskrant’s tone of media coverage of the EU increases, the tone of the media coverage of the EU of ‚de Telegraaf’ decreases and thus gets slightly more negative.
When focussing on the development of the sentiment of the news articles over time, as can be seen in figure 6, it is apparent that the most significant decrease takes place between 2000 and 2001. While from then on until 2007 the sentiment increases. On the other hand, the period from 2008 until the end of 2014 is characterized by moderate and steady development. It has to be said that due to the fact that 97.310 articles are analyzed, the differences in sentiment are small. While the most positive sentiment is .90 and the most negative sentiment -1.0, the scale of figure 6 in which one sees the mean differences is mainly between .02 and .04. This illustrates how small the differences really are. Besides, it shows that the assumption that the sentiment of Dutch news media coverage of the EU is overall small is not true. Therefore Hypothesis 1 is rejected.
!
Figure 6. Sentiment Dutch newspapers’ coverage of EU
!
!
!
!
!
!
!
!
!
!
!
!
!
Comparison
In the last part of this section the different found trends are analyzed and compared with each other. In order to do illustrate the difference in one graph all scales have to be the same. Therefore all trends are expressed in annual percentages of the total growth. By doing this, the GDP rates should be negative at 2 or 3 points in time. However, since this tremendous influences the graph, it is chosen to have 0 as a minimum. Furthermore, are the unemployment rates mirrored in order to describe the same positive and negative trends. Of course, the real values of the lines disappear in this graph, however, since this paper is mainly interested in the trends it is suitable. The first graph in figure 7 shows the European trends.
!
Figure 7. European trends
From this graph it is visible that sentiment and GDP seem to follow the same pattern. Since there are too few values, a linear regression analysis cannot be conducted. Instead it is chosen to measure the Pearson Correlation coefficients. It turns out that the correlation between GDP and
sentiment is not significant. The only significant correlation is between sentiment and Life Expectancy (p=< .05, r=.574).
The same applies for the Dutch trends in figure 8. Only the correlation between sentiment and Life Expectancy is significant (p=< .05, r=.541). Therefore, only sub-hypothesis H2d is rejected and thus H2a, H2b, H2c and H3 are accepted.
!
Figure 8. Dutch trends
!
!
!
!
!
!
!
!
Discussion
In the first part of this section conclusions will be drawn from the presented results. In the second part limitations of this study will be discussed as well as the possible implications for future research.
!
Conclusions
First of all, when looking at the formulated hypotheses, it strikes that the first assumption that the tone of news media coverage of the European Union has a negative tendency, is false. When focussing on individual articles one finds of course some negative sentiments. However, when smoothing the results and just zooming in on the mean sentiment each year, one cannot conclude that Dutch newspapers reported overall negative about the EU between 1999 and 2014. These findings are not in conflict with the studies that show that there is a lot of negative media coverage of the EU (Hawkins, 2012, de Vreese, 2007, de Vreese et al., 2007) and the ones that prove that negative news has more effect than positive news (Kleinnijenhuis et al.,2006, Soroka, 2006), however, it does not seem to be the case that the consequence of its findings are that negative sentiments prevail in the Dutch newspapers’ coverage of the EU. A similar sentiment analysis (van Dam, 2014), focused on the Dutch newspapers’ coverage of the European Parliament Election in 2014, neither indicated that the tone in the media coverage was negative. Therefore, one should be careful by stating that media in general are Eurosceptic, fuel the lack of trust in the European Union or even are responsible for its democratic deficit.
Secondly, the results of this study prove that - as assumed - the Dutch media coverage of the European Union between 1999 and 2014 does not seem to reflect real-world trends. GDP Growth Rate (although showing more or less the same trend), Unemployment Rate and Educational Expectancy Rate developed in other ways than the sentiment of the media coverage. The only exception is the Life Expectancy Rate, which turned out to correlate with the coverage of the Dutch newspapers. One explanation for this finding could be that the expected age one has increases very slowly and its trend is therefore characterized by a steady course, which can as well be noticed in the overall development of the sentiment from 2002 on. The increase of the Educational Expectancy Rate was small as well and this trend developed steady too. However, one can notice two small peaks in this trend, that is why there is no significant correlation between media coverage and Educational Expectancy.
Finally, it turned out that the public opinion - the trust people have in European institutions - neither is reflected in the coverage of the European Union by Dutch newspapers. While public trust
in the European Union suffers major decline from the Eurocrisis, started in 2008, the sentiment in the media coverage of the EU does not seem to have a similar development. Therefore, it can be concluded that as well as most real-life trends, the public opinion about the European Union is not reflected in the coverage of the Dutch newspapers.
To conclude, overall this study showed that - contrary to earlier expectations and presumptions - Dutch newspapers tend not to cover the European Union in a notable negative way. So to answer the header of this paper; there does not seem to be a disproportionate degree of Euroscepticism and negative sentiment in the media coverage of the European Union. One question that therefore needs to be asked is whether media are that Eurosceptic in general. Besides, the findings of this sentiment analysis demonstrate that the media coverage of the European Union does not seem to correlate with most real-world trends. This leads to the following answer on the research question; Dutch newspapers seem to reflect real-life trends and public opinion to a very small extent. On the one hand, there does not seem to be a reflection of GDP Growth Rate, Unemployment Rate, Educational Expectancy Rate and public opinion trends in the media coverage. On the other hand, the media do seem to follow the same trend as the Life Expectancy Rate. As Soroka (pp.30, 2003) suggested, it remains the question whether media coverage should simply be the product of these real-world indicators or whether it should aim for its own influence on the public, so-called ‚media effects’. In line with this issue, this study adds that the public opinion does neither seem to be reflected by the media coverage.
!
Limitations
Although this computer-assisted sentiment analysis enables one to analyze 97.310 news articles, this method and study suffer several major drawbacks. First of all, when it comes to measuring the polarity of an article, this method relies on the lexicon of De Smedt & Daelemans (2012) and the way the computer analyzes the article. Although their lexicon proved to be reliable and the computer just executes the commands, there is no room for different interpretations of articles. An article in which, for example, the European Union is evaluated slightly positive but in which the city of Brussels or the public transport to get to the European Union institutions are described by using an abundance of negative adjectives, would score negatively. Another limitation of this method is the fact that De Smedt & Daelemans’s lexicon contains 1,044 adjectives, of which are 48% positive, 36% negative and 16% neutral assessments. This overrepresentation of positive adjectives might influence the objectivity of measuring polarity.
Furthermore, this study on media coverage of the European Union focuses in the first place only on Dutch media. Besides, it only analyzed the newspaper coverage, not the television, radio or online media coverage. Therefore, one should be very cautious when generalizing the results of this study to other member states and other media platforms.
Another limitation of this research is the fact that it only analyzed trends of the last 15 years. Of course, the biggest enlargements of the European Union took place in this period, which is therefore an interesting time period to analyze. However, in order to compare trends one should look at long term trends and one can raise the question whether the results would be the same if this paper would focus on trends in the last 30 years.
Finally, it is important to bear in mind the fact that real-world trends in this study are defined by just 4 indicators. It is possible that when replacing these indicators with other indicators, the sentiment in news coverage of the European Union does reflect all (or some) of these real-world trends.
!
Implications for future work
Despite its limitations, the findings of this study may help us to understand and to consider our views of the role of the media in the public discourse about the European Union. Future studies on different types of media outlets, in different member states and on different time periods are required to establish the findings that the media do not report disproportionate negative about the European Union. And if this is true, how can claims of de Vreese et al. (2007), who state that Euroscepticism is „a product of the diet of information one consumes about European affairs” or Statham (2006), who says that the media is the one to blame for the democratic deficit of the European Union, then be explained? And should the media reflect real-life trends in the European Union more in order to inform their public properly? These are research questions that might be interesting for future work and will provide us more insights in political communications related to the European Union.
!
References
Van Atteveldt, W., Kleinnijenhuis, J., Ruigrok, N., & Schlobach, S. (2008). Good news or bad news? Conducting sentiment analysis on Dutch text to distinguish between positive and negative relations. Journal of Information Technology & Politics, 5(1), 73-94.
Behr, R. L., & Iyengar, S. (1985). Television news, real-world cues, and changes in the public agenda. Public Opinion Quarterly, 49(1), 38-57.
Brouwers, L. (1989). Het juiste woord: Standaard betekeniswoordenboek der Nederlandse taal (7th edition; ed. f. claes). Antwerpen, Belgium: Standaar Uitgeverij.
Doms, M. E., & Morin, N. (2004). Consumer sentiment, the economy, and the news media. Divisions of Research & Statistics and Monetary Affairs, Federal Reserve Board.
de Smedt, T. & Daelemans, W. (2012). Vreselijk mooi! (terribly beautiful): A subjectivity lexicon for Dutch adjectives. Proceedings of the 8th Language Resources and Evaluation
Conference (LREC’12), 3568 - 3572.
de Vreese, C. H. (2007). A spiral of Euroscepticism: The media’s fault?. Acta Politica, 42(2), 271-286.
Eurobarometer (2013). Public Opinion in the European Union. Retrieved from: http://ec.europa.eu/public_opinion/archives/eb/eb79/eb79_first_en.pdf
European Parliament (2009). Retrieved from: http://www.europarl.europa.eu/aboutparliament/en/ 000cdcd9d4/Turnout-(1979-2009).html
Eurostat (2014). School Expectancy. Retrieved from: http://epp.eurostat.ec.europa.eu/portal/page/ portal/education/data/main_tables
Eurostat (2014). Real GDP Growth Rate. Retrieved from: http://epp.eurostat.ec.europa.eu/tgm/ table.do?tab=table&init=1&plugin=1&language=en&pcode=tec00115
Eurostat (2014). Harmonised Unemployment Rate by Sex. Retrieved from: http:// epp.eurostat.ec.europa.eu/tgm/table.do?
tab=table&init=1&plugin=1&language=en&pcode=tec00114 Eurostat (2014). Life Expectancy by Age and Sex. Retrieved from:
http://appsso.eurostat.ec.europa.eu/nui/submitViewTableAction.do77
Goidel, R. K., & Langley, R. E. (1995). Media coverage of the economy and aggregate economic evaluations: Uncovering evidence of indirect media effects. Political Research Quarterly, 48(2), 313-328.
Hawkins, B. (2012). Media coverage of the European Union is key to understanding eurosceptic attitudes within the UK. LSE European Politics and Policy (EUROPP) Blog.
Hollanders, D., & Vliegenthart, R. (2011). The influence of negative newspaper coverage on consumer confidence: The Dutch case. Journal of Economic Psychology, 32(3), 367-373. Hopmann, D. N., Vliegenthart, R., De Vreese, C., & Albæk, E. (2010). Effects of election news
coverage: How visibility and tone influence party choice. Political Communication, 27(4), 389-405.
Jijkoun, V., & Hofmann, K. (2009, March). Generating a non-english subjectivity lexicon: Relations that matter. In Proceedings of the 12th Conference of the European Chapter of the
Association for Computational Linguistics (pp. 398-405). Association for Computational Linguistics.
Kleinnijenhuis, J., van Hoof, A. M., & Oegema, D. (2006). Negative news and the sleeper effect of distrust. The Harvard International Journal of Press/Politics, 11(2), 86-104.
Scheufele, D. A., & Tewksbury, D. (2007). Framing, agenda setting, and priming: The evolution of three media effects models. Journal of communication, 57(1), 9-20.
Schuck, A. R., & De Vreese, C. H. (2006). Between Risk and Opportunity News Framing and its Effects on Public Support for EU Enlargement. European Journal of Communication, 21(1), 5-32.
Shoemaker, P. J., & Reese, S. D. (1996). Mediating the message. White Plains, NY: Longman. Soroka, S. N. (2003). Media, public opinion, and foreign policy. The International Journal of Press/
Politics, 8(1), 27-48.
Soroka, S. N. (2006). Good news and bad news: Asymmetric responses to economic information. Journal of Politics, 68(2), 372-385.
Soroka, S. N. (2012). The gatekeeping function: Distributions of information in media and the real world. The Journal of Politics, 74(02), 514-528.
van Dam, M. (2014). News coverage of the elections for the European Parliament in Dutch national newspapers. An automated sentiment analysis approach (Unpublished master thesis).
University of Amsterdam, Amsterdam.
van Spanje, J. and De Vreese, C. H. , 2012-05-24 "Effects of News Media Coverage on Eurosceptic Voting: Evidence From the 2009 European Parliamentary Elections" Paper presented at the annual meeting of the International Communication Association, Sheraton Phoenix
Statham, P. (2006). Political journalism and Europeanization: pressing Europe?. Centre for European Political Communications.
Young, L., & Soroka, S. (2012). Affective news: The automated coding of sentiment in political texts. Political Communication, 29(2), 205-231.
