Looking at Ankara from Strasbourg: An Empirical Assessment of meps' Voting on and Framing of Turkey
STEFANO BRAGHIROLI
University of Tartu, Estonia
this article examines the way in which the Members of the European Parliament (meps) frame Turkey and how this affects their voting stance towards Ankara in the parliamentary debates. Recent studies (Baldwin and Widgren 2005; Braghiroli 2012; Canan-Sokullu 2011) have demonstrated that the debate on Turkey's European Union (eu) membership produces a very divisive impact on the voting dynamics and voting alignments in the European Parliament (e p) in the light of its national and political significance. The parliamentary positions on the 'Turkey discourse' range from enthusiastic support to open Turkophobia. What is even more striking is the wide variety of individual positions generally identifiable within the same political/ideological area. To what extent are meps' different perceptions and representations of Turkey reflected in the way they vote when Turkey is at stake in the e p? And, what is the impact of this state of things on groups' internal cohesion? This study addresses these two fundamental questions using two different sources of data. Elite survey data is used in order to capture meps' perceptions of Turkey, while meps' voting behavior is assessed in the light of the expressed votes.
Key Words: European Parliament; Turkish membership; perceptions; voting behavior
INTRODUCTION When the European Council decided unanimously to start the accession negotiations with Turkey in December 2004, the decision was confirmed by the ep, with 407 votes in favor and 262 against.1
Despite the unanimous pledge of the eu governments, their commitment lost momentum within a few months and Turkey was in-
volume 6 | 2013 | number 1 | 3—80
Stefano Braghiroli
creasingly confronted with open or implicit opposition of a number of member states and key stakeholders. According to the Independent Commission on Turkey (2009), 'in several countries such public discourse coincided with elections, giving the impression that domes-[4] tic political calculations were involved.' At the same time, the functional use of the 'Turkey discourse' also gained ground among the mainstream parties, both at the national and ep level. As witnessed by Nicholas Sarkozy and Angela Merkel's recent negative remarks towards Turkey's accession, ideological and/or functional opposition towards Ankara's eu ambitions has increasingly emerged as a practical short-cut to convey popular concerns about immigration, unemployment, multiculturalism, and Islam (McLaren 2007). Moreover, the functional use of the 'Turkey discourse' seems no longer a monopoly of the radical and populist movements, but it has been increasingly interiorized by the mainstream conservative and moderate political forces.
The growing skepticism is reflected by the new Negotiating Framework formally agreed in the Luxembourg European Council and endorsed by the ep in 2006. While Turkey's accession is defined as 'the shared objective of the negotiations,' the negotiations are presented as 'an open-ended process, the outcome of which cannot be guaranteed beforehand' (European Commission 2005).
This article examines the way in which the meps frame Turkey and how this affects their voting stance towards Ankara in the parliamentary debates. Recent studies (Baldwin and Widgren 2005; Braghiroli 2012; Canan-Sokullu 2011) have demonstrated that the debate on Turkey's eu membership produces a very divisive impact on the voting dynamics and voting alignments in the ep, in the light of its national and political significance.
The present article proceeds as follows. Following the definition of the ep setting and of the actors involved in the first section, the methodology adopted is discussed in the second section. In the third and fourth section, the analysis of the expressed votes and of the declared opinions is performed; while in the fifth section a comparative exercise is presented. In the concluding sections, some boarder considerations are discussed in the light of the results.
ijems
Looking at Ankara from Strasbourg
DEFINING THE sETTING AND THE AdoRs Following the formal redefinition of Ankara's accession prospects, some eu governments openly argued in favor of the cooperation arrangements clearly alternative to the full membership, such as a 'privileged partnership.' They emphasized the exceptionality of the Turkish [5] case when compared to the other waves of enlargement.
So far, only few, non-mainstream, ep party groups are openly against Ankara's eu membership, while the majority of the political forces in the ep formally support it, at least on paper. However, as time passes and the negotiation outcome becomes more unpredictable, the 'Turkey discourse' appears increasingly hostage of partisanship with the European center-left emerging as the herald of a pro-Turkey stance, while a growing number of conservative meps appear increasingly tempted to adopt a more populist approach in order to attract protest vote in an electoral perspective (Braghiroli 2012).
Parliamentary support and opposition to Ankara's European ambitions range between functional/interest-based and ideal/ideological stances. The pragmatic stance appears more frequent among the mainstream parties, while the identitarian approach characterizes more extreme and protest parties. A clear example of the functional opposition side is provided by a recent report commissioned by the German Christian-Democrats warning against a 'too big, too poor [Turkey], with too dangerous borders and insufficiently "European" to join the Union.'2
In the light of the salience of the issue at stake, the wide range of conflicting positions seems to have a very relevant disruptive potential on the parliament's voting dynamics when the 'Turkey discourse' is at stake. To what extent are the parliamentary voting dynamics on Turkey a function of meps' different perceptions and representations of Turkey? So far, no clear answer has been given to this very basic question.
This study represents one of the few empirical attempts to look at the dynamics of the debate on Turkey from a parliamentary perspective involving meps' perception-based framing of Turkey. The scholarly attention on the 'Turkey discourse' has mainly been focused on the eu's executive institutions (the Council and the Commission), while the ep
volume 6 | 2013 | number 1
Stefano Braghiroli
has been generally depicted as a sort of 'irrelevant other.' However, as LaGro and Jorgensen (2007) warn
[...] the institutions to decide on the faith of Turkey will not be na-[6] tional parliaments on the recommendation of their respective governments, but the peoples of Europe and, of course, one must not forget, the European Parliament, which is gaining power exponentially within the e u institutions.
In this respect, the ep represents the only eu institution directly legitimized by citizens' vote. It is not only the sole legitimate representative of the people of Europe, but, given its multi-national nature and ideological composition, it is also more likely to reflect their attitude in voting dynamics.
As the analysis is addressing a relatively unexplored ground, this study is conceived as an exploratory analysis towards a more precise understanding of the relationship between meps' perceptions and voting behavior in the specific case of the 'Turkey discourse.' For this reason, this study will not propose a formal set of hypotheses to test.
methodology Two different sources of data have been used in the analysis. In order to capture the meps' perception of Turkey, a feeling thermometer question included in the 2008—10 waves of the European Elite Survey/Transatlantic Trends Leaders3 has been used, recoded according to a 0 (lowest level of sympathy) — 1 (highest level of sympathy) scale.
When it comes to the mep' voting behavior, the available roll-call votes (rcvs) held on Turkey-related issues between 2009 and 20124 have been collected. The procedure that has been adopted to score the meps' votes according to their connotation towards Turkey implies three successive steps. First, for every bill considered, the sections concerning Turkey and Turkish membership are recorded. Second, every vote is assigned a score in the light of the connotation it gives to Turkey.5 Third, a final measure is calculated for every mep on the basis of each legislator's valid votes portraying the meps' overall voting position when Turkey and Turkish membership are at stake. Therefore,
ijems
Looking at Ankara from Strasbourg
0.00—0.20
113%
0.21—0.40	30%
0.41—0.60	19%
0.61—0.80	14%
0.80—1.00	23%
figure 1 Distribution of meps' individual voting scores
if mep 'x' supports a piece of legislation favorable to Turkey or opposes one labeled as negative towards Turkey he/she gets score 1, vice versa he/she gets score 0. In case of abstention, he/she gets score 0.5. The final measure represents the mep's average score and ranges from 0 (highest level of Turkey-friendly voting behavior) to 1 (lowest level of Turkey-friendly voting behavior).
The final analysis will be conducted by crossing the meps' perception of Turkey and their voting behavior at an individual level6 and by assessing the level of compliance between the two. This will allow us to understand whether the meps tend to vote according to their preferences when it comes to the 'Turkey discourse' or whether they are driven in one way or the other by domestic or parliamentary pressures and behave pragmatically.
[7]
THE VOTING SIDE In the following sections, the two analytical dimensions considered will be discussed. The measures of homogeneity and cohesiveness will be calculated on the basis of the meps' partisan affiliation and nationality.
In total, nine votes were included in the analysis respecting the 75:25 ratio; five were coded as positive/favorable towards Turkey and four as negative/unfavorable. The rcvs analyzed are all related to the meps' scrutiny of the Commission's annual progress reports.
Figure 1 charts the distribution of the voting scores among the 735 meps included in the research. The votes clearly do not appear normally distributed. If we look at the two polar voting categories, respectively expressing the highest level of negative votes towards Turkey (0—0.20) and the highest level of positive votes (0,81—1), the chart shows that the latter is by far the most frequent category with 174 meps, which constitutes more than 24% of the total. In this respect, those who
volume 6 | 2013 | number 1
Stefano Braghiroli
expressed the most negative voting stance represent the smallest of the five categories with 98 meps (13%).
Looking at the general trend, what emerges is a slight prevalence of positive scores (given an average ep score of 0.53), while the meps ex-[8] pressing a 'moderately negative' voting attitude towards Turkey (0.21— 0.40) represent the modal group with 220 meps (30%). The general picture seems fairly balanced and the gap between the 'negative' and the 'positive' group seems very narrow, also considering the 143 meps that fall in the median category (19%).
Figure 2 represents the average voting scores by country. Also in the light of the great domestic political salience of Turkey's e u accession in many member states, the results presented display a significant level of variance among the national delegations. A 50% gap emerges between the delegation expressing the most negative connotation and the delegation expressing the most positive one.
Against an average ep score of 0.53 (denoting a fair balance of negative and positive votes); the member states representing the lowest rank score are Cyprus with a score of 0.2 and Greece with a score of 0.27 Greece. It is worth noting that if we ignore these two outliers, the gap narrows to 33%. While it is no surprise that Nicosia and, to a lesser extent, Athens' delegations present a cold voting stance towards Turkey, more puzzling are the other low scoring delegations. In total, only seven out of 27 delegations are characterized by a majority of negatively expressed votes. Among them it is worth mentioning the Austrian (0.37), Hungarian (0.42), Dutch and Polish (0.46), and French (0.49) delegations.
While in the case of the Austrian, French, and Dutch meps the cold voting stance seems to reflect the long lasting negative bias towards Ankara's membership often fuelled by the presence of relevant migrant communities from Turkey (McLaren 2007), more confusing are the cases of the Hungarian and Polish delegations. In this case, the average negative factors are possibly determined by incidental factors that will be possibly clarified by the analysis of the inter-group variance.
Interestingly, the German delegation (0.55) appears not only characterized by a majority of positively expressed votes, but it also scores
ijems
Looking at Ankara from Strasbourg
mt	0.70
se	0.67
es	0.64
DK	0.62
ro	0.61
cz	0.61 [9]
si	0.60
ee	0.60
f1	0.59
ie	0.57
uk	0.56
lt	0.56
sK	0.55
de	0.55
pT	0.54
lv	0.54
e p*	0.53
B G	0.52
iT	0.51
be	0.50
lu	0.49
f r	0.49
pl	0.46
nl	0.46
hu	0.42
aT	0.37
el	0.27
cy	0.20
figure 2 Distribution of average voting scores by national delegations (* average)
higher than the ep average. The Scandinavian meps and those from the Central and Eastern Europe (cee) express the most favorable voting stance towards Turkey, along with the Mediterranean Spaniards (0.64) and Maltese (0.7).
The political support of the Nordic countries, such as Sweden (0.67), Denmark (0.62), and Finland (0.59) to Ankara's European ambitions has been well documented in a number of studies (Adam and Moutos 2005; Muftuler-Bac and McLaren 2003) and our data seems to confirm the same Turkey-friendly stance in the voting dynamics of the Scandinavian delegations. However, the high scores of most of the cee delegations — Romanian and Czech (0.61), Slovenian and Estonian (0.6) meps — seem due to the well documented phenomenon of
volume 6 | 2013 | number 1
Stefano Braghiroli
s & D
I0.87
Greens/efa alde/adle ep average
I °.53
I0.56
I 0.71
ECR epp
|°.5I
l°.34
GUE-NGL EFD NI
I °.3I
| °.29
| °.22
figure 3 Distribution of average voting scores by group
enlargement solidarity (Falkner and Treib 2008; Rahman 2008; Zielonka
Figure 3 charts the average voting score distribution by political group. A number of recent studies (Hix and Noury 2009; Rasmussen 2008) have demonstrated that votes in the ep are generally expressed along the political lines, rather than the national ones. Other studies claim that the political groups in the ep also represent the main source of discipline when it comes to the meps' individual voting behavior as mirrored by the high level of cohesion in the Parliament (Hix 2002). In this respect, the results presented above appear very relevant.
A point that emerges clearly from the figure 3 is that the m e ps' voting stance towards Turkey seems to reflect a very evident left-right divide, thereby presenting a clear ideological/partisan connotation. Worth noting is that the range between the parliamentary group expressing the most negative stance and the group expressing the most positive one equals 75% and is therefore far larger than in the case of the national delegations discussed above. In this respect, it is useful to divide the political groups in the ep in three clusters. The right side of the political spectrum (including extreme right, Euroskeptic right, and moderate-conservative European People's Party7) presents the scores far below the ep average, thereby reflecting a majority of negatively expressed votes. The center of the spectrum — including liberal-democrats (alde) and democratic Euroskeptic affiliated to the group of the European Conservatives and Reformers (ecr) — presents the scores aligned to the ep average, thereby suggesting a combination of different voting options and a less ideological approach for the centrist groups. The left side of the political spectrum — including the social-

ijems
Looking at Ankara from Strasbourg
democrats (s & d) and the Greens — presents the highest scores and the highest level of Turkey-friendly votes.
Looking at the scores of the groups, two relevant exceptions seem to emerge with respect to the ideological characterization of the Turkey-related votes. In particular, the radical left and the democratic [11] Euroskeptics seem to present a relevant mismatch in this respect. The radical leftist group of the European United Left-Nordic Green Left (gue-ngl) presents extremely low voting scores (0.31) comparable to the radical and Euroskeptic right. This seems to be due to two specific factors. On the one hand, it is worth mentioning the long lasting support expressed by many constituent parties for the Kurdish cause, which is reflected by a widespread functional opposition towards Turkey and its alleged assimilation and repression campaigns (Gunes-Ayata 2003). On the other hand, another important factor that contributes to the explanation of the exceptionality of the group is represented by the key relevance of the Cypriot communist delegation within the gue-ngl. In this respect, despite the limited size, the Cypriot communists express the only head of government from the ranks of the gue-ngl, thereby making the Cyprus-issue a very sensitive one for the group.
In the case of the Euroskeptic ecr, the relatively high scores recorded seem due to their strong support for a faster enlargement strategy of the eu as way to make the Union more plural and to weaken its alleged federal character. In this respect, the conservatives' support for Turkish membership therefore seems functional.
the elite opinion side Figure 4 charts the distribution of perception scores among the 176 meps included in the ees/ttl surveys. In this case, the meps appear more normally distributed than in the case of the voting scores presented in the previous section. In terms of the connotation of Turkey, the declared opinions appear positively-oriented. In this case, while the two negative categories (0—0.20 and 0.21—0.40) account for 24% of the total, the percentage grows to 29% if we consider the two positive categories.
Moreover, what emerges as the most relevant difference in compar-
volume 6 | 2013 | number 1
Stefano Braghiroli
0.00—0.20 0.2I—0.40
0.41—0.60 0.61—0.80 0.80—1.00
I6%
47%
23%
7%
figure 4 Distribution of meps' individual scores in the feelings thermometer
ison to the distribution of the voting scores is that the modal group in the elite opinion distribution is represented by the central category (0.41—0.60) capturing neutral or moderate scores in the feelings thermometer and accounting for 47% of the total. In general, we can therefore say that not only do the declared opinions appear on average more normal than the expressed votes, but are also more moderate and less polarized.
Figure 5 charts the national delegations' average declared feelings towards Turkey and compares them with their average voting scores presented above. Only the national delegations with at least 10 interviewees were included in the computation in order to grant a fair degree of generalization. Moreover, for the same reason, the distributions presented have been weighted according to the relative size of each party in the respective national delegation.
If we compare the m e ps' image of Turkey with their actual voting scores in the 7 largest delegations included, no major mismatch seems to emerge. Moreover, all scores do not distance themselves too much from the average (0.55) and the gap between the most friendly delegation and the most negative one is much more narrow than in the case of the expressed votes, thereby equaling 14%. Interestingly, in most of the national cohorts, the gap between declared perceptions and expressed votes is of a few decimals. This is the case in the Spanish (+5%), Romanian (—3%), Italian (+2%), and French (+3%) delegations. In these cases, the m e ps' image of Turkey seems to almost perfectly reflect the way they vote.
Partial exceptions to this state of things are represented by the British, German, and Polish delegations. On one hand, in the case of the Polish delegation, the image of Turkey that representatives have in mind is more positive than the one that emerges from the voting scores. The positive mismatch emerged appears consistent and it equals 8%. On the other hand, in the German and British cases, the
XXXXXXXXXXXX^ ijems
Looking at Ankara from Strasbourg
c ■	0.54
Spam
°.59
Romania ^^^^^^^^^^^^^^^^^^^^^^^^^^^ °
0.58
0.53
EES average
0.55
Poland	046
0.54
T 1	O.5I
Italy	O.53
France
o.52 0.55
[13]
Germany	o 47
United Kingdom ^^^^^^^^^^^^^^^^^^^^^^^^^^^ 0.56 5	0.45
figure 5 Distribution of average scores in the feelings thermometer by national delegations (dark — voting scores, light — feelings thermometer)
meps' representation of Turkey is much more negative than their actual voting behavior. Moreover, in both cases their voting scores suggest a moderately-positive attitude (respectively 0.55 and 0.56), while their declared opinions highlight the existence of moderately-negative attitudes (respectively 0.47 and 0.45), thereby confirming the existence of a negative mismatch. The existence of relevant mismatches appears — among others — related to the high voting cohesion achieved within the major group, which seems to induce pragmatic, rather than idealistic behavioral styles in the affiliated meps.
Figure 6 charts party groups' average feelings towards Turkey and compares them with the voting scores presented in the previous section. Also in this case, the distributions presented have been weighted according to the relative size of each national party in the national delegation. Looking at the overall picture, what emerges is the fact that the inter-group variance is higher than in the case of opinion distribution by national delegations, but smaller than in the case of the vote-based analysis of the party groups.
Looking at the individual groups, the most significant negative mismatches are represented by the group of the European Social-democrats (—23%) and by the Greens (—17%). In this respect, the meps belonging to the groups that presented extremely high voting
volume 6 | 2013 | number 1
Stefano Braghiroli
scores appear to have more moderate feelings. Although — on average — they still present a very positive connotation towards Turkey, their positive attitudes appear more tempered than it appears from their voting stance.
[14] Interestingly, the biggest gap is represented by the positive mismatch registered among the meps belonging to the radical left, where the difference between declared opinions and expressed votes equals 27%. In the previous section, the study discussed the potential factors behind gue-ngl's extremely low score, such as the well-known concern for the Kurdish issue and the key role played by the small Cypriot delegation (not included in the ees/ttl sample). In this respect, many of the leftist meps appear to have far more moderate ideas than those expressed by the voting stance of their group, denoting a very pragmatic behavior. Other significant positive mismatches are represented by the epp group (+13%) and by the far-right non-attached meps who mark a difference of+17%, despite retaining a very negative stance towards Turkey. In this respect, shifting from expressed votes to declared opinions seems to implement the process of 'normalization of the extremes.'
The ideological diversity in many groups, witnessed by the mismatch between declared opinions and expressed votes, reflects the existence of frequent pragmatic behaviors as hypothesized above. In this respect, it seems that a relevant number of meps — if let free to act according to their preferences — would adopt a more positive or a more negative voting stance towards Turkey than the one sponsored by their party group. The emergence of a consistent gap proves the existence of a relevant group of meps who only partially follow their belief structure when voting, thereby prioritizing group's interests or other exogenous instances.
However, those with strong negative perception of Turkey appear less likely to behave pragmatically than those characterized by a positive percentage.
closing the circle In this final section, the results of our experiment are discussed, thereby crossing the meps' perceptions of Turkey and their voting behaviour
ijems
Looking at Ankara from Strasbourg
S & D GUE-NGL EES average Greens/efa alde/adle ep ECR
0.31
°-34
I 0.64
0.58
°.53 I °.55
0.54 0.56
0.48 I 0.47
o.5i
0.46
0.7i
0.87
[15]
0.39
0.29 0.26
figure 6 Distribution of average scores in the feelings thermometer by party groups (dark — voting scores, light — feelings thermometer)
0.22
at an individual level, after having explored them separately in the previous sections.
Figure 7 provides a graphic representation of the 87 meps' distribution along the two dimensions. In general, the trend emerged confirms that — as expected — the two dimensions are positively and significantly correlated. In this respect, as the meps' perception of Turkey shifts from unfavorable to favorable, their likelihood to support Turkey-friendly legislation and to oppose the unfavorable one is also supposed to increase. However, as proved by the slope of the interpolation line and by the ^-squared coefficient (0.204), the match appears imperfect and in many instances fairly weak. In particular, around 30% of the analyzed cases do not fall in the expected quadrants if we assume a positive relationship between perceptions and votes. The two unpre-dicted quadrants are marked in light grey in the figure.
This state of things seems to suggest that generally, the m e ps' image of Turkey is not the only and (often) not the strongest criterion according to which the legislators take their voting stance when the 'Turkey discourse' or Ankara's membership are at stake. The presence of a relevant number of meps in the unpredicted quadrants confirms
volume 6 | 2013 | number 1
Stefano Braghiroli
that meps' pragmatic behavior seems to play a very relevant role in the voting dynamics related to Turkey, as mirrored by the frequent mismatches highlighted in the previous section.
Interestingly, pragmatic behaviors are not equally distributed among the two deviant categories. In this respect, the meps who display a positive perception towards Turkey are much more likely to behave pragmatically — that is to support unfriendly legislation towards Ankara (see upper-left quadrant) — than those who display a negative perception towards Turkey. However, the latter appear much more unlikely to ignore their negative feelings and support friendly legislation towards Ankara (see right-lower quadrant). In other words, while Turkey's supporters tend to support the votes favorable to Ankara, but can accept voting pragmatically due to group's loyalty or national interest, the opponents of Turkey only very rarely vote against their beliefs and therefore appear more ideological and less pragmatic.
Does the mismatch observed have any divisive impact on the groups' voting cohesion? Is there any evident difference among the groups considered? In order to answer these questions it is worth looking at the groups' level of cohesion in the Turkey-related votes and to compare them with the average level of cohesion of the groups in the 7th e p (figure 7).
As expected, in most cases, despite the relevant mismatch between
ijems
Looking at Ankara from Strasbourg
registered perceptions and expressed votes, the level of cohesion of the groups does not seem to suffer from the gap. Particularly significant seems to be the disciplining potential of the group in the case of the radical left (gue-ngl) and of the social-democrats (s & d) that presented a mismatch equaling respectively +27% and —23%. In both [17] cases, almost no difference is registered when comparing the voting cohesion in the case of Turkey-related votes with the overall level of cohesion. In general, all major groups do not seem to suffer from ideological mismatch among the affiliated m e ps. However, particularly interesting are the cases of the rightist meps affiliated to the Euroskeptic group of Europe of Freedom and Democracy and of the democratic Euroskeptics (ecr) where the impact on groups' discipline is rather remarkable. In the first case, the 'Turkey discourse' seems to play the role of identitarian glue, thereby fuelling the group's cohesion from 50% to 74% in the case of Turkey-related votes. The high ideological coherence of the efd group in the Turkey-related votes is clearly reflected by the almost perfect match between expressed votes and declared opinions, discussed in the previous section. The opposition to Ankara's membership represents a part of the ideological dna of the efd group, as evident by the words of its charter: 'Peoples and Nations of Europe have the right to protect their borders and strengthen their own historical, traditional, religious and cultural values.'8
In the second case, despite the modest mismatch between expressed votes and declared opinions, the 'Turkey discourse' seems to play a significant and divisive role within the ranks of the European Conservatives and Reformers, thereby determining the lowest level of cohesion among the eight group (—16%). A more in-depth analysis of the voting defections seems to suggest the presence of a deep-rooted disagreement when it comes to Turkey between the two main components of the group, the British conservatives and the Polish nationalists, with the latter sponsoring a more intransigent stance.
conclusions
The present article has the ambition to be a pioneering attempt to explore the nature of the 'Turkey discourse' looking at the meps' perception-based representation of Turkey and at the way it reflects
volume 6 | 2013 | number 1
Stefano Braghiroli
their voting behavior. As this ep perspective is generally ignored by the mainstream literature on Eu-Turkey relations, the revealing potential of our results appears even higher. In this respect, the ep seems to represent a perfect laboratory to study the impact of cross-cleavage [18] issues, such as eu Turkey's bid, on the voting dynamics given its multinational, multilingual, and multicultural nature. A comparative analysis of the ep voting dynamics on the 'Turkey discourse' vis-à-vis the perspective of the Eu's executive institutions (the Council and the Commission) seems increasingly necessary also in the light of the ep's growing stake in the enlargement process due to the recent treaty reforms.
Having in mind the ep's exceptional nature and multi-dimensionality, our primary objective was to assess how the meps frame Turkey and how this vision affects their voting stance towards Ankara in the parliamentary debates. In the analysis, the results are presented according to the following two criteria: the m e ps' partisan affiliation knowing that the general patterns of competition and coalition in the ep are largely based on the ideological left-right division, and the meps nationality knowing the high domestic salience and significance of the 'Turkey discourse.'
Practically, the analysis performed in this study has been twofold. First, it described separately how the meps look at Turkey and how they vote when Turkey-related votes are at stake, using respectively ees/ttl survey data and Rcv data. Then the analysis crossed these two dimensions at an individual level in order to assess the level of match between the m e ps' declared opinions and expressed votes.
The goal of the analytical efforts was to identify the pragmatic or idealistic/identitarian behavioral styles affecting the meps' voting decisions and groups' internal coherence.
In both cases, our analysis proved successful and particularly revealing, thereby demonstrating that the nature of the voting dynamics is much more complex than it might appear at first sight. In general, the study found that the meps' declared opinions on average appear not only more normally distributed than the expressed votes, but also more moderate and less polarized. However, the score distribution analysis displayed that the inter-group variance is higher than in the case
ijems
Looking at Ankara from Strasbourg
of opinion distribution by national delegations, but smaller than in the case of the vote-based analysis of the party groups. Therefore, the results seem to confirm the prevalence of politically-driven votes over nationally-driven ones and highlight a significant gap between the meps' perceptional representation of Turkey and their expressed [19] votes.
The separate analyses of survey data and voting records revealed that the meps' voting stance towards Turkey seems to reflect a left-right divide, thereby presenting a clear ideological/partisan connotation. Three clusters emerged reflecting political groups' different levels of support: the right (and moderate) side of the political spectrum representing a majority of negatively expressed votes; The center of the spectrum representing a combination of different voting options and a less ideological approach for the centrist groups; and the left side representing the highest scores and the highest level of Turkey-friendly votes.
Looking at the level of variance among the national delegations, the analysis revealed the existence of national delegations' clusters characterized by a strong voting support towards Turkey, mainly including the Scandinavian delegations and the meps from cee. The high scores of most of the cee delegations seem to reflect the phenomenon of enlargement solidarity.
When it comes to the second part of the study, crossing the meps' perceptional representation and expressed votes, the analysis revealed the existence of frequent pragmatic behaviors witnessed by a mismatch between declared opinions and expressed votes. Our results suggest that a relevant number of meps would adopt a more positive or more negative voting stance towards Turkey than the one sponsored by their party group, while voting consistently with the latter. The emergence of a consistent gap between potential and actual behavior proves the existence of a relevant group of meps who only partially follow their belief structure when voting, thereby prioritizing group's interests or other exogenous instances.
The results of our final experiment therefore suggest that the meps' image of Turkey is not the only and (often) not the strongest criterion according to which they take their voting stance. Moreover, the
volume 6 | 2013 | number 1
Stefano Braghiroli
presence of a relevant number of meps in the unpredicted quadrants confirms that the meps' pragmatic behavior seems to play a relevant role when Turkey is at stake. Interestingly, pragmatic behaviors are not equally distributed among the two deviant categories. The meps who [20] display a positive perception towards Turkey are much more likely to behave pragmatically than those who display a negative perception towards Turkey.
A further evidence of legislators' pragmatic behavior is also represented by the fact that, in all the major groups, despite the relevant mismatch between registered perceptions and expressed votes, the level of cohesion of the groups does not seem to suffer from the gap.
In conclusion, our attempt to penetrate the nature meps' perception-based representation of Turkey as reflected by the parliamentary dynamics, far from being exhaustive, seems to provide a useful map to identify the key dimensions of conflict and the triggering factors related to the identified voting patterns, while representing a valuable stress test of groups' capacity to achieve high voting coherence despite significant internal ideological variance.
acknowledgments
This research has been supported by the Estonian Research Council, under grant project erm o s 95, and by the Scientific and Technological Research Council of Turkey (tubitak), under grant project 2216 — Research Fellowship Program for Foreign Citizens (ref. B.02.1.TBT.0.06 .01-216.01-5/282).
notes
1	The minutes of the parliamentary debate are available at http://www .europarl.europa.eu/sides/getDoc.do;jsessionid= C3356i02E8CABA5A5A9066FC77A2B3E4.nodei?pubRef=-//EP// TEXT+PRESS+DN-20041215-1+0+DOC+XML+V0//EN &language=EN#SECTIONi.
2	For further details, see Hughes (2004) and Boehm (2010).
3	The ees/ttl is a panel project (initiated in 2006) whose aim is to examine the attitudes of meps and top Commission and Council officials towards foreign policy and transatlantic issues. The project is coordinated by the Centre for the Study of Political Change (circap)
ijems
Looking at Ankara from Strasbourg
of the University of Siena in cooperation with other European universities and is supported by the foundation Compagnia di San Paolo.
4	The record of the votes held is available at http://www.votewatch.eu/ search.php. Only the votes with the modal voting option lower than or equal to 75% have been considered in the analysis. In total nine votes were included in the computation.
5	Accordingly, the score equals '+' if the overall body of the proposed text is mostly favourable/positive towards Turkey; it equals '—' if the overall body of the proposed ext is mostly unfavourable/negative; it equals '=' if no position or neutral position is reported.
6	While the ees/ttl sample includes 176 meps, the final experiment crossing expressed votes and declared includes 87 meps for which the comparison was possible, accounting for 18% of the total.
7	Interestingly the moderate epp, with a score of 0.34 presents a level of voting scepticism very close to the non-attached extreme right meps (0.22) and to the Eurosceptic right (0.29).
8	The charter of the efd group is available at http://www.efdgroup.eu /about-us/who-we-are/charter.html.
REFERENCES
Adam, A., and T. Moutos. 2005. 'Turkish Delight for Some, Cold Turkey for Others?: The Effects of the EU—Turkey Customs Union.' CEsifo Working Paper Series 1550, CEsifo, Munich.
Baldwin, R., and M. Widgren. 2005. 'The Impact of Turkey's Membership on eu Voting.' cepr Discussion Papers 4954, Centre for Economic Policy Research, London.
Boehm, W. 2010. 'One day Turkey Will Run the eu.' Die Presse, September 28. http://www.presseurop.eu/en/content/article/ 348111-one-day-turkey-will-run-eu
Braghiroli, S. 2012. 'Je t'aime ... moi non plus! An Empirical Assessment of Euro-Parliamentarians' Voting Behaviour on Turkey and Turkish Membership.' Southeast European and Black Sea Studies 12 (1): 1—24.
Canan-Sokullu, E. 2011. 'Turcoscepticism and Threat Perception: European Public and Elite Opinion on Turkey's Protracted e u Membership.' South European Society and Politics 16 (3): 483—97.
European Commission. 2005. 'Commission Presents a Rigorous Draft Framework for Accession Negotiations with Turkey.' ip/05/807, June 29.
volume 6 | 2013 | number 1
[22]
Stefano Braghiroli
Falkner, G., and O. Treib. 2008. 'Three Worlds of Compliance or Four? The Eu-15 Compared to New Member States.' Journal of Common Market Studies 46 (2): 293—313.
Günes-Ayata, A. 2003. 'From Euro-Scepticism to Turkey-Scepticism: Changing Political Attitudes on the European Union in Turkey.' Journal of Southern Europe and the Balkans 5 (2): 205—22.
Hix, S., and A. Noury. 2009. After Enlargement: Voting Patterns in the Sixth European Parliament.' Legislative Studies Quarterly 34 (2): 159—74.
Hughes, K. 2004. 'Turkey and the European Union: Just Another Enlargement?' Friends of Europe Working Paper Series, September 2004. http://www.cdu.de/en/doc/Friends_of_Europe_Turkey .pdf
Independent Commission on Turkey. 2009. 'Turkey in Europe: Breaking the Vicious Circle.' http://www
.independentcommissiononturkey.org/pdfs/2009_english.pdf
LaGro, E., and K. E. J0rgensen. 2007. Turkey and the European Union: Prospects for a Difficult Encounter. Basingstoke: Palgrave MacMillan.
McLaren, L. M. 2007. 'Explaining Opposition to Turkish Membership of the Eu.' European Union Politics 8 (2): 251—78.
Müftüler-Bac, M., and L. Mclaren. 2003. 'Enlargement Preferences and Policy-Making in the European Union: Impacts on Turkey.' Journal of European Integration 25 (1): 17—30.
Rahman, J. 2008. 'Current Account Developments in New Member States of the European Union: Equilibrium, Excess, and euPhoria.' imf Working Paper Wp/08/92, International Monetary Fund, Washington, d c.
Rasmussen, A. 2008. 'Party Soldiers in a Non-Partisan Community? Party Linkage in the European Parliament.' Journal of European Public Policy 15 (8): 1164—83.
Zielonka, J. 2002. 'How New Enlarged Borders will Reshape the European Union.' Journal of Common Market Studies 39 (3): 507—36.
ijems