Looking for the Babel-fish? Language discrimination and EU law

Giovanni Zaccaroni, Ph.D (University of Bologna), Junior lawyer FratiniVergano 

Introduction: Language differentiation in the EU

There is no need to create language differentiation in Brussels as there was in the ancient Babel. There are already 24 official languages, according to Regulation n. 1/1958, and the EU Charter of Fundamental Rights protects the rights of EU citizens to address the EU Institutions in their own languages. The working languages mostly used within the Institutions are three: English, French and German; however, from a legal standpoint, all the documents concerning the Institutions need to be translated in all the official languages of the EU.

This was still manageable until the great enlargement of 2004, but after that it has become increasingly difficult for the Institutions to communicate in the plethora of different languages. This creates quite a complex linguistic regime which could sound odd to a person who is not confident with EU issues (surely not as much as the double seat of the European Parliament). This has rightly brought Advocate-General Kokott (in C-566/10-P) to quote the Babel-fish of the "Hitchhiker's guide to galaxy" and to refer to the aspiration coming from certain Member states of having within the EU Institutions the brand new version of this legendary beast.

So it was easily predictable that sooner or later, in the franglais-friendly environment (with some solid German stronghold) of the EU Institutions, one of the unlucky countries where these three languages are not easily spoken would have attempted to challenge the validity of an act of the Institutions on the ground that it was violating language discrimination. This was precisely what occurred when Italy and Spain appealed in front of the General Court, and then the Court of Justice, several EPSO (the inter-institutional body in charge with the selection and recruiting of a substantial part of the EU staff) competitions which appear to be waiving the linguistic regime established since the creation of the European Community by the Regulation n. 1/1958 (as successively amended), the Staff Regulations and the Charter of Fundamental Rights.

1. The facts of the cases at stake

The judgments of the Court arise from the appeal, brought by the Italian Republic, of the judgement of the General Court in Joined Cases T-166/07 and T-285/07, where the General Court dismissed an action for annulment against an EPSO competition based precisely on linguistic discrimination. The General Court held that there was no violation of the principle of non discrimination on the ground of language (as it is enshrined in the various EU legal instruments). In particular, the General Court maintained that the Commission assured an adequate publicity of the competition (while not translating the text in all official languages) and that there was no proof of the fact that the limitation of the selection of the second language to English, French and German was discriminatory.

The case was appealed in C 566/10-P on the ground that the decision of the General Court was wrong in law, as it was failing to consider that the duty not to discriminate on the ground of language was extended also to the hiring procedure of the EU Institutions. The Italian Republic won the appeal in the case, and that made easier for the General Court to rule on subsequent appeals of EPSO competitions brought by Italy and Spain in Joined Cases T-124/13 and T-191/13 (released in September 2015) and T-275/13 (released on 17 December 2015). However, (as it is testified by the reasoning of the Court in paragraphs 70 ss. in case T-166/07) the case law of the Court of Justice was quite vague, up to 2012, as regards the application of the principle of non-discrimination on the ground of language to the EU Institutions and to the specific case of EPSO competitions.

It was to a certain extent obvious that each Institution, in this case the European Commission, was free to impose some language requirements in order to assure that the members of the staff hired were able to take up immediately the duties connected with their civil servant status. However, since the entrance into force of the Amsterdam Treaty, which provided the legal basis (now Article 18 of the TFEU), inter alia, for the twinned directives 2000/43/EC (concerning race discrimination) and 2000/78/EC (concerning discrimination on four other grounds), and the entrance into force of the Charter of Fundamental Rights, the situation concerning the prohibition of discrimination has slightly changed. The fact that the Charter of Fundamental Rights is addressed primarily to the EU Institutions leaves very little discretion to the Court as to the application of the regime expressed in Regulation 1/1958 and by the Staff Regulations to the activity of the EPSO (an inter-Institutional body), and the strict rules about the burden of proof in cases of discrimination imposed on the Commission to justify each derogation from the general prohibition of discrimination of which Regulation 1/58 is an expression.

2. The equality scrutiny and the strict concept of equality

The reason why these cases are worthy of comment is that they are a very useful example of the changing attitude towards equality scrutiny in Court of Justice case law of the last few years. In particular, they are a sign of the deeper implementation at the level of judicial protection of a strict notion of equality (in which equality is presumed, unless the contrary is proved), which seems to leave apart the more traditional "relational concept" of equality (where, on the contrary, much more attention is devoted to the selection of the appropriate comparator). One would say, nothing new. The regime of anti-discrimination law in the EU after Amsterdam is built in this way: the discrimination is presumed, unless the contrary is proved. But the fact that as a general trend in anti-discrimination law there is no comparison as such and that the structure of the equality judgment switches straight to the justification regime implies that the proof become increasingly difficult.

As to the traditional understanding of the notion of equality, the structure of the judgment has always been clearly divided in two main different steps: comparison and justification. The comparison implies the choice of the appropriate comparator, and, if the situation of the people subjected to the equality scrutiny is different from the one of the comparator, then there is no discrimination. In these cases (T-124/13, T-191/13 and T-275/13, but also in C-566/10-P) there is no evidence of the fact that the Court of Justice has made the choice of selecting an appropriate comparator. It is only presumed, positively or negatively, that there is no interest of service justifying a derogatory linguistic regime for the European Institutions. On the other side, if one looks at paragraph 99 of the decision in joined cases T-166/07 and T-285/07, which was the basis of the appeal of the first case (C-566/10-P), one find the opposite approach: a more detailed analysis and an evaluation of the differences (a comparison) between the case of people speaking multiple languages and people speaking only one.

From this point of view, the switching of the balance of the equality scrutiny to the justification step means that the outcome of these cases turns upon the strict approach to justification. In this case it is undeniable that there is a Regulation and a linguistic regime applicable to the EU Institutions, which leaves very little room for derogations. However, are we really sure that equality cases should presume discrimination and go straight to the justification issue, without indulging in a comparison? This approach is surely extremely useful when it comes to a situation of discrimination in which it is difficult to unveil the discriminatory behaviour, but in term of costs-benefit approach, to use it in the inter-Institutional relationship is like using a tank to kill a bee.

3. The “interests of service” versus the presumption of discrimination and the justification regime

However, it cannot be denied that there is a problem, from the viewpoint of the application of anti-discrimination law, in the apparent inconsistency between the language regime of the EU Institutions, written in the Treaties, and the behaviour of the same Institutions. The EU Institutions apply in the recruitment procedure a golden rule, which implies (reasonably from the viewpoint of people who have ever worked inside the EU) that every single civil servant should be able to communicate effectively, other than in his or her native language, in one of the so-called “working languages” (English, German and French). However, in the EU Treaties as well as in the various pieces of secondary legislation there is no sign of these “working languages”, other than the possibility set out in Regulation 1/58 for the Institutions to approve their own linguistic regime (note that the only EU body that has approved a specific linguistic regime which has passed the Court’s scrutiny is OHIM, the EU trademark agency).

Not one of the Institutions has so far introduced such a regime, and the outcome is that the linguistic regime of the European Commission had, in the three cases examined, been scrutinised for justifications. During the justification scrutiny what should be proved is that, provided that the aim to be reached by the Institutions is legitimate, there is an evident need for the application of a specific derogatory linguistic regime, relying on several pieces of quantitative and qualitative evidence.

In fact, the presumption of discrimination which lies at the core of the actual EU anti-discrimination legislative framework, and which is applied in the three cases by the Court, leaves the proof of the absence of the discrimination entirely on the perpetrator of the discriminatory behaviour. And in case that the proof is not available, also the proof of the objective justification lies on the person, or on the Institution, which is upholding the discriminatory behaviour. From this perspective here is where the apparent weakness of the argument of the Commission in these judgments lies: the Commission thinks that the existence of an interest of service is sufficient to escape the proof of the absence of the discrimination. The Court replies that unless specific and detailed evidences are produced, which demonstrate the contrary, the “interest of service” does not allow per se the Commission to derogate from the linguistic regime. Interestingly, the Commission attempted, in particular in Case T-124/13, to provide evidence on the need for the selection of a certain number of languages within the interest of the service. The Court first, and the General Court later on, have however considered this argument insufficient, on the ground that the principle of non discrimination, of which the linguistic regime of the Institutions is an application, has a specific structural importance for the survival of the whole EU machinery.

4. Costs and benefits of the outcome of the judgments at stake

At the end of the day, what stands out of the crowd in these judgments is the fact that the EU Institutions, and the EPSO, should translate in all the official languages of the EU the text of the exams of the competitions, and that each single EU citizen can apply to the competition provided that he or she speaks at least two EU languages, no matter which languages they are. However, it is easy to doubt that an applicant who is able to find his way through the tricky EPSO competition will immediately be able to take up the duties connected to his service if he or she only speaks, for instance, Italian and Polish. And what can be said about the linguistic regime of the recruiting procedure applied by other Institutions outside the EPSO scheme?

For instance, the Court of Justice itself requires a very good knowledge of French for its positions available within the Research and Documentation DG, implying also the effective ability to write in this language. Can such a requirement resist the linguistic regime applicable to the EU Institutions and the equality scrutiny applied by the very same Court? These are all very interesting questions, but until the Institutions and the Member States effectively regulate the issue of the working languages, it is highly likely that we will see more and more judgments like these, having as a main effect to paralyze the already complex machine of the EPSO, and to render even slower the process of hiring new staff, delaying once more the crucial generational turnover within the EU Institutions.

Barnard and Peers: chapter 3

Photo credit: gizmodo.com

Babelfish credit: Douglas Adams

The CJEU transforms family reunion for Turkish citizens

Steve Peers

In this week’s judgment in Dogan, the Court of Justice in effect established a new set of rules for family reunion for most Turkish citizens living in the EU – although those rules will still differ in each Member State.

The case concerned Mrs. Dogan’s application to come to Germany to live with her husband, after spending thirteen (by now sixteen) years apart. She was refused on the grounds that her German language skills were not satisfactory, for even though she passed the relevant language test, her written German was not satisfactory due to illiteracy.

This raised two legal issues. First of all, since her husband ran a business in Germany, did the national rule breach the 1970 Protocol to the EU/Turkey association agreement, which prohibits new restrictions on establishment or the provision of services? Secondly, did the national rules comply with the EU’s family reunion Directive, which permits Member States to require family members to comply with ‘integration measures’?

The Court of Justice decided to answer only the first question. This means that its ruling is only relevant to Turkish nationals, rather than all third-country nationals. Having said that, Turkish nationals make up a large proportion of all third-country nationals in the EU, and the Court’s judgment on this point applies to all Member States, including the UK, Ireland and Denmark, which do not apply the family reunion Directive.

However, it should be noted that the Advocate-General’s opinion also argued that the integration requirement breached the family reunion Directive (see discussion here), and that there is another case before the CJEU which raises this issue (notably the K and A case; see further the Commission's recent guidance on the Directive, discussed here).

The judgment

Previously, the CJEU had held that the standstill on new restrictions on establishment and provision of services was binding, had direct effect, and prevented any new measures making it more difficult to exercise self-employment or provide services. But the previous case law only concerned those persons who were actually self-employed or providing services. Could it extend also to the rules regulating the family members of such persons?

The CJEU said it could. In the Court’s view, where the national law made family reunion ‘difficult or impossible’, the establishment of a self-employed person could be ‘negatively affected, since that person would ‘find himself [or herself] obliged to choose between his [or her] activity in the Member State concerned and his [or her] family life in Turkey’. In this case, the national rule made family reunion more difficult, and so violated the standstill clause.

Finally, the Court noted that a measure infringing the standstill clause could be permissible, if it could be ‘justified by an overriding reason in the public interest’, and was ‘suitable to achieve the legitimate objective pursued and does not go beyond what is necessary in order to attain it’. In this case, assuming that the objectives of the national law (preventing forced marriages and promoting integration) were overriding reasons in the public interest, the national law failed the proportionality test, since the refusal of an application on linguistic grounds was automatic, without taking account ‘of the specific circumstances of each case’.

Comments

In order to assess the impact of the Court’s ruling, it is necessary first of all to determine its scope. The geographical scope, as noted already, is all Member States. Also, while the family members of Turkish nationals will usually reside in Turkey, the judgment should logically apply also if those family members reside in another third State, or in a Member State. Although in the latter case, it is possible that EU immigration law, such as the long-term residents’ Directive, might also apply, that legislation is without prejudice to more favourable rules in treaties between the EU and third States.

The temporal scope, as established in the previous Dereci case, is 1973 for the first nine Member States (the date of entry into force of the 1970 Protocol to the association agreement), and the date of joining the EU for all other Member States. That means that the national law in force on that date regarding family reunion cannot be made worse for those joining Turkish nationals. Moreover, as established in Toprak and Oguz, if the national law is made more liberal after that date, the standstill also prevents any reversion to the more restrictive law.

Next, what is the personal scope of the judgment? There are three facets to this issue: the sponsor (ie the Turkish citizen in the EU); the family members who seek to join them; and issues of nationality law.

For the Turkish citizen in the EU, the standstill which the Court discussed here also applies to providers of services, but not (as the Court established in Demirkan) to recipients of services. There is also a parallel standstill relating to Turkish workers: according to Article 13 of Decision 1/80 of the EU/Turkey Association Council, the rules on access to employment of Turkish workers and their family members cannot be made more restrictive either. The case law on this provision (see most recently the Demir judgment) makes clear that the standstill on workers and the standstill on establishment must be interpreted the same way (although for the first nine Member States, the standstill on workers took effect in 1980, rather than 1973). In particular, the standstill on workers applies to measures concerning their first admission. It must logically also apply to rules on the first admission of the worker’s family members, since workers, like self-employed persons, would otherwise face a choice between carrying on with employment in a Member State and enjoying family life in another country.

As for family members, while this case concerned spouses, it would logically apply to any other family member as well (Mrs. Dogan had also originally applied for two of the couple’s four children to be admitted also). Simply put, the Turkish citizen in the EU could equally face a choice between his or her economic activity and family life with unmarried partners, children, parents or other family members as well.

Next, as for issues of nationality, one question is whether the primary right-holder and/or the family members concerned can still rely on the rule, if they are dual citizens of Turkey and another country. This question appeared to be answered by the CJEU in Kahveci and Inan, in which it ruled that dual citizens of Turkey and a Member State can rely on the EU/Turkey association rules. But in the pending case of Demirci, concerning social security, an Advocate-General has recently argued that dual citizens cannot invoke the rules. The CJEU will likely clarify this issue soon.

Another nationality-related question is whether third-country national family members of Turkish nationals (in this context, meaning family members who are not nationals of Turkey or a Member State) are covered by the standstill rules. It is clear that they are, by analogy with the recent judgment in Dulger.

Next, what is the material scope of the judgment? In other words, what type of rules relating to family reunion are covered? The test is whether the national rule makes family reunion difficult or impossible. This is apt to cover waiting periods, in-country applications, income requirements, fees for applications (see the Sahin judgment), age limits for spouses or children, and accommodation or sickness insurance rules. Furthermore, the standstill should equally apply as regards different categories of sponsors: for instance, more favourable rules that apply to groups such as refugees, long-term residents or highly-skilled workers cannot be made more restrictive either.  

Finally, what about the possibility of justification? The CJEU did not rule directly in this case as to whether prevention of forced marriages and integration were possible grounds of justification, although in Demir it ruled that prevention of irregular immigration could be such a ground. An example of a new rule which could surely be justified would be a ban on admission of family members if they have received a five-year prison sentence, as compared to a ten-year threshold previously. On the other hand, it seems doubtful that a national rule which simply aims to reduce the numbers of Turkish persons’ family members entering the country can be justified, in light of the overall objective of the association agreement of eventually providing for free movement of people and preparing for Turkish accession.

In any event, it is clear from the Dogan judgment that even if a new national restriction is justified, refusals cannot be automatic and must take account of individual cases. The requirement that the restrictions be ‘suitable’ also suggests that they must be the only possible means to achieve their end. More generally, it is surely the case that like any derogation from EU law, such restrictions and their application must also be judged in light of the EU Charter, namely the right to family life and rights of the child. 

Barnard & Peers: chapter 24, chapter 26