'One billion euros' sounds a lot, in any language

The European Union takes its citizens' attachment to their mother tongue so seriously that multilingualism was actually written into the Treaty of Rome. EU citizens have the right to communicate with its institutions in the language of their choice; and its regulations and decisions must be translated into all of the EU's official languages before they become valid. As the EU has expanded, that has become an increasingly cumbersome task. Brussels now spends more than €1bn a year on translation and interpretation.

As well as rising costs, there are serious practical implications. In 2003, an agreement to allow developing countries to bypass patent laws and import cheap generic medicines was delayed by more than a year, because of the necessity of translating it.

Translation also increases the time and the costs involved in filing a patent in Europe: it takes on average more than four years, and costs €129,000 in the EU, compared to just over two years, and €16,500 in the United States. Linguistic affairs are no laughing matter, as MEPs will attest: with their debates simultaneously translated into the European Union's official languages - of which there are now 23 - they have been asked to avoid making jokes, to avoid confusion. In a new CEPR paper, Researchers Jan Fidrmuc and Shlomo Weber and their co-author, Victor Ginsburgh, set aside fraught questions of national pride, and analyse data about the linguistic abilities of Europe's citizens, to identify the optimal set of official languages for the new EU. In the hope of extricating the EU from this linguistic quagmire, Fidrmuc et al. set out to narrow down the current list of 23 languages, by asking which alternative, smaller set would leave the fewest EU citizens 'linguistically disenfranchised' - unable to participate in EU politics, because they don't speak or understand any of its official languages.

In order to answer this question, the authors analyse the results of an EU-wide Eurobarometer survey, carried out in 2005, which asked people which languages they spoke 'well enough to have a conversation,' allowing them to list up to three.

Using this data, the authors are able to quantify the citizens who would be genuinely excluded by a given list of languages.

How many languages would be optimal depends on the political question of what level of linguistic disenfranchisement is acceptable. English is the most widely-spoken language across the EU, for example - but making it the only official language would leave out as many as 63% of EU citizens, who cannot speak it. Moreover, in 20 of the 27 member states, more than half the population would be disenfranchised.

To identify a more politically acceptable solution, the authors build up a 'nested set' of languages: English must be in the optimal language-set, because it leaves the fewest citizens disenfranchised; the next option on the policy menu is a set including English, plus the language which, together with it, would leave the smallest group of non-speakers. This turns out (narrowly) to be German. The next set is English, plus German, plus the language which brings the largest number of non-English and non-German speakers into the fold - French - and so on.

If English, German and French were made the EU's three official languages, 38% of citizens would still be linguistically disenfranchised, because they speak none of these three. Increasing the set to six, by adding Italian, Spanish and Polish, leaves just 16% of the population disenfranchised; adding Romanian reduces that to 13%.

In order to get a better idea of the political constraints facing linguistic reformers, Fidrmuc et al. examine how far down the list of language sets it would be necessary to go to win support for a change through the decision-making system of Qualified Majority Voting.

The authors note, of course, that any changes to EU official languages currently require unanimity among member states (and as such are highly unlikely to win support); but they examine what would happen under the less onerous QMV system.

On this basis, as few as seven official languages could win support if each country were willing to vote for a proposal which disenfranchised 40% or less of its population. If that seemed too high a price, and countries would only accept a proposal that disenfranchised 10% or less of their population, the list of official languages would have to expand to 11. In other words, the current system is one the member-states would be highly unlikely to agree themselves, under QMV. Even a list of 11 would slash the number of official languages in half.

Survey evidence suggests there may be considerable public support for greater linguistic conformity: a recent study carried out across the EU showed that 54% of respondents 'tend to agree' that European institutions should adopt a single language to communicate with EU citizens.

That support is likely to increase over time: the Eurobarometer survey of language abilities allows the authors to measure the proficiency of different generations in foreign languages. Among the younger respondents, there was a far greater tendency to speak foreign languages. Using English as the EU's one official language would only leave out 45% of young Europeans, compared to 63% of the population as a whole, for example. Considering only 15-29 year olds, something between three and seven languages would be needed to secure a majority under QMV, depending again on the levels of disenfranchisement each government was willing to accept among its own population.

How easy it is for non-speakers to master the chosen official languages may also be relevant. For that reason, the authors carry out another analysis taking into account the 'distance' between one language and another, i.e. how similar it is (Dutch speakers tend to be able to master German, for example, more easily than Latvian speakers can pick up French). If speakers of similar languages are counted among those who could understand the new list of official languages, again, no more than seven languages would be needed to win a vote under QMV.

Linguistic disputes can be an emotive cocktail of fierce national or ethnic pride and social identity. In this paper, however, Fidrmuc and his co-authors seek to cut through these emotive questions with a practical analysis, which shows that reform could gather broad political support - and if it were achieved, would cut the costs of translation, reduce the risks of misunderstanding and delay, and perhaps even allow MEPs to unleash once more their sense of humour.

DP6367: Ever Closer Union or Babylonian Discord? The Official-language Problem in the European Union
Jan Fidrmuc, Victor Ginsburgh and Shlomo Weber