1.

The Rules of Procedure of the Commission (1) specify that all acts adopted by the Commission shall be attached to a ‘summary note’ (in the oral procedure) or ‘day note’ (in the written, empowerment or delegation procedure) (2), in the authentic language or languages, i.e. in the official languages of the Communities in the case of instruments of general application and the language or languages of those to whom they are addressed in other cases (Article 18 of the Rules of Procedure).

2.

Although ‘comitology’ Committees come under the Commission itself (3) and Council Regulation (EEC) No 1/1958 of 15 April 1958 applies to their work (4), an explicit reference to the language regime was deliberately not introduced in the Standard Rules of Procedure (5), so as to allow each committee to find pragmatic solutions tailored to its real needs. Documents that would, in any case, be translated at a later stage (in particular, draft measures whose final version must be published in the OJ) are in principle already translated for committee meetings.

3.

A note from the Secretariat-General (6) issued in June 2004 specifies that ‘for those implementing measures which are addressed to a reduced number of Member States only, a translation in the corresponding languages would be sufficient, together with a communication published in the Official Journal of the European Union announcing the adoption of the decision and informing the reader of the website on which it is accessible. Full translation is required only if the publication in the Official Journal of the European Union is a legal precondition for their coming into force (i.e. for regulations, directives and decisions of the EP and the Council adopted in codecision and Commission regulations and directives addressed to all Member States, but not for Commission decisions).’

4.

The May 2004 Communication distinguishes between documents resulting from legal obligations (subdivided into nine categories) and documents of political importance (subdivided into 11 categories). The translation of certain categories of documents, such as publications other than the Official Journal ofthe European Union, is not covered by the Communication and must be decided on a case-by-case basis through dialogue between the requesting DG and DGT (7).

5.

The main categories of documents translated into all languages are: legislative proposals, explanatory memoranda, reports required by legal texts, announcements of State aid and antitrust procedures published in the Official Journal, as well as the subsequent final decisions published in the Official Journal, anti-dumping decisions, Communications of the Commission (including white and green papers), calls for tenders or for manifestation of interest and press communiqués.

6.

Other documents are only translated into the procedural languages (English, French and German) (in particular communications to the Commission, follow-up reports of Council decisions and Commission proposals), or into the language of the addressee (i.e. letters to permanent representations, replies to individual requests and financing proposals). Technical annexes are translated into no more than three languages, or are not translated at all, depending on the ‘real needs’. Replies to written parliamentary questions are translated into the language of the Member of European Parliament who has asked the question and into one procedural language.

7.

Pursuant to Rule 138 of the Parliament's Rules of Procedure, ‘all documents of the Parliament shall be drawn up in the official languages’. Article 150 states that ‘amendments shall be put to the vote only after they have been printed and distributed in all the official languages, unless Parliament decides otherwise. Parliament may not decide otherwise if at least 37 members object.’

8.

On 19 April 2004 the Bureau of the European Parliament adopted a Code of Conduct which entered into force on 1 May 2004 and replaced a similar Code of Conduct of 8 February 1999. Article 1(2) of the Code of Conduct states the following: ‘Language facilities in Parliament shall be managed on the basis of the principles governing “controlled full multilingualism”. Accordingly, the right of Members to use in Parliament the official language of their choice, pursuant to Parliament's Rules of Procedure, shall be fully respected. The resources to be devoted to multilingualism shall be controlled by means of management on the basis of users' real needs, measures to make users more aware of their responsibilities and more effective planning of requests for language facilities.’

9.

The documents with the highest number of target languages are amendments, draft opinions and draft reports. For all the other document types, target languages range from one (mainly English) to four. Table 8, drawn up for the purpose of the audit, gives an overview of the types of documents which are translated at the Parliament.

10.

A document published in the Official Journal of the European Union or voted on in plenary will always have to be translated into all official languages, whereas translation of working documents can be restricted to the language needs of the readers.

11.

Pursuant to Regulation No 1 and Article 14 of the Council's Rules of Procedure (8), translation into all official languages is required for documents discussed at ministerial meetings, except as otherwise decided unanimously by the Council on grounds of urgency. Any member of the Council may oppose discussion if translations are not available.

12.

The Annex to the Guide for producing documents for the Council and its preparatory bodies lists the ‘core documents’ which have to be translated into all languages (9). As stated in the guide, documents that are not on this list may not be translated into the EU-10 languages. These documents may be translated into the EU-15 languages, on a selective basis, if the volume of work of DG A III allows it, and the deadlines laid down in the rules of procedure are respected. DG A III is given authority to refuse the translation of any document not on the list.

1.

The full cost (cost before tax and pension), as estimated by the auditors, includes all the direct and indirect costs related to the translation activity.

2.

By deducting the Community tax paid by staff translators from the ‘cost before tax and pension’ and by adding the fictitious employer's pension contributions for staff translators (which are equal to twice the contribution paid by the official), the cost to the Community budget is obtained (cost after tax and pension).

3.

Direct costs are the staff costs of the activities related to translation, performed by translators or secretaries (including pre-translation by the translation units, revision and formatting, and excluding the time spent on tasks other than translation).

4.

Indirect costs include the salaries and allowances of staff who are not directly involved in specific translation tasks (terminologists, working method specialists, Heads of Unit, language advisors and senior management, freelance unit and planning) as well as a pro rata for time spent by translators on activities other than translation or revision, plus overhead costs such as IT, buildings and office furniture, human resources management, etc., which in part are not directly managed by the translation services. The split between direct and indirect cost was agreed with the institutions.

5.

To work out the staff cost the following elements have been used:

6.

The amount of the IT cost was worked out in the light of the 2003 budget implementation for the specific budget lines and the average acquisition and maintenance cost of the IT equipment as established by the IT and Finance services.

7.

As internal translators carry out quality control and revision of translations made by freelance translators, part of their time and cost was allocated to external translation.

8.

The total translation costs of the three institutions were then divided by the number of pages translated per year to obtain the cost per page (see Tables 2, 3 and 4). For the Commission and the Parliament the official number of pages translated, as produced by DGT and DG VII, were used, as the page count methods used by these two institutions (see footnotes 8 and 9 in the report) lead to comparable results. As the official Council figures for 2003 were simply based on the number of printed pages of the documents, instead of on the number of characters to be translated, the Court calculated the average correction factor for a sample of 100 documents (2). By multiplying this average correction factor by the official number of printed pages issued by DG A III, a comparable production figure was obtained.

9.

The cost per page calculated by the Court is not the full cost of multilingualism, as it is does not include the cost of revision by the lawyer-linguists or the additional costs of publishing documents in 20 languages. As during the legislative process many consecutive versions of the same draft legislation are translated, the actual cost of adopted legislation is a multiple of the cost per page mentioned in Table 2 of the current report. As an example, 15 versions of the Constitution for Europe were translated.

1.

Within the European Parliament, the number of working languages has not been reduced. This means that all sittings documents and documents for meetings of official bodies are translated into all the working languages of the Union.

13-14.

In connection with translation work, Parliament applies its Rules of Procedure and the Code of Conduct on Multilingualism adopted by the Bureau, which entered into force on 1 May 2004.

While the code of conduct does not of itself constitute an authorisation system, it does establish a ranking of bodies authorised to request translations, lays down a set of priorities concerning the types of document to be translated, both in-house and externally, during the various legislative procedures, and an order of priority for the languages requested, and lays down deadlines for submitting translating requests as well as limits on the length of the texts to be translated. The code of conduct also lays down conditions requiring express authorisation of translation requests made by entities that are not included on the list of authorised users. Such requests are submitted to the Bureau and authorisation is given only if the document in question is of essential importance to Parliament's business.

15-16.

Access to the system is granted solely at the request of the head of a requesting service. However, as part of the overhaul of the GEPRO IT system (ITMS project) all the current accesses will need to be re-authorised and, as far as is possible, the overall number will be reduced. In Parliament's view, the risk to which the Court refers is minimal owing to the restrictions set in place by the afore-mentioned provisions.

See also the reply to points 13-14 above.

17.

Although this does not happen frequently, in some cases Planning reduces the number of languages requested, either in accordance with the rules on language profiles or with other specific exceptions of which the requesting services may have failed to take account.

Approximately 90 % of the time, translations are delivered prior to the meeting date. In cases where the deadline cannot be met, the text is translated only if it will serve as the basis for the next stage in the procedure.

20.

The situation was as follows in April 2006: 238 FTE translators (FTE = full time equivalent), of which 25,75 CS, 25 ET, 26 HU, 24,5 LT, 28 LV, 23 MT, 31 PL, 26,75 SK and 28 SL, and 5 heads of unit (CS, LV, PL, SK, SL), plus one acting head of unit (MT), were responsible for translating into the EU-10 languages.

23.

Within Parliament, translations from the new languages have almost always been performed by translators whose mother tongue is the same as the target language.

Parliament does not restrict the number of working languages and does not recognise the concept of ‘procedural languages’. (See also the reply to point 1).

26.

The concept of ‘procedural languages’ does not exist within Parliament. (See point 23).

32-33.

The concept of ‘language profiles’ has been introduced under the code of conduct. As the Court points out, documents considered by the parliamentary committees are translated only into the languages of the full and substitute members of the committees.

36.

In Parliament's views, the approaches are not inconsistent because they meet the specific needs of each institution, and those needs differ from institution to institution.

37.

Parliament would draw the Court's attention to the quality action plan drawn up by the Interinstitutional Committee for Translation and Interpreting (ICTI) at the request of the secretaries-general of the institutions, which will be implemented in the near future. The plan sets out 14 practical measures, some of which apply to both translation and interpretation, in areas such as human resources, the technical and linguistic quality of originals and translations, documentation, terminology and tools. Provision has been made for an official in charge of quality to be appointed in each of the services concerned.

39.

It is important to point out that within Parliament systematic checks are made for all types of error, and this includes checking that the correct terminology has been used.

40.

Although a specific tool to measure user satisfaction has not been introduced, there is a procedure for submitting complaints, for which a specific form exists. Furthermore, the ‘clients’ in the EU institutions, particularly MEPs, have sufficiently close links with the Secretary-General to be able at any time, and without having to follow a specific procedure, to express their dissatisfaction should they have cause to do so.

47.

This provision of the code of conduct cannot be implemented under the current system. However, the ITMS project currently under development will contain a billing module designed for this purpose.

52.

As the Court points out, it is important not to focus solely on the price indicator which, taken in isolation, cannot reflect the realities of the translation process; account should also be taken of factors such as quality, deadlines and document types.

53.

Parliament's translation and planning services have started to compile statistics covering three of the points raised, namely:

These statistics are intended to provide performance indicators enabling management to take appropriate action.

56.

Planning of this kind is indeed necessary. However, some aspects of the work of a political institution are inherently unforeseeable.

66.

The EP has initiated a preparatory study into a new management system (T-Flow). This system, which will enable Parliament to meet the concerns raised by the Court, is currently in the development stage and will be operational in the autumn of 2006.

67.

The relevant IT departments are currently looking into the ‘scattered documents’ issue raised by the Court. The translation units do, nonetheless, have a translated-work repository in which to search for texts already translated.

69.

The ratio is 1 to 2 for the EU-15 languages. For the EU-10 languages, the secretaries/translators ratio is currently approximately 0,7 to 2 and should not go higher than 0,8 to 2.

70.

It is true that IT use is not yet uniform throughout the translation service and that working methods sometimes differ. This is due in particular to the large differences in the age of staff. However, as a result of the extensive experience of the more senior translators, per capita productivity is much the same between age groups. The EP Bureau would like to ensure better use of new technologies and has instructed the Secretary-General to submit a report on the issue. Parliament's administration is currently in the process of preparing this report.

73.

It is important to put the remark concerning the French Unit's internal capacity in its proper perspective. That capacity was not underused; it was used for other, additional work. The FR Unit was obliged to devote a particularly large amount of time to setting up and managing the new Polish and Lithuanian Units and to language courses to prepare translators to translate from the EU-10 languages. Its additional tasks also included providing relay translations for other language units and transcribing and revising the Verbatim Report of Proceedings of part-sessions and the ‘economic and monetary dialogue’.

74.

The remark requires clarification because, as the Court notes, the outsourcing of translation work, excluding the CRE, was suspended during the period referred to.

As regards the stable outsourcing rate, the apparent contradiction with the lower overall workload does not bear scrutiny. The reason why the average outsourcing rate hardly changed despite the decrease in the workload is that the main bulk of the workload fell in the first part of the year, which resulted in more work being outsourced during that period.

83.

The EP cannot endorse the Court's conclusion that productivity is lower in the two units referred to, given the difference in the substance and nature of some of their tasks compared to those of other units. This difference requires the deployment of additional human resources. The fact that the English and French units produce fewer-than-average pages translated in-house can be explained by the fact that they serve as a relay for translations from lesser-used original languages, by larger volumes of originals to revise and edit, by a larger volume of verbatim reports of proceedings to transcribe and revise, by the need to manage some of the new units and by EU-10 language training needs. Furthermore, since the autumn of 2005, the English Unit has been sharing responsibility with the Tabling Office for systematic verification of originals.

See also the reply to point 73.

95.

In the event of an overly heavy workload in the language units, Planning first looks into whether workload balancing is feasible on the basis of the language combinations involved and availability in the other institutions.

The EU-10 languages have been excluded only for the time being. Once the EU-10 language units have a full complement of staff, they will be involved on the same basis as the EU-15 language units.

It should be emphasised that active cooperation between the institutions' planning units in connection with workload balancing did not start until 2006. The statistics for the first quarter of 2006 show a significant increase in the volumes exchanged.

96-109.

The EP notes all of the Court's recommendations and will take appropriate action thereon. In particular, an ad hoc working party has started work on a revision of the code of conduct. IT projects covering some of the issues raised by the Court are also under way.

III.

Although the GSC does not have written translation guidelines, the Coordination and Production unit (Central Coordination) ensures a controlled production of translations.

IV.

The problems experienced in 2004 for the EU-10 languages had to do with the difficulties in recruiting staff (see comments on paragraph 19 below). Quality and timely delivery of translation improved significantly in 2005. In 2006, this improvement can be seen by the reduction in requests for translations sent to freelance agencies.

V.

Since the second half of 2005, the Head of Central Coordination resumed close contacts with Permanent Representations, especially from the EU-10, with a view to:

A mailbox is to be launched in the very near future. That mailbox will concentrate all the comments from the delegations, provide an overall view of the perceived quality of the translations produced by the Council and improve the synergy between translation teams and delegations. The delegations were informed of this on 5 May 2006.

VI.

The General Secretariat of the Council intends to set up a system which would allow a permanent review of the cost of translation. There are, however, reasons why GSC translations are expensive, and which are discussed in greater detail elsewhere in this document: the specificities of the Council's work and its deadlines, the need to ensure adequate quality in legislative documents near the end of the production chain, and the temporary structural reasons linked to the restructuring of the Directorate.

VII.

The GSC intends to reduce overcapacity, for example by means of internal redeployment (to other DGs) and interinstitutional mobility (to translation departments in other institutions).

Moreover, the GSC is examining all possibilities provided by the Staff Regulations.

VIII.

The GSC is a major contributor to the workload-balancing system and has accounted for the translation of 55 % of the total number of pages processed under this system in 2005.

IX.

In 2004 and 2005, the GSC developed a coherent policy on the use of modern technology. All the new language units have worked with the available IT instruments since their creation, for the ‘old-language’ units the policy is currently being rolled out in line with the development of appropriate IT tools.

6.

The GSC does not have a freelance unit, or an IT unit attached to its translation department. The structure in place in the GSC is therefore lighter than the other institutions.

13.

The GSC has clear procedures, although these might be improved.

It should be noted that those who can request a translation are defined in the General Secretariat of the Council and known to Coordination. Similarly, types of documents which should be translated are clearly defined in the list of core documents.

14.

The GSC is reluctant to take a differentiated approach to EU-10 and EU-15 languages which could leave it open to accusations of discrimination.

15.

The conclusion of the Court on the translation request is correct. However the acceptance of the requests depends on the control structure in place.

16.

In the Council, Central Coordination, which is upstream of the Translation Department, always checks the appropriateness of a translation before sending the text to one or all the 20 units. In case of problems the Head of Central Coordination intervenes and contacts even with the Presidency may ensue. Central Coordination, namely Coordination DG and Linguistic Coordination, is in constant contact with DGs and also to a lesser extent with the Presidency, and this enables it to make informed judgments on the appropriateness of translations and to prioritise translations in the event of insufficient translation resources.

18.

The authorisation and refusal of translations is a task of the Coordination (especially Tour de Contrôle). The situation is satisfactory in the Council's view. Improvements are however possible. Excluding the ‘check by management’, statistics on refusals may be prepared for further analysis. No measures are required as far as authorisations are concerned.

The fact that the ‘acceptance of translation requests (1) often depended more on the availability of translation capacity than on actual needs’ is inevitable, due to:

Translations that are really needed are never refused.

19.

The recruitment made through EPSO has turned out to be more difficult than expected as a result of the short number of suitable candidates, despite the best efforts made by the institutions to raise the awareness of Member States and other parties to this problem.

20.

It cannot be implied that the fault of not achieving the recruitment targets falls to the individual institutions. Unfortunately, this situation (shortfall in staff numbers) will continue even after that date (late 2006), not only as regards translators, but also as regards Heads of Unit, quality controllers and secretaries.

23.

It is correct to state that very few translators in the EU-15 units other than EN and FR have knowledge of an EU-10 language. That is the result of a deliberate policy decision carefully designed to increase the number of official languages from 11 to 20 without requiring a commensurate increase in staff levels. In preparation for the 2004 enlargement, the GSC has been organising training for translators in the EN and FR units in one or more EU-10 languages, and this has led to the situation presented in the following table (situation as of 22 May 2006).

EN and FR translators knowledge of EU-10 languages (situation on 22 May 2006)

Moreover, the GSC is also examining other ways of providing coverage for all official languages, including using retour translation (having translators working from their native language into EN and FR).

26.

The EN unit participates extensively in the workload balancing system. However, it should be noted that even though there is some overcapacity in the EN unit, intended to ensure full multilingualism in that unit, this should be compared with the cost of the previous system whereby every language unit had the capacity (approximately a minimum of six translators with knowledge of each EU language) to translate from every official language even though the number of pages to be translated from most of those languages represented less than 1 % of the number of pages of the source documents.

Nevertheless, the GSC has always said that it would review its policy in the light of experience (a minimum of two years). This review will take place in the second half of 2006/early 2007.

35.

Central Coordination does control the compliance with the ‘milestones stages’ policy, namely by assessing the need of a translation at a certain stage and refusing to translate certain documents. On the basis of the resources available and the time constraints, Linguistic Coordination ensures that the appropriate number of milestone versions is translated. A number of intangibles such as the political importance of a text, and its priority for a Presidency programme, and time constraints are elements which are taken into account by Linguistic Coordination when deciding on the translation of a milestone version.

38.

In certain cases, when a balance has to be made between timeliness and quality, Linguistic Coordination makes recommendations about the non-revision of a text. This can only occur at intermediate stages.

However, as the Council only translates core documents, a high level of quality is normally required.

39.

The GSC is putting in place a quality control procedure at language unit level and has recently implemented a formal procedure for handling quality complaints and allowing our clients to provide feedback on translations.

40.

See comment to paragraph 39.

41.

The GSC is pleased to note that the Court of Auditors acknowledges that its clients are satisfied with the quality of the service provided. It is however ready to put in place harmonised quality assurance procedures.

42.

The general feedback from EU-10 delegations is now also generally satisfactory. The revision of external translations is a problem. Translations are outsourced because of understaffing problems and their revision would reduce the benefit of such outsourcing (use of valuable resources for revision).

44.

The Council will examine, in the framework of statistical developments in Central Coordination, how to assess the level of compliance with the translation deadlines. However, core documents that comply with the Rules of Procedure are always ready at the meeting for which they are required.

46.

This was due to the dramatic lack of capacity in those languages. However, there is no evidence that the adoption of any legislation was delayed because of the unavailability of documents.

47.

The GSC intends to build on the experience obtained during this audit to set up a system for keeping the cost of translation under review. The GSC is also considering the possibility of systematically sending DGs and the Presidency the ‘cost’ for certain translations especially those giving rise to excessive overtime.

51.

As the Council is at the end of the legislative procedure, it has a particular responsibility for the quality of the adopted (and published) text. Quality comes at a price and has implications for the productivity of the translators. Moreover, as explained earlier, it is not appropriate to send legislative and politically sensitive documents to freelance agencies. This is the reason why the general policy of the GSC is to avoid the use of external translation.

53.

56.

The argument of the Court of Auditors on the importance of accurate forecasts is valid, but in reality it is often beyond the reach of the translation services, e.g. the progress of legislative texts in the Council depends on the progress achieved in different working groups, committees and Council meetings, sometimes even in bilateral meetings.

61.

The staff complement of 25 translators is based on the normal anticipated workload, which in turn is based on the ‘core documents’ policy. The surge or peak in workload at the end of each Presidency is resolved by a much stricter management of the work in the units, renegotiation of deadlines, but most importantly of all by the use of overtime at the end of the working day and, exceptionally, during the weekend or public holidays.

The fact that Council translations are only provided shortly before the meeting is due to the fact that 62 % of Council documents have deadlines of less than 72hrs. It is therefore unavoidable that the translations are available shortly before the meeting.

64.

Correct, but they are anyway experienced enough to know the peaks and in most of the translation units they are ‘helped’ by translators acting as full-time coordinators. Both the Head of Unit and the coordinator are kept constantly informed by Central Coordination. At the Council, which only works with short-term deadlines, the initiative on deadlines is left to Central Coordination which is the unit with a global view on the volume of work.

65.

The Council has changed its system of counting pages for estimating workload and it now uses a system of weighted standard pages, called ‘net’ pages.

67.

This task is performed by the Central Coordination using technical means.

69.

The Council has made considerable progress on this front since the period covered by the Audit, and the ratio of secretaries to translators is still improving.

70.

The Council has invested a lot of time into optimising the use of IT tools. An implementation plan involving the development of tools (EURAMIS and in-house tools), a training programme and the setting up of a special TRADOS cell in Central Coordination is currently being rolled out. The Council intends however to introduce a system of statistical indicators to monitor the use of TRADOS in the Translation Department.

78.

The private sector is concerned less about quality and consistency than the institutions. The experience of the institutions in general, as confirmed by the work of CIEQ, is that legislative texts should not be sent to external agencies. The experience of the Council is that even when texts are sent out to agencies during the early stages of negotiations within the Council (Working Party level), this has not been of much help during the translation of subsequent versions of the text.

85.

There are indeed very limited possibilities of internal redeployment. Translators constitute approximately 66 % of AD staff (790 LA excluding JLs; 450 A) at the Council.

It should also be noted that in the eight language units with supernumerary staff, there is a very high proportion of staff in the age bracket 55-65.

Proportion of staff in language sections in the age bracket 55-65

It is difficult to retrain and redeploy translators in great numbers as translators come with a specialist language background. Nevertheless, the Council intends to look at every possibility of redeployment, both internally and interinstitutionally. Appropriate action is under way.

86.

In addition to the tasks mentioned in the Court of Auditors' report, the English language unit also provides direct assistance to the Presidency (one translator works on the premises of the Presidency) and to other DGs.

88.

The performance indicators and statistics mentioned will be implemented gradually as soon as the resources available for workflow development permit the necessary enhancements.

90.

The statistics used by the Council were changed in 2005 to reflect the practice in other institutions. Statistics for 2003 and 2004 were amended retroactively to bring them into line with the new methods of calculation.

105.

Central Coordination in the Council tries to forecast translation workload and to plan the workload accordingly. In this way, it seeks, within the political constraints, to balance workload and translation capacity.

I.

The Commission is committed to support multilingual written communications through equal linguistic treatment of all official languages. It also sustains multilingualism in the Union, which has become a sensitive political matter. Matching real needs, high quality and cost effectiveness are the guiding principles followed by the Commission in providing translations.

The Commission is the institution that has the initiative of making legislative proposals and therefore is often faced with translations of completely new texts, subsequently modified in the Council and the Parliament. This entails a higher difficulty and more research, mainly in the field of terminology.

II.

Since the completion of the audit, the Commission has introduced further changes in its translation practices and procedures which, in its opinion, address the observations of the Court and are highlighted in the Commission's reply to the relevant paragraphs.

III-IV.

The Commission will update its translation strategy e.g. by including guidelines for non-core documents. DGs will be encouraged to streamline their internal procedures for requesting translations.

V.

The Commission is currently examining the most appropriate ways of measuring quality. DGT has intensified its contacts with Commission DGs and services which allows for regular feedback on the quality of service offered to them. It considers that regular user satisfaction surveys at the level of DGs and services will not provide significant added value.

The Commission's quality control procedures are based on risk assessment and cost-effectiveness, with the aim of producing translations that are fit for purpose in every case. This policy renders a single ‘quality indicator’ based on number of errors per page inappropriate (see also point 100 below).

VI-VII.

The Commission calculated its translation cost and cost per page for 2002 and suspended subsequent updates of these calculations awaiting the outcome of the audit which is subject of this report.

The increase in the cost per page in 2005 for the Commission (compared to 2003) results from the implementation of a strict demand management reducing demand equally across the different languages and below available capacities in ‘old’ language departments. This allowed the Commission to reduce its need for resources for the new language departments thereby creating significant immediate savings for the Commission's administrative budget equivalent to some 380 posts.

Resulting capacity surpluses in the ‘old’ languages were already partly addressed in 2004 and 2005 by means of compensating measures, such as internal redeployment, new methods for increased flexibility, drastic reductions of freelance translations for the ‘old’ language departments, etc. These measures were stepped up at the end of 2005 with the introduction of a strategy for resources allocations within DGT, confirmed in the Commission's APS 2007 decision, aiming at a better balance between supply and demand and improving cost-efficiency, thereby creating additional savings for the Commission's administrative budget, in the medium to long term. This should ensure that differences in work rate and outsourcing percentages are evened out in the medium term.

VIII.

The Commission supports interinstitutional cooperation including the workload balancing mechanism as a means to share spare translation capacity between institutions and has been the highest net contributor throughout the pilot project.

The Commission will continue to contribute towards the fine-tuning of this procedure. However, its potential will be reduced in future as a result of measures currently being taken to adjust resource allocations in accordance with real needs. Moreover, there are limits to how far one can go because of the different nature of work and timetables and agendas of the different institutions and workload balancing cannot replace freelance work, nor will it allow complete absorption of idle capacity across institutions.

IX.

While no information was available in 2004 and 2005 on the average use of the IT tools among Commission translators, a survey concluded in late 2005 concluded that use of these tools is at least 76 %, while in the ‘new’ EU-10 languages it reaches almost 100 %.

The Commission is currently setting up statistics on the use of IT tools. Work is underway to automatically register actual use and on the setting of other indicators.

The Commission's centralised translation demand management unit has significantly improved its capacities and performance in forward planning, negotiating realistic deadlines and consistency in outsourcing practices but these efforts do not yet encompass the totality of translation activity in the Commission. The forthcoming review of the Commission's translation strategy will strengthen the coordination role of DGT for all translation activities of the Commission.

11.

Since the completion of the audit, the Commission has introduced further changes in its translation practices and procedures which, in its opinion, address the observations of the Court and are highlighted in the Commission's reply to the relevant paragraphs.

13-18.

The Commission has established its procedures for requesting translations to DGT, supported by a web-based application (POETRY) which channels all translation requests from its DGs and services to DGT and which is used by DGT in handling these requests in the light of its translation strategy. Within these constraints, each client service organises its internal workflow to channel its requests through ‘POETRY’.

Given the lack of consistency in the translation authorisation procedures in the Commission, DGT's screening of demand plays an important role in harmonising translation practices among DGs. As of 2005, the authorisations given within the Commission's planning unit are supervised more systematically, especially with respect to categories of documents for which no guidelines have been set in its translation strategy.

The Commission will address the authorisation procedures and the screening of translation requests within its DGs and services and will establish guidelines also for the translation of non core documents as part of its updated strategy for the period 2007 and beyond (scheduled for end 2006). This will strengthen its capacity for a consistent screening of demand.

19-20.

The levels of effective recruitment per 31 December 2004 are to be assessed against the decisions of the budget authority to phase-in resources for enlargement over the period 2004 to 2006.

In spite of staff shortages and fragile structures due to the high numbers of temporary staff, the Commission was able to meet its legal obligations and has therefore satisfied all crucial translation needs for these languages. This has been achieved through consistent demand management, staff in the EU-10 departments working considerable overtime, and heavy reliance on external translation.

The original target of 90 full-time equivalents, set before the transitional translation strategy was adopted, has been revised in line with the estimated impact of this strategy on the demand for translation was set at 75 full-time equivalents (as confirmed by the Commission's APS 2007 decision).

21-23.

The Commission organises its translation activity in its central premises in Brussels and Luxembourg in order to facilitate cooperation between language departments, multilingual coordination as well as management, training and IT support.

The Commission nevertheless needed to set-up translation facilities in most of its representation offices as a temporary measure to complement scarce capacities available during the phasing-in period for the new languages. The role of these facilities is currently being reassessed.

The Commission actively prepared for translations out of the EU-10 languages and in 2004 more than 25 % of total ‘old’ language translators had taken or were taking courses in the ‘new’ languages (216 in training, 87 proficient and 68 partial).

The Commission uses (0,9 % of the total production) translation or revision by native speakers of the source rather than the target language to deal with capacity shortages for some language combinations and increase cost-efficiency.

25.

The rising trend in translation volumes for the Commission was halted in 2004 as a result of the introduction of its demand management strategy. The slowdown of legislative activity highlighted by the Court of Auditors had only a limited impact as large volumes of legislative texts such as agricultural acts, antidumping, competition, State aids, etc., are not affected.

In addition, the anticipated build-up of a backlog of translations into the EU-10 languages did not materialise.

With respect to direct outsourcing and grey translations (referred to in footnote 13), the Commission agrees that currently there are no reliable estimates for the scope of grey translations or direct outsourcing and is therefore launching a survey within its DGs and services to assess the scope of such practices. The result of this survey will be used for a Commission update of its translation strategy for 2007 and beyond.

37-43.

The Commission's translation quality control policy has focused on producing translations that are ‘fit for purpose’ and it has not so far devised a unified quality indicator as such. DGs and services may not always give sufficient consideration to the differentiated quality requirements in accordance with fit-for-purpose principles in assessing the quality of freelance translations. The Commission is currently examining the most appropriate ways of measuring quality.

All freelance translations are evaluated on the basis of a sample check of pages and are rejected if the quality is found on that basis to be unsatisfactory.

The quality problems for outsourced legislative proposals submitted by the Commission for the new languages during the start-up period in 2004 are explained by the limited capacity for internal revision in that period. While autonomous acts were fairly systematically revised, co-decision acts often were not as these would still be read at least once (by Parliament or Council) before publication. This temporary problem has been fully addressed in 2005 and more than 90 % of all in-house and freelance translations is being revised or checked.

The Commission acted on experiences of poor quality of freelance translations, especially for the new contractors for the EU-10 languages, by (i) organising workshops for freelance translators at the beginning of 2005 in the capitals of the new Member States; (ii) ensuring the growing availability of translation tools and reference material and (iii) providing systematic feedback to external translators for all translations not assessed as ‘very good’ and containing more than five pages. As a result, the frequency of quality problems with freelance translations has dropped in the second half of 2005.

44-46.

The improvement in the Commission's respect of deadlines for translations is due to a better translation demand management and pro-active programming.

DGT experiences problems with deadline keeping much more as an issue of improving quality of planning by the services and the institutions, rather than shortages in translation capacity. The Commisison considers translation as a part of the enacting process whereby translation deadlines need to be fully integrated into the programming and DGT is participating in agenda planning and coordination networks. The Commission has now set up a monitoring system on the timely delivery of translations in accordance with the Commission and Council meeting schedules and results so far demonstrate that the translation services offered in this context are compatible with legal obligations and operational needs.

Deadlines set for mid and end of 2004 for translations into the EU-10 languages were indeed longer due to a lack of resources as pointed out under paragraphs 19-20. As of 2005 the deadlines are shorter and programming smoother.

The Commission is launching a survey to quantify the magnitude of grey translation and direct outsourcing and will address these issues in the context of the forthcoming update of its translation strategy for 2007 and beyond.

47-49.

The Commission's costing project referred to by the Court of Auditors was developed in 2001 and applied until mid-2003. It did achieve its objective of raising awareness on the cost of translation. Nevertheless, maintaining this system by providing up-to-date information on the total cost of translations requested was not considered a priority as this would provide little additional value to client DGs in a context where their translation demand was scrutinised in detail against the transitional translation regime.

Following the May 2004 Communication, the Commission shifted its focus from raising awareness on the cost of translation to a pro-active demand management resulting in a significant reduction of total translation costs.

The Commission will propose for its Preliminary Draft Budget 2007 (PDB 2007) the creation of a new policy area ‘31-Language Services’ which will group the budgetary impacts of all language services (but no longer allocated to the different policy areas).

The Commission is currently updating its permanent cost calculation system, in line with the methodology applied by the Court of Auditors, to support management in monitoring the efficiency of the translation process through the evolution of the costs incurred in providing its translation output (and other products).

51-52.

In spite of the near doubling of the number of official languages the total cost of translation for the Commission in 2005 only increased by some 20 % compared to 2003. This was achieved by a strict implementation of the demand management strategy. It nevertheless resulted in a drop in the work rate and a linked increase of the cost per page for the Commission which is of a transitional nature and can be explained by:

The Commission proceeded with some internal redeployments already in 2004 and 2005 and has set more ambitious targets on internal redeployments for DGT in the Commission's APS 2007 decision to adjust fixed costs to changing demand patterns by language groups. Other corrective measures include reducing freelance translations (although outsourcing may be required in view of specific subject matter or language expertise not available internally), two-way translation, the temporary secondment of translators to other DGs, and the expansion of the Commission's Editing Service.

53. (f)

The Commission considers that, as of 2005, the number of translations carried out in full or in part but not needed because of subsequent changes made in the original is marginal due to its policy of not beginning a translation until a stable text exists.

The only exceptions are those documents which, for political reasons, confirmed at the highest level, are subject to tight deadlines that could not be respected if translations were not initiated before the stable document existed.

54.

The Commission is launching a survey to assess the extent of grey translations and direct outsourcing and the result of this survey will feed into the forthcoming update of the Commission's translation strategy for 2007 and beyond.

56-59.

As of 2005 the Commission improved significantly its capacity for forward translation programming: a large part of the Commission's translation workload is covered by agenda planning and regular monitoring with DGs in coordination meetings. DGT's demand management unit maintains regular contacts with translation correspondents in every DG and these efforts have resulted in a situation in which most DGs send regular forecasts, especially when there are important issues at stake.

The need to rely on a more robust programming instrument has been recognised by the Secretariat-General and steps are being taken to integrate translation better in the programming instruments of the Commission.

However, such systems for forecasting demand will always be subject to uncertainties, linked to the political nature of our work, unforeseen events which require new initiatives, etc. Moreover, even when in-house capacities are available, freelance capacity may still be needed for such documents (e.g. with a high degree of technical terminology) for which internal expertise does not exist.

60-62.

DGT fixes deadlines with the requesting DGs in the light of their requirements linked to the destination/purpose of the document (e.g. for discussion at specific Council or Commission meetings). This approach is preferred over tabling deadlines as it allows a more flexible and efficient use of available resources and reduces the need for additional capacity to deal with peak workloads.

67.

Pre-processing in the Commission is highly automated and does not require its central translation planning unit to do any specific search: all documents (except confidential ones) to be translated are checked automatically against Euramis, a huge linguistic database containing thousands of documents divided into sentences linked to corresponding translations and these checks also include the content of Eur-Lex.

68-71.

The organisation of the translation workflow within the Commission's translation units is supported by tools showing the availability of individual translators. However, these tools do not fully reflect all aspects of translator availability such as specialist or language knowledge. Therefore the Head of Unit's judgment remains an important factor.

With respect to the use of IT tools, the Commission actively promotes their use to increase cost-efficiency. Actions in this area have consisted of awareness-raising, training (in particular with a view to enabling end-users to make efficient use of these tools), and ensuring that productivity-enhancing tools run smoothly with other applications (e.g. TWB with LegisWrite). The use of such tools has been constantly advocated by senior management and the Commission issued recently some ‘best practices guidelines’ for its translators and special efforts are also made to train newly recruited translators and secretaries to use computer tools. It remains however partly up to the translator to decide what tools and resources are best suited to his job, depending on the pre-processing reports and his own background, knowledge and experience.

73-74.

The Commission's decision to outsource or not depends heavily on the types of documents which are under consideration. In a situation where a 30 % externalisation target was imposed and translation workload within the Commission was very high, certain decisions were taken to automatically outsource certain documents.

After the introduction of the new demand management policy in 2004, the Commission abandoned its earlier outsourcing targets. There was, however, a certain delay before outsourcing practice changed. In the meantime, outsourcing is largely limited to cases of real need (i.e. to complement in-house capacities). Nevertheless, in exceptional cases where the necessary expertise on specific subject matters is lacking within a given language department, translations will continue to be outsourced, irrespective of available in-house capacity.

75.

The Commission's translation strategy will be extended to include also guidelines on the translation of non-core documents and will strengthen the coordination role of DGT through channelling all information on translation requests through DGT.

76.

The Commission understands ‘productivity’ to refer to ‘workrate’ meaning the production by translator per day taking demand into account, as productivity cannot be measured by simply quantifying the number of pages by FTE.

78.

In comparing translation in the EU institutions with the private sector the following additional factors have be taken into consideration:

The Commission is therefore of the opinion that the cost advantages of external translation do not always outweigh quality-related risks in political organisations such as the European Institutions.

79-80.

The Commission underlines that the evolution of the average work rate per translator for 2004 and 2005 is atypical and not presenting a fair view on translators' performances. It rather is the result of drops in demand creating structural imbalances between demand and supply in the language departments, the lack of experienced translators, terminology bases and linguistic memories for the EU-10 languages, and time spent by EU-10 translators on the revision of the numerous freelance translations (the work rate statistics presented by the Court have not been adjusted for this aspect).

Demand, and the linked work rate, can fluctuate with the level of policy and legislative activity at the Commission. Part of the rationale for a relatively high-cost in-house translation service is to allow coverage of urgent political needs, which implies a certain measure of slack capacity at times when activity is low.

The Commission has taken a number of initiatives in response to the structural decline of demand levels for the EU-15 non-procedural languages:

These measures will be further pursued and the Commission aims to adjust staff levels for all its language departments to real needs by the end of 2006. It will also address the issue of direct outsourcing and grey translations by DGs without any supervision by DGT in the context of its forthcoming update of its translation strategy for 2007 and beyond.

81.

The Commission will intensify its efforts to achieve a greater balance in the workload of the various units by (a) increasing intra-departmental flexibility and (b) streamlining portfolios among the units within the language departments, mirrored by similar structures for the sectors of the demand management unit.

The Commission nevertheless points out that the characteristics and constraints of different language departments should be taken into account in comparing their respective work rates.

87.

The Commission has made significant efforts in 2004 to adjust such statistical systems to cover also relevant indicators for the implementation of the translation strategy decided in May 2004. It is currently developing a Data Warehouse to incorporate more detailed data on translation workflows. A review of certain key statistical concepts such as ‘daily pressure’ is planned for 2006.

88.

The Commission prefers to use a work rate at an aggregate level (per unit/department; averages per translator; etc.) which it considers adequate for monitoring the performance of its translation workflow. Statistics at a more detailed level should be used with care as different factors impact on such a statistic: the availability of translation memories and terminology, the subject matter expertise of the translator, the language competency of the translator with respect to the source language (translators third, fourth or even fifth language), quality assurance efforts in accordance with the fit-for-purpose approach, etc. Such statistics are complex and costly to set-up considering the complexity of the translation workflow.

The Commission's translation data warehouse, currently under construction, will incorporate more detailed data on quality control on the basis of an agreed definition of revision, cross-reading, proof-reading, random sample checking and other quality control operations, and an understanding of the place and purpose of each in assuring a product that is fit for purpose.

With respect to correlations between workload and outsourcing, it should be borne in mind that certain translations sometimes have to be outsourced, regardless of available in-house capacity (also see comment on points 73 and 74).

The Commission has no established performance indicators on the impact of IT translation tools on productivity but has some statistics, which have to be interpreted. Data on yearly production and work rate are available which can be correlated with IT translation tools expenditure. Nevertheless, this correlation is not obvious, since impact is always bigger in the first stages.

89.

The Commission is of the opinion that text segments translated automatically should be taken into account when calculating the impact of investments in IT tools on productivity.

Adjusting the Commission's workload statistics for the impact of translation memories or other translation support tools would represent a significant investment. Other elements impacting on productivity (e.g. the disadvantage of translating from a fourth or fifth language or the advantage of translating a well-known subject) are more difficult to factor-in.

91-95.

The workload balancing scheme was a pilot project to build up mutual confidence. Results were very positive: the Commission offered to the other institutions its full spare translation capacity in the EU-8 languages and agreed with all institutions that the EU-10 translation departments were still not in a position to fully adhere to the workload balancing scheme due to their scarce resources; they will be involved during 2006.

Workload balancing figures in 2006 show a steady and large increase of the total volume. The Commission is fully committed to use and further develop the workload balancing mechanism but notes that a certain margin of internal capacity must always be available to meet unforeseen demand.

The potential of the workload balancing scheme will decrease as the result of ongoing actions to adjust resources allocations for the language departments in line with real needs. These corrective actions remain the first priority for the Commission.

If an average of 20 % of the global workflow is being outsourced by the institutions instead of being exchanged among them, this is mostly due to the lack of coincidence between available idle capacity on the one hand and request for exchange on the other hand, over the whole period needed to meet the request.

The additional savings of 11 million euro estimated by the Court of Auditors relating to freelance activities is not relevant for the Commission which concentrated, as of May 2004, its freelance translations mostly on EU-10 languages, thereby reducing potential further savings of freelance budget to practically zero.

96.

Procedures and tools are in place for channelling translation requests from Commission DGs to DGT and for subsequent handling of such requests.

97.

The Commission has been able to deliver on its political and treaty obligations in terms of translation. It also points out that the acceding countries have the primary responsibility in developing their translation markets and the Commission has repeatedly communicated this message to the respective governments.

99.

The lack of internal capacity to revise freelance translations for translations into the EU-10 languages was limited to the first few months following enlargement and limited to documents which benefited from the safety net provided by subsequent readings in Parliament and Council.

100.

Product quality controls (revision and other checks) on internal translations are performed with a view to providing texts that are fit for purpose, but do not produce a quality indicator as such, and may not be applied to all translations (risk assessment). External translations are evaluated on the basis of a sample of pages and those found of poor quality are subsequently revised.

102-103.

The Commission calculated its translation cost and cost per page for 2002 and suspended subsequent updates of these calculations awaiting the outcome of the audit which is subject of this report.

104.

The drop in the work rate and the structural imbalance of supply and demand for translation across the different official languages result from the impacts of the translation strategy introduced in May 2004. These imbalances are nevertheless of a transitional nature and are currently being addressed.

This strategy has created a basis for significant savings to the Community budget: immediate savings have been obtained by setting recruitment targets for the new languages significantly below historical averages. Additional savings will be obtained by ongoing redeployments out of the old languages and other human resources management actions as well as continuous re-adaptations of the translation strategy.

105.

The Commission has improved significantly its forward translation planning capacity: the network of translation correspondents in the DGs has been revamped to provide specific contact points for DGT's demand management unit and to promote both coordination of translation issues within DGs and upstream translation programming.

Closer attention is now paid to translation aspects at both Commission and interinstitutional level: close follow-up of translations of legislative proposals has been organised in cooperation with the Commission, Council and European Parliament registries. Efforts are being pursued with the SG to further improve translation programming tools in the context of eGreffe. Meetings with the Council Presidency, focusing on the planning of translation and the establishment of priorities, have become a regular feature, thereby contributing to enhanced forward planning. Also, regular inter-services meetings on translation are organised with the Commission DGs.

The Commission is determined to continue on this road in order to further improving its forward programming of translation.

106.

Although some transitional problems did occur in 2004 the Commission has now adjusted its outsourcing policy and is reducing outsourcing to practically zero for language departments with spare capacity. However, outsourcing will be maintained to respond to specific shortages in terms of language competency (i.e. specific language combinations) or subject matter expertise (e.g. international accounting standards, banking directives, etc.).

107.

The Commission is determined to continue its role of promoter and active contributor to the inter-institutional workload balancing scheme. However, it believes that the Court of Auditors' estimate of EUR 11 million in potential savings is not relevant to the Commission (see point 95).

Workload balancing remains subject to restrictions linked to differences in the nature of work in the different institutions, timetables and agendas of the different institutions and the need to maintain some capacity to respond to unforeseen and urgent demand.

109.

The Commission considers that management information to follow-up on the translation cost is adequate. Nevertheless, statistical databases are being updated and improved on a continuous basis to adjust them to developing needs for targeted management information.

Moreover, the Commission favours aggregated statistics which are deemed adequate for monitoring purposes.