Much has already been said about the computing services and the support work undertaken by the AGL Directorate. In this section we summarize briefly the roles of those units most directly concerned with the specification, choice, implementation or integration of new computing tools.
Let us start with SdT-03, the small unit responsible for modernisation of work methods. As its name implies, it is the job of SdT-03 to look at the organisation of the work of SdT as a whole, analysing both the activity of the Translation Service and the administrative, organisational and technological environments within which this activity takes place, and to follow how these environments are evolving and changing. The unit also keeps an eye on what is happening elsewhere, in other organisations and institutions. On this basis, the unit tries to work out what the possible options are for improvements throughout the SdT, and to encourage a favourable attitude towards change and innovation.
What this boils down to is trying to rethink radically the whole process of translation, seeing it as in symbiosis with computer tools. Critical questions concern the relationship with the requestors and questions of how resources can be shared. This latter can be restated as trying to download as much as possible of the translator's know-how into a form that can be shared by others. Here, it is important to distinguish between recurring problems, where the problem and its solution can simply be recorded for future use, and more complex problems which are perhaps the domain of the specialist. The former involves above all trying to improve the communication channels, so that information does not get lost by being passed through a long and complex channel.
AGL 4 is responsible for the development of multilingual tools. In collaboration with AGL 3 (see below) and in consultation with SdT-03, AGL 4 studies what translators' aids tools are needed within the SdT, draws up functional specifications for preprocessing tools, authoring aids and standardisation tools, as well as for the terminology data bases. In collaboration with D.G. XIII, AGL 4 contributes to the development of machine translation systems in conformity with user needs. The Unit is also responsible for the choice of computer tools and linguistic resources, and for following their integration into the SdT environment. These responsibilities imply a need to keep abreast of developments in the language industry and in research. In the latter context, AGL 4 coordinates the SdT's participation in the language industry related RD programmes managed by D.G. XIII and application of the results obtained by such programmes.
AGL 4 works very closely with SdT-02, the computing services, who as we have seen, are responsible for all the information technology in the SdT (including faxes and photocopiers, although not telephones). The job can be summarized as keeping the machines and the users operational, through maintenance, help desks and training. SdT-02 is also involved in the choice of new commercial products to be tested, their testing and their integration into the computing environment of the SdT. One problem repeated here was mentioned frequently by many of the AGL and computer staff: many, if not most, of the translators' aids tools available on the market are aimed at the single user. Adapting them to use in a very large service where information needs to be shared and double work avoided can be far from straightforward. We have already seen some examples of this problem in discussion of, for example, the use of translation memories. A second problem is that of integration, which is important both from the point of view of sharing resources and of user-friendliness. The user will typically have a work station with access to office tools, management tools and translation tools. It is important that these are all integrated in such a way that the same manipulation, typing ``exit'' for example, always means the same thing. This is a huge job, since different products are developed independently, and each manufacturer will at best strive for compatibility only between his own products.
Market forces also have an effect, sometimes undesirable, on the effort to keep all languages equal from the point of view of computer treatment. Text processors provide a good example here. English text can be dealt with through an English interface, and French through a French interface, but Finnish and Greek are typically dealt with through an English interface.
A PC-based client-server architecture also creates some tensions: an increasingly computer literate population of users is aware that almost anything can be done on a PC. But in a large organisation where resources have to be shared, some constraints have to be accepted if sharing is to be effective. It is not always easy to give the users what they want and still respect the constraints that make best use of the resources available, whilst still at the same time taking into account that another part of the targeted population is not at all computer literate.
We have already noticed that the size of the organisation has effects on the speed of change that can be accepted. In the interests of preserving a certain stability that will allow the user to be maximally productive, computerization in the SdT must always be slightly behind the state of the art whilst remaining keenly aware of likely future developments. This too is a delicate balance to maintain.
AGL 3 is the final link in the chain between the support services and the users. Its function is to make available a coherent set of language services and computerized tools to the translators and to the other officials of the Commission. Thus, the Unit provides a linguistic help desk, which can offer help with terminology and other linguistic problems, manages and develops the linguistic resources provided through central and local terminology bases, and helps to create new linguistic resources which will be integrated into translators' aids. The Unit is responsible for encouraging the use of language engineering tools and for their distribution, as well as for collaboration with the other institutions on harmonisation and exchange of terminology.
Before leaving this section we should mention the work of the Modernisation Committee and its successor, the Follow-up Committee, to whose work the groups mentioned in this section contributed very largely. The modernisation Committee was set up in 1993 and decided on a number of projects, including the creation of manuals for the data bases available, the introduction of e-mail throughout the Translation Service, the creation of topical glossaries and a project on rationalisation issues. The Follow-up Committee was created to follow the work of these projects. In the summer of 1995 it was felt that the aims of most of these projects were either achieved or well on their way to achievement, and that it was appropriate to make an assessment of the results of each action, and of where the results could be found. This has led to an information effort, in the form of a series of two to four page long documents (on yellow paper, in order to draw attention), under the name of ``Mini-bilan'', which attempt to summarise the results fo each action.
Currently, a slimmed down version of the Follow-up Committtee is looking at future modernization issues. The aim here is not to redesign the future shape of the SdT, but to look at practical measures like the introduction of POETRY into the requester services, starting with one Department and trying to ensure that every document coming in comes through PEOTRY, or trying to find alternative ways of easing the problem of information transfer which has been mentioned many times in this study and is widely acknowledged to be a difficult issue.