Easy access to good terminological resources is of critical importance to both translators and revisers. This has long been recognised by the SdT, which maintains a staff of terminologists. Their work can loosely be classified as proactive or reactive.
Proactive work involves identifying new terminology, updating and maintaining EURODICAUTOM, working towards greater collaboration with the other institutions and planning for the future.
EURODICAUTOM should, ideally, be a multilingual, multicultural terminology base, easily accessible through user friendly interfaces and covering all Union languages. Although widely acknowledged to be an invaluable tool, in its current state it does not meet those ideals. That this is so follows inevitably from its age: it is based on old technology, the number of languages it should cover has grown over time so that coverage of languages is unequal and a certain amount of indiscriminate up-dating has led to the existence of double or multiple entries for some terms. Nonetheless, it is available to all officials, and to the citizens of the Member States at no cost except the price of communication, is therefore very heavily used and highly competitive with commercial terminology bases. For the future, the challenge will be to modernize without loosing the invaluable information which has already been collected.
Most modern terminologists think in terms of concepts. Creating a new entry involves answering questions like whether the concept being dealt with is stable, whether it is the same across different disciplines and different environments and so on. Such questions are answered through background reading, through research in data bases and through reflection. Several terminologists pointed out that one way in which a terminologists's work differs from that of a translator is that the translator is always under time pressure: he has to find a solution to the problem posed by his current text. The terminologist can afford to search for a good solution, even if that takes some time. Amongst the tools used by the terminologists, the data base of press releases (RAPID) and the various corpora of newspapers were frequently mentioned as being very useful.
A term is accompanied not only by the equivalents in the other languages of the base, but also by a classification code which indicates the subject field in which the term is used. The terminologist is thus also presented with a problem of classification. Subject codes are, by their nature, open to interpretation, and since it is impossible to predict who the user of a terminology base may be, the terminologist has to develop what one of the people interviewed called "a sort of universalism" which will help to maximize the chances that the user's interpretation will match that of the terminologist.
Other information attached to a term includes managerial data which, for example, gives an indication of the age of the information and its validity, and encyclopaedic information which helps the user to identify the nature of the concept underlying the term.
The creation of new terminology is also contracted out to freelance terminologists. As with freelance translation, problems of quality have sometimes been encountered. This is an acknowledged danger of paying for terminology as piece-work.
Terminology is also obtained from other sources through collaboration agreements. Although material from other terminology bases is an obvious source for finding new terminology economically, it can give rise to consistency problems once incorporated into EURODICAUTOM.
Updating a terminology base involves inspecting entries to discover whether the information they contain is correct and is correctly presented. A term may, for example, fall into misuse or change in the way it is used. Here, the questions to be answered remain essentially similar to those involved in creating a new entry.
A special problem is caused by the existence of multiple entries for the same term. A natural desire to be as complete as possible has led sometimes to a new entry being added for a term for which an entry already exists. Notice, though, that this does not necessarily mean that the later entry is redundant. We have just mentioned that a term may change in usage over time, and we noted earlier that the same term may be used rather differently in different environments. Over long periods of time, though, the result can be that the user is presented with a great deal of noise, and has to sort through many entries to find the one that he really wants. Until quite recently, even identifying cases of multiple entry was difficult and time consuming, involving individual searches and a certain hit and miss factor on whether a multiple entry was found or not. The recent implementation of batch mode searching in EURODICAUTOM facilitates identification of multiple entries, and makes it possible to envisage some of the preparatory work for cleaning up the existing entries being done by freelances. Sometimes, when a multiple entry has been found, it is not a question of choosing what entries should remain and what should be eliminated, but of synthesizing two or more of the multiple entries to make a new single entry. This is reported to be difficult but fascinating work.
Reactive work in terminology essentially consists in trying to help the translator who is confronted with an immediate problem. The translator makes use of the SVP service, run by the terminology help desk, whose staff reply not only to linguistic questions but also to questions about the use of existing tools. The help desk staff become expert in the use of EURODICAUTOM, and point out that experience leads to discovering ways of searching, based for example on the subject codes, that the average translator does not know about but which lead to a higher success rate. The help desk staff suspected that translators sometimes did not consult them about how to use EURODICAUTOM because they felt that it was embarrassing to have to ask what they believed they should already know. Other translators, they felt, tried to search for themselves, but became discouraged if they did not find the information quickly, for example because of having used an ill-chosen subject code; hence the suggestion that creating short courses on "clever" ways to search EURODICAUTOM might break down some of the barriers to self-help.
Because the job of the help desk is to be on call in order to react to requests for help, it is clearly not possible to plan the work. There are periods of intense activity, but they are impossible to predict. Most of the help desk staff therefore participate heavily in other activities such as evaluating new products or helping to develop SYSTRAN. We shall return to these activities in section and in section.
As well as the centralized terminology staff concentrated in AGL 3, each language unit has its own terminology correspondent. There are also local terminology bases scattered throughout the SdT, whose creation is sometimes the result of a concerted effort by a unit or a department, sometimes the initiative of an individual translator. Commercially available products for the management of such local data bases are currently under evaluation and we have already noticed the potential of such local bases as sources of terminology. However, several of the people interviewed pointed to the dangers inherent in having a proliferation of local terminology bases at the expense of the central resources, and to the problems of validation associated with sharing which have already been mentioned. Indeed, one of the people interviewed characterized the major problem of terminology management not as a problem of planning but as a problem of how to share knowledge.