Dream tools

During the interviews many ideas for useful tools were suggested, either in the course of the conversation or in response to specific prompting. In a few happy cases, the tools already existed, and it was possible to pass on information about them either directly or by asking the SdT coordinators of the study for help. Great efforts are made within the service to ensure that information about what is available circulates freely, but the sheer size of the SdT, coupled with the complexity of its organisation and the heavy workload inevitably means that sometimes information does not get through to the person who needs it. It was pleasing to be able to offer useful information from time to time.

Some of the tools suggested were already being studied by the computing services or by AGL 4. Others, although they may not yet exist on the market, are perfectly feasible in the state of current technology. Still others would require a research effort for their development, but are imaginable in the current state of the art. Yet others are tools for the future: they would require major breakthroughs to find solutions to problems that until now have proved intractable.

Some of the ideas are not tools at all, but changes which would facilitate the work of the SdT (or in one case, eliminate it!).

It is perhaps worth noting that willingness to suggest new tools was strongly associated with an awareness of the usefulness of those tools which exist already.

The ideas suggested are simply collected together here, with no attempt to say what already exists, what is feasible and what is not.

• A single language for Europe.
• An ergonomic advice sheet: on this would be recorded all the data needed for an individual to create a work environment tailored to his physical needs: e.g., required height of chair, slope of chair back, height of table, position of screen, etc.
• A standardized complete modernization of administrative language.
• Full standardisation of texts.
• A standard observed even for relatively minor features of a text, like how footnotes are treated.
• Easy access to information available on CD-Roms, e.g. corpora, specialized dictionaries, encyclopaedias.
• Easy access to any data base which exists.
• Make sure that officials in the requesting departments also have access to terminology, to background documents and to reference documents.
• Persuade Council to accept electronic documents.
• Allow teleworking from anywhere in the Union.
• All staff completely networked.
• Tools that would facilitate translators working together in a group.
• Access for freelance translators to all the resources available to in-house translators.
• A scheduling programme which works out when a job should be started so that it can be finished on time, and does it dynamically to take account of new work coming in. Each day it would signal what should be done that day.
• A tool that would allow profiles of individual users to be set up, so that interfaces could reflect the individual's own interests and what he uses.
• A multilingual spelling checker which automatically recognizes what language is being dealt with and switches to the appropriate spelling checker for that language.
• A spelling checker that learns what mistakes the particular user makes and corrects them automatically.
• A tool that would check the quality of a freelance translation and set a warning flag if the translation should be thoroughly checked. It was suggested that using existing spelling and grammar checkers might be a point of departure for such a tool. It was also suggested that looking for characteristic translation errors might have potential.
• A style checker tailored to Commission jargon and style.
• A grammar checker that could learn a house style.
• An editing pool which would take documents on their arrival in the SdT, would sort out any problems in the original and would do as much preparation as possible before the document was passed onwards for translation.
• A multilingual authoring tool. The user would construct his text by putting together predefined textual elements chosen from a list of such elements. The system would generate equivalent texts in all other Union languages.
• Voice recognition for computer commands.
• Voice recognition for e-mail addresses, thus avoiding mistypings of non-natural strings.
• Good voice dictation systems for all languages
• A system that would associate spoken target language templates with words in the original text. This would facilitate dictation.
• A tool combining speech input and conventional text-processing: when revising, for example, one could use the mouse to highlight the section to be replaced and then dictate the modification.
• A system that would take a text, discover what all the relevant reference material was, generate that material and make it available.
• Inter-Institutional collaboration on reference materials.
• A system that would extract new terms and identify their equivalences.
• A tool that would check for the consistency of terminology across a text or series of texts.
• A dedicated computer scientist as part of the terminology staff.
• A tool that would identify all the repetitious elements in a text, and maybe identify what texts are suitable for preprocessing.
• A tool that would be able to recognize recycled text, even when there are changes in the order of text elements or minor changes of vocabulary.
• A tool which would examine a text and give advice on what other tools could be used on it, for example Tman preprocessing, machine translation, term extraction, translation memory.
• A translator's workbench with a translation memory which would find all terms for which a validated translation exists.
• A hierarchically organised terminology search system, giving preference first to validated terms, then to EURODICAUTOM terms, then to terms recovered from translation archives.
• A data base of external experts who can help with terminology, what their specialities are and how to contact them.
• A translation memory which would find all the old translations relevant to the current translation. Where the individual user had a preference for a particular translation, this would be signalled and used.
• Training courses for new staff to teach them how to use SYSTRAN in practice.
• Machine translation even of inferior quality from the lesser-used languages into English and French.
• A tool which integrates the best possible translation memory system with the best possible machine translation system and is then used as an aid to human translation.
• Fully automatic high quality translation.
• Machine translation with voice input.
• A CD-Rom jukebox giving access to all COM, SEC and C documents.
• A Commission-wide fully integrated document management and archiving system. The ideal version of this tool would include:
  • Every document and every translation ever produced, whether on paper or electronically
  • cross-indexing to show relationships between documents
  • all documents multilingually aligned
  • facilities for keyword search
  • facilities for content-based search
  • terminology resources, including full information on each term, such as
    • who uses it
    • how often
    • what the reference is
    • grammatical information
    • "views" to allow human use or computer use of the same information
• A project to think about all the different ways that massive amounts of linguistic material could be searched and used.
• An electronic thumb-print for every document so that it could be uniquely identified for ever.
• The ability to search for documents with a similar thumb-print to the one you are immediately dealing with.
• A tool to carry out consolidation of legislation which has been modified by other legislation, modifying each change in each text.
• An eye-pointing device to position the cursor on the screen.