DigiLab - Neue Technologien im Konferenzdolmetschen

Dolterm ist ein auf die Arbeitsprozesse beim Konferenzdolmetschen hin optimiertes, simultanfähiges Terminologie-Datenbanksystem. Es wurde im Rahmen einer Dissertation in Zusammenarbeit mit 40 AIIC-Dolmetschern nach dem Prinzip der agglomerativen Clusteranalyse entwickelt und befindet sich weltweit im Einsatz.

More Info

Abstract:

To activate terminology with the speed required by scientific conferences, interpreters train association clusters. A mind mapping based terminology system offering flexible views of prestructured content is currently being developed as part of a Ph.D. project at the IÜD. The basic structure was developed by professional conference interpreters and today is in fact being used by more than 40 aiic interpreters in the booth. The DolTerm project investigates the problem of organising vast amounts of knowledge and terminology so that they can be retrieved within only a few milliseconds as needed during SI. This marks a radical departure from earlier approaches, which were MT (machine translation) or CL (computer linguistics) oriented. Again, the interpreter is the focus of attention, technology is there to improve his performance, to support his workflow and to improve the working environment.

Notebook computers, ubiquitous in current working situations already host specialised dictionaries, encyclopaedias, Internet Content Trees (Websleuth) and other reference resources. The information is, however, not organised to match the needs during simultaneous interpretation and thus only serves as a reference source during breaks. Efficient and intuitive navigation are expedient to ensure smooth integration of these systems into the concentration allocation patterns[1]. As an illustration of the challenge one might try to retrieve information in a computer on the passenger seat of a car while driving and talking on a cell phone. Early integration of NDM (neural drill modules) into interpreter training will ensure perfect dovetailing of these personalised knowledge management systems, so that they become virtual extensions of the brain. The interface of course is the major problem in this context. Complex onomasiological structures, such as word fields, association patterns and collocation and syntax-based anticipation clusters can only be displayed and navigated using large screen holographic projections on the inside of the booth window. This also creates room to manage competing information streams, such as speaker view, manuscripts and terminological information. By using customisable display options, information is context optimised and displayed selectively. Figures and verbs are highlighted automatically and meta-information on syntax is displayed as thin graphic lines in the text. Unfortunately, technology is not quite adequate in all fields so that this project is still in progress. The Ph.D. thesis will however set forth a concrete requirement profile which may then be implemented by other projects in the respective scientific or academic disciplines, or by industry partners.

The implications for interpreting didactics are obvious: the analysis of professional workflows requires the theory of interpreting didactics to accommodate neural drill tactics. These are currently being developed as part of the DigiLab project at the IÜD. Following a systematic collation of supcompetences in interpreting based on current best practice in interpreting didactics and the relevant literature, training modules were devised. Typical syntactic constellations are one example, they are grouped into ndm modules to automate and expedite solution strategies. Efficient interpreting strategies[2] liberate concentration resources which are then available for active co-ordination of sub-processes. These partial competences are outlined and discussed in theoretical exercises (seminars) and then trained slowly and separately in a virtual language lab. Here, a computer program provides the interpreter with exercises which adapt and evolve along his or her personal progress. The acquisition of interpreting skills is a highly idiosyncratic process: meaningful automated s