Extended Translation Memories for Multilingual Document Authoring
Jean-Luc Meunier, Marc Dymetman
This paper discusses multilingual document authoring, viewed as providing computer support for a user to author a document in some source language while automatically generating the same content in one or many target languages. A kind of unanticipated use of multilingual authoring appeared in the service sector, in situations where an employee is servicing customers by answering their requests, or helping them, via written electronic communication. Decoupling the employee’s language from the customer’s language may open up new perspectives and motivated this work, where we propose a small set of extensions to be made on a translation memory to support multilingual authoring more efficiently. We describe how an instance of such extended formalism can be conveniently created thanks to a domain specific language and describe how we implemented a full system. Finally, we report on the experiment we ran in a real business setting.
BUCC, 7th Workshop on Building and Using Comparable Corpora, Reykjavik Iceland, 27 May 2014.
2014-005.pdf (629.89 kB)