Sketch of a Noisy Channel Model for the Translation Process

The paper develops a Noisy Channel Model for the translation process that is based on actual user activity data. It builds on the monitor model and makes a distinction between early, automatic and late, conscious translation processes: while early priming processes are at the basis of a "literal default rendering" procedure, later conscious processes are triggered by a monitor who interferes when something goes wrong. An attempt is made to explain monitor activities with relevance theoretic concepts according to which a translator needs to ensure the similarity of explicatures and implicatures of the source and the target texts. It is suggested that events and parameters in the model need be measurable and quantifiable in the user activity data so as to trace back monitoring activities in the translation process data.
Michael Carl is a Professor with special responsibilities at the Department of International Business Communication. His current research interests are related to the investigation of human translation processes and how advanced computer tools (such as machine translation) can fruitfully complement and support the human (translation) activities. Furthermore, he is the director of the Center for Research and Innovation in Translation and Translation Technology (CRITT) at IBC.