Pulling strings: Concordance searches as manifestations of translation problems

Current forms of web-based external support can easily log user activity and provide researchers with large quantities of an additional type of process data. This paper aims to present a multilingual concordancer used by translators at the EU institutions as an instance of this new data source. To analyze concordance data, a methodological approach will be developed to include searches across multiple language combinations. A concordance search will be broken down into individual components that will be analyzed both quantitatively and qualitatively with a special focus on the problem unit and the cognitive aspects of problem solving. Results show that searches can be both static (spot searches) and dynamic (search sessions). Moreover, translators often tend to refine their queries by reducing the initial search and progressively highlighting the actual problematic item.