Using Regular Expressions in Translation Memories

Abstract. The paper presents the idea of using regular expressions in translation systems based on translation memories. Regular expressions are used in: search for a match in the instance sentence, search for an appropriate example in the translation memory and in the transfer of the instance sentence into its equivalent. The application of transfer rules to translation memories support the thesis, put forward in the paper, that Machine Translation and Computer-Aided Translation converge into the same direction. “While the original aim of the MT pioneers was fully automatic MT systems, there has also been, at least since the 1966 ALPAC report and possibly even earlier, the view that computers could be used to help humans in their translation task rather than replace them”. This citation comes from Harold Somers (Somers, 2003). The below figure may illustrate this view: The idea of co-operation between man and machine goes in two directions: one direction is via Machine Translation (MT), possibly pre-edited or/and post-edited, towards Fully Automatic Machine Translation (FAMT) where there is nothing to do for a human. The other direction is via Computer-Aided Translation (CAT), where the user – human translator – is handily equipped with necessary glossaries and dictionaries and has the possibility to re-use his/her own translation, towards the use of large Translation Memories where there is almost nothing to do for a computer (except for just matching an instance text to an example that has once been translated). It appears that the idea of Computer-Aided Translation is nowadays beating the idea of Machine Translation. The reason is mainly economical: The creation of a good-quality MT system is much more costly than that of a good-quality CAT syste