ABSTRACT
The chapter presents the corpus-driven methodology of researching inverse legal translation, and the selected findings of an ongoing InLeTra (Inverse Legal Translation) research project. The focus of this chapter is on multiword units (MWUs) related to the structure of statutory provisions, observable in non-native English translations of Polish legal acts. Special attention was given to potential untypical collocations, understood as non-standard variants or unnatural frequency of an established MWU (cf. Mauranen 2006, 2007). An attempt was made to discover the reasons for the appearance of untypical collocations and to evaluate the influence they may have on the naturalness of the translation. The comparable-parallel corpus method enabled the identification of several untypical collocations, prevalent in the material. The findings show that the translations generate distinctive phraseological patterns that are predominantly shaped by the interference from the source language. In consequence, the transfer of system-dependent textual patterns into another legal system results in the reduced textual fit of the translations to the non-translated language. Moreover, the results provide evidence for the normalisation and levelling-out hypotheses (cf. Baker 1993, 1996).
