A computational analysis of hedging in English to Polish translations of film subtitles.

The thesis presents a new pragmatic annotation scheme for the phenomenon of hedging which allows for a computational analysis of its occurrences in English and Polish film subtitles. First and foremost, the definition of the concept of hedging is provided, as well as an overview of its evolution within linguistic studies and its several classifications. The general notions concerning linguistic annotation and parallel corpora are discussed in the following sections, along with some references to the relevance of a computational analysis to other fields of study, such as computer-assisted translation. The second chapter introduces the Opensubtitles and the ParTy corpora and explains the selection of texts which, having been converted into CONLL-U format, are subsequently the object of a quantitative and contrastive analysis of various occurrences of hedges in English original material and its Polish translation. Thus assembled evidence constitutes the basis for the development of a pragmatic annotation scheme specific to hedges. Lastly, the application of the aforementioned scheme to the chosen texts allows for a more thorough analysis and discussion of certain cases presenting the phenomenon in question. The thesis ends with some considerations on the value of the annotation scheme for future study.