Part-of-Speech and Pragmatic Tagging of a Corpus of Film Dialogue: A Pilot Study

Abstract

AbstractThis article presents how a pilot study for automatically POS-tagging a corpus of orthographic transcriptions of film dialogues (Pavia Corpus of Film Dialogue) was dealt with and the related issues solved. The software CLAWS4, which is freely available on UCREL’s website, was used for the sake of comparability with reference corpora such as the BNC (both 1994 and 2014) and all the English corpora available on english-corpora.org (former BYU interface). The study highlights that automatic POS-tagging needs readjusting when applied to film dialogue and the accuracy of the tagging greatly benefits from the introduction of tags for pragmatic categories. This integrated approach of grammatical and pragmatic automatic tagging was realised through the writing of a Python script which post-processes the data output of CLAWS4.