Corpus with Speech Function Annotation: Challenges, Advantages, and Limitations
Lidiia Ostyakova, Maria Molchanova, Ksenia Petukhova, Nika Smilga, Daniel Kornev, Mikhail Burtsev · Computational Linguistics and Intellectual Technologies · 2022
Creating a corpus labeled with dialog acts is one of the most difficult tasks in corpus linguistics. Dialog acts can reflect various aspects of the utterances in the dialogues, but most often represent the pragmatic intentions of the speaker or the features of the dialog discourse. The pragmatic and discourse functions of one utterance can be interpreted in different ways depending on an annotator. We used speech function theory that are similar to DA theories to annotate casual conversations and analyzed labeled data. This article is devoted to the first steps and challenges of creating a corpus with speech function annotation.