Pragmatic Markers in Russian Spoken Speech: an Experience of Systematization and Annotation for the Improvement of NLP Tasks

Abstract

Pragmatic markers are an integral part of spontaneous spoken speech, however, they still have no systematic scientific description. These speech elements perform mostly pragmatic functions and are characterized by almost complete absence (or significant weakening) of lexical and/or grammatical meaning. The frequency of pragmatic markers in speech exceeds that of almost all content words. Because of that, for the improvement of many current NLP tasks, it is very important to obtain proper systematization of pragmatic markers and to develop effective and reliable schemes for their annotation. In current research, we describe the preliminary set of pragmatic markers categories and present the results of two stages of their pilot annotation made independently by a group of experts.

Keywords

• speech technologies
• NLP
• Russian
• everyday spoken speech
• spoken discourse
• pragmatic markers
• pragmatics
• annotation