Recognition of Manipulative Text: Structure and Dominant Language Parameters

Abstract

The problem of recognition of the manipulative text in the news on the example of Kazakhstani media is posed. One of the most important tasks of text research in the context of manipulativeness is to find something that, although it has a linguistic representation, is obviously pragmatically “obscured”, hidden from the addressee. It is emphasized that in such cases, the danger of manipulation increases at times, since it is addressed both to large social groups and to society as a whole. The relevance of the work is that the destructive nature of manipulation can lead to large-scale social upheaval. The study is based on the hypothesis of a uniform structure of the manipulative text in the context of big data research for further machine processing. Based on the formed working definition of manipulability, an algorithm for analyzing media texts has been developed, which includes a complex of speech and language indicators. The work was carried out on a corpus of 1,000 media texts, selected from 10,000 fragments of manipulative discourses. Both unambiguous and mixed components are distinguished, that is, linguistic units capable of performing dual functions due to their semantic polysemy. A 4-component universal structure of manipulative texts is compiled, including language parameters extracted from manipulative texts.