sunnyJy
commented
6 years ago (1) I think sentimental analysis may be one case of studying tone changes. It belongs to the third collection of the studies. Page 41 of Evans and Aceves (2016) lists papers studying on both individual and collective levels. Sudhof et al (2014) traced users' sentiment shift. Does it use any techniques you are looking for?
(2) "how different parts of the text support or negate other parts of the text?" I think this belongs to the analysis of higher-level linguistic discourse. Page 28 mentions that this is a limitation of current content analysis: "Another limitation relates to the lack of sophisticated models for higher-level linguistic discourse, such as how sentences relate to one another (Stymne et al. 2013) and aggregate into paragraphs and more or less effective arguments, although this is one of the targets of sociological text analyses (e.g., message complexity and popularity in Bail 2016)."
This was a very interesting read on the state of text analyses in sociology (very good overview for someone without a sociology background!). Most the studies are done at the micro level (n-grams, microinteractions, sentence structure, etc), and this seems to be predicated on the assumption that each text is more or less monolithic. Do we have to tools conduct analyses at a higher level, with more nuance? For example, how the tone of a text changes from the beginning to end, or how different parts of the text support or negate other parts of the text? And further, how effective (however this can be measured) these techniques are?