Illocution on Twitter: The Construction and Analysis of a Social Media Speech Act Corpus
Bell, Laura Beth
Portner, Paul H
In the years since J.L. Austin (1962) first proposed Speech Act Theory (SAT), the literature has taken it in many directions. One recurring point of discussion is the extent to which the direct illocutionary force of an utterance is overtly encoded (Searle and Vanderveken, 1985; Bach and Harnish, 1979; Kissine, 2009; among others). A related question, which has received notably less attention, is the exact identity of the elements proposed to be responsible for such encoding. This dissertation began with the goal of conducting an empirical investigation into the existence and identity of these illocutionary force indicating devices (IFIDs, Searle and Vanderveken, 1985).This investigation was realized through the construction and analysis of a corpus of posts from the social media platform Twitter. The data was annotated, using Amazon Mechanical Turk, for a variety of features including direct illocutionary act category, indirect illocutionary act presence and category, and hashtag functionality. The annotated corpus was analyzed using logistic regression and random forest methodologies, with direct act category as the response variable, in an attempt to identify potential IFIDs. Additional hypotheses that arose during the course of planning the project were also assessed. For example, the corpus was utilized to investigate stylistic accommodation on Twitter as a secondary goal.The corpus analysis identified several features as having a significant effect on the model, though no features were identified by the model as individually significant for direct illocutionary act classification. The dissertation proposes that IFIDs may be realized as templates of typical features as opposed to categorical entities. The analysis also identified several stylistic features to which Twitter users appear to accommodate over time. The dissertation proposes that several of these features may be explained by a style that promotes authenticity and is uncertain about its own level of formality, both attributes of Twitter attested in the literature.
Showing items related by title, author, creator and subject.
Barnes, Michael Randall (Georgetown University, 2019)This dissertation aims to deepen and broaden the philosophical literature on subordinating speech. It aims to demonstrate that how language subordinates is more complex than has been described by most philosophers writing ...