Files in this item

FilesDescriptionFormat

application/pdf

application/pdfHINDERLITER-DISSERTATION-2016.pdf (2MB)
(no description provided)PDF

Description

Title:The evolution of online asexual discourse
Author(s):Hinderliter, Andrew Clinton
Director of Research:Terkourafi, Marina; Shih, Chilin
Doctoral Committee Chair(s):Terkourafi, Marina
Doctoral Committee Member(s):Girju, Corina Roxana; Sadler, Randall W
Department / Program:Linguistics
Discipline:Linguistics
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:Ph.D.
Genre:Dissertation
Subject(s):Corpus linguistics
Asexuality
Abstract:Technological, social, and economic changes in recent decades have led to new possibilities for communication and for forming communities that are not tied to a specific geographical location. This creates new opportunities and challenges for studying language change, including the language of online communities. This dissertation provides such a case study by examining the development of, and changes in, online English language asexual discourse from the second half of the 1990s until late 2013, focusing on lexical items and multi-word expressions. This dissertation combines three major research approaches---archival research, corpus research, and survey research . Using the Way Back Machine, databases of newspaper archives, and academic databases and references, historical conceptualizations of asexuality can be seen well before the emergence of asexual communities online, but I can find no evidence of asexual organizing prior to the 1990s. Largely using qualitative analysis of asexual websites, I give a historical account of the development of online asexual communities, and I argue that there have been at least two major conceptual shifts in the conceptualization of asexuality in the time period under consideration, which I call the "AVEN shift" and the "rise of intermediate categories." I then discuss the construction of four corpora: I scrapped the largest asexual website (Asexual Visibility and Education Network [AVEN]) and a similar sized message board on a different topic as a control (Non-asexual corpus). I subdivided the AVEN data into two sub-corpora, based on the sub-forum topics. Most subforums are in the AVEN-core corpus. Some (e.g. "Just For Fun" or "Off-A") are mostly about topics other than asexuality, and were grouped as the AVEN-other Corpus. In addition, I scraped several asexual blogs and asexual communities other than AVEN (e.g. a LiveJournal community): these comprise the Asexual-other corpus. Using a multinomial Naive Bayes classifier, I found moderate distinguishability between the AVEN-main and non-asexual corpura at the level of individual posts when only considering individual words. To rule out the possibility that the classifier was distinguishing AVEN vs. non-AVEN discourse, I first created an algorithm to remove from consideration words that probably refer to users. Second, I used the same classifier on the AVEN-other corpora and the asexual-other corpora. Results for the asexual-other corpus are similar to results for the AVEN-main corpus, while results for the AVEN-other corpus are not. This suggests that the classifier is identifying asexual discourse vs. other discourse. I used the AVEN-core Corpus to generate a list of "key-words" that well-characterize asexual discourse, and then investigate the evolution of three sets of these: intermediate category terms, romantic orientation terms, and the terms repulsed and indifferent. Results provide strong support for the "rise of intermediate categories" hypothesis, and also provide evidence of terminological change for the other two domains. To test the "AVEN shift" hypothesis, I conducted an online survey in Early 2012 about people's self-understanding prior to and after finding an online asexual community. Results provide evidence for all predictions of the "AVEN shift" hypothesis, although the changes were by no means monolithic. Through this research, I illustrate the utility of applying methodologies from corpus linguistics and from machine learning for investigating the language of specific online communities. Further, I provide novel methodologies (or novel uses of existing methodologies) for problems likely to be faced by other researchers using corpus linguistics to study online language change in specific (sub)communities.
Issue Date:2016-12-02
Type:Thesis
URI:http://hdl.handle.net/2142/95433
Rights Information:Copyright 2016 Andrew Hinderliter
Date Available in IDEALS:2017-03-01
Date Deposited:2016-12


This item appears in the following Collection(s)

Item Statistics