Identifying the conversational structure of online communications between child sexual abusers

Computer mediated communication (CMC) plays an important role in shaping our social and professional interactions, as a great part of those is carried out online. It is employed in a variety of situations, such as building community, finding support, education, and many other functions. CMC has multiple points in common with face to face communication as it exhibits oral features, and multiple dissimilarities such as less coordination regarding sequence organisation and the obvious lack of prosodic and behavioural cues (Meredith, 2019). The discussion around the concept of CMC is interesting and fruitful and online conversations are noteworthy as a conversational phenomenon with context-specific characteristics (Herring, 2010).

In this project, I aim to combine Conversation Analysis and Corpus Linguistics in order to analyse online conversations. Both Conversation Analysis and Corpus Linguistics examine naturally occurring data, and the former analyses linguistic phenomena through the study of large collections of such data. Conversation Analysis is the analysis of talk as action. It views actions as constructed and understood within an interaction and analysed through a close examination of sequential organisation, turn-taking, adjacency pairs, and repair strategies (Benwell and Stokoe, 2006). It is a method with roots in ethnomethodology and, while initially used solely for analysis of spoken conversation, it has expanded to the examination of online interaction (Meredith, 2019). The combination of Corpus and CA is valuable as it would enable researchers to analyse actions systematically and extensively (Housley et al, 2019). Thus, the research objective is to complement the two methodologies making it possible to carry out sizeable corpus analyses that are informed by CA and to support CA using corpus-based methods.

This combined methodology will be devised through the analysis of various sources of online conversations. The existence of CMC allows researchers to gather big, searchable datasets and projects with large scale CMC conversations corpora have been carried out by NLP and AI (Mehri and Carenini, 2017). A substantial CA and Corpus project would contribute to the literature in novel ways, by focusing on actions. The datasets used will be, among others, the Ubuntu Chat Corpus for Multiparticipant Chat Analysis (Uthus and Aha, 2013) and a corpus of authentic Dark Web chat logs comprised of communication between child sexual abusers exchanging illicit imagery.

This project will have important methodological applications, as the aim is to enable researchers to study actions on a large scale. There are also important practical applications, for example insights on multiparticipant chatrooms are beneficial for academics and practitioners that work on Dark Web data relating to child sexual abuse, terrorism, drug dealing or arms trading (Weimann, 2016; Faizan and Khan, 2019). Moreover, work on multiparticipant chats is central to understanding how environments that present an overload of information work, which can be of vital importance in educational or military settings (Shanker and Richtel, 2011).