Guest post: Linguistics & cybersecurity
Our guest bloger is Alex Krendel who graduated from our BA English Language & Linguistics and MA Applied Linguistics. She will be starting her PhD at Lancaster University this autumn to study how gender and socio-political beliefs affect online impoliteness.
I attended the Behavioural and Social Sciences in Security (BASS18) conference in Lancaster this summer to discover the links between cybersecurity and linguistic research. I attended talks on Operational and Linguistic Information Processing, Cybersecurity and Actors, Understanding Who and Why, and Analysing Communication. However, as possibly the only linguist attending (as a guest or as a presenter), I was a fish out of water! Despite this, I found plenty of research areas in cybersecurity that would be improved by the input of enthusiastic linguists.
Much of the presented research included corpora built for the purposes of machine learning i.e. training machines to use Natural Language Processing (NLP) to detect pro-terrorist social media accounts. As a result, there is room for computational linguists to get involved with NLP for cybersecurity! However, these NLP systems often solely focus how often certain words appear, while often ignoring grammatical features, syntax and punctuation. Using the corpus analysis skills taught in the Applied Linguistics MA Researching Language In Use and Forensic Linguistics modules at Sussex, a linguist could build a corpus of online hate speech, and apply existing linguistic methods of content analysis such as conversation analysis and appraisal theory to the data.
I noticed much of the research presented at BASS18 was heavily quantitative, which leaves a handy research gap for linguists with a love of qualitative discourse analysis methods! For instance, one researcher analysed ISIS propaganda magazines for key themes discussed in the text and what words have positive and negative emotion associated with them. Following this analysis, she used these criteria to automatically identify pro-terrorist Twitter accounts using machine learning.
In fact, many content analyses carried out by the researchers at BASS18 reminded me of the BA Discourse of Social and Personal Identity module at Sussex, as psychologists at BASS18 analysed how communities of online hackers interact, and how right-wing extremists respond to terror attacks on right-ring forums. Additionally, the link between language and gender was explored to an extent, as one talk explored how extremist groups perceive women, and attempt to persuade women to join their cause.
At a glance, the biggest potential linguistic contributions towards cybersecurity are computational linguists improving NLP, and discourse analysts using methods that computer scientists and psychologists do not know about. However, there were other talks I unfortunately could not attend as I do not own a Time Turner. Talks from BASS18 that I missed looked at police interview techniques, lie detection, and detecting security threats from within an organisation – all topics that a conversation analyst or a grammarian would devour! The possibilities are endless.
I returned home with one very strong takeaway message: linguists need to be attending these conferences, making new contacts, and showing psychologists and computer scientists how linguists can contribute to the important work they are carrying out.