Casey Whitelaw & Maria Herke-Couchman

Macquarie University, University of Sydney

Tackling the system: Using state of the art computational techniques to automate systemic choices of interpersonal distance

The challenge of automating the complete tool kit available to the manual Systemic Functional analyst is enormous. However, given the time demands of comprehensive manual analysis, it is only by meeting the challenge and breaking the barrier into high volume text analysis, that we can both verify our registerial hypotheses as well as augment our instantial understanding of texts with a comprehensive understanding of the system.

Just as our understanding of the instance in terms of the sf framework continues to positively impact literacy education, a comprehensive understanding of the system promises an even more powerful influence on issues directly contributing to literacy.

This paper reports on research that attempts to take a micro-step toward tackling the computational challenge. Focusing on just the pronominal and determination system and the role it plays in constructing interpersonal distance, we propose a simple computational model of SFL as the basis for extracting simple semantic features from documents. Using machine learning techniques and three corpora with characteristic and distinct interpersonal distance: newswire, transcripts of spoken text, and ""Nigerian emails"", we demonstrate that the systemic features give clear separation between registers with different interpersonal distance.

This is an encouraging result for our future work which will seek to extract richer semantic features using SF theory.