Previous work on social power modelling from linguistic cues has
been limited by the range of available data. We introduce a new corpus of dialogues,
elicited in a controlled experimental setting where participant roles were
manipulated to generate a perceived difference in social power. Initial results
demonstrate successful differentiation of upwards, downwards, and level communications, using a classifier built on a small set of stylistic features.
Original languageEnglish
Pages (from-to)9 - 26
JournalInternational Journal of Computational Linguistics and Applications
Issue number1
Publication statusPublished - 1 Jun 2015
Event16th International Conference on Intelligent Text Processing and Computational Linguistics - Egypt, Cairo, Egypt
Duration: 14 Apr 201520 Apr 2015


Dive into the research topics of 'Identifying Linguistic Correlates of Power'. Together they form a unique fingerprint.

Cite this