Clustering Language Features: Scaling up from Micro to Macro Variation

• Activity Number: 51 - Making Sense of Discrete Data: Challenges, Inferences and Applications
• Type: Invited
• Date/Time: Sunday, July 28, 2019 : 4:00 PM to 5:50 PM
• Sponsor: Social Statistics Section
• Author(s): Ivy Liu* and Richard Arnold and Miriam Meyerhoff and Shirley Pledger and Lingyu Li
• Companies: Victoria University of Wellington and Victoria University of Wellington and Victoria University of Wellington and Victoria University of Wellington and Victoria University of Wellington
• Keywords: Clustering; Linguistics

Abstract

A fundamental, but as yet untested, assumption in linguistics is that the variation that exists between languages must arise at the level of the individual speaker within a speech community.

A significant amount of variation exists within and between speakers in any speech community. The causes of this variation are manifold -- including differences by age, gender, social status, social setting, contact with other languages, other external influences as well as spontaneous innovation.

An interesting question is whether the variations that most strongly distinguish languages from each other are the same as, or at least analogous to, the strongest variations present within individual speech communities.

To test this we have data on variables charaterising and differentiating the speech of 19 individual speakers of Bequia Creole English, and map these onto variables that differentiate varieties of English. We use clustering methods to identify the variables that characterise the variation between the 19 speakers and carry out the same clustering on these variables coded for 76 varieties of English. We compare and discuss the outcomes of these two clustering exercises.