Great new article about random speaker effects in sociolinguistic data analysis

Posted on Updated on

Heya. I just wanted to bring the following nice article by Daniel Ezra Johnson to everyone’s attention:

Getting off the GoldVarb Standard: Introducing Rbrul for Mixed-Effects Variable Rule Analysis,
Daniel Ezra Johnson , University of York,
Language and Linguistics Compass 3/1
(2008): 359-383, doi: 10.1111/j.1749-818X.2008.00108.x
PDF:  http://www.blackwell-synergy.com/doi/pdf/10.1111/j.1749-818X.2008.00108.x

The article addresses the need for random speaker effect modeling in sociolinguistic data analysis and why researchers should switch from a Goldvarb standard to mixed effect models. The paper also describes an implementation available in R (Rbrul) that does affords both ordinary and multilevel regression modeling and is capable of formatting output in ways that either follow standard regression conventions or the Varbul standard which is more common in sociolinguistics and variationist work. I think the paper is really well written and provides some compelling arguments to use the more advance mixed effect models. Spread the word. There are still plenty of people out there who are hesitant to leave Goldvarb behind despite the obvious shortcoming that it does not support random effects.