That’s Cool. Computational Sociolinguistic Methods for Investigating Individual Lexico-grammatical Variation

The present study deals with variation in the use of lexico-grammatical patterns and emphasizes the need to embrace individual variation. Targeting the pattern that’s adj (as in that’s right, that’s nice or that’s okay) as a case study, we use a tailor-made Python script to systematically retrieve grammatical and semantic information about all instances of this construction in BNC2014 as well as sociolinguistic information enabling us to study social and individual lexico-grammatical variation among speakers who have used this pattern. The dataset amounts to 4,394 tokens produced by 445 speakers using 159 adjective types in 931 conversations. Using detailed descriptive statistics and mixed-effects regression models, we show that while the choice of some adjectives is partly determined by social variables, situational and especially individual variation is rampant overall. Adopting a cognitive-linguistic perspective and relying on the notion of entrenchment, we interpret these findings as reflecting individual speakers' routines. We argue that computational sociolinguistics is in an ideal position to contribute to the data-driven investigation of individual lexico-grammatical variation and encourage computational sociolinguists to grab this opportunity. For the routines of individual speakers ultimately both underlie and compromise systematic social variation and trigger and steer well-known types of language change including grammaticalization, pragmaticalization and change by invited inference.