Problems around linguistic dispersion, association, and keyness: summary and critique of Gries
What is it about?
How much of corpus-based linguistics research can we trust? Gries (2024) claims to have found serious problems with the statistical measures used. These include measures to identify keywords and measures of dispersion and association, used heavily for studying words (lexicography) and how words/features are used in association (collocation). Gries has also offered some bold solutions. This book review gives a concise summary of the problems and solutions. Afterwards, a critical reflection notes various limitations with reference to simple examples and other research. Areas for further research to improve our methods and understanding are suggested.
Why is it important?
It's easy to get lost in the various arguments and detailed statistics, so a concise overview of the latest research on corpus-linguistic measures is helpful to organise our thoughts. Furthermore, this review makes novel criticisms of this recent research and the dominant way of thinking, allowing corpus linguists to make more informed decisions about their statistics. New directions for advancing the field are also suggested that could provide better solutions to the problems with corpus-linguistic measures.