On the Perils of Linguistically Opaque Measures and Methods: Toward Increased Transparency and Linguistic Interpretability

In studies whose primary goal is linguistic description of characteristics of features, we have to do our utmost to ensure that we measure and analyze said features in a way that is linguistically interpretable. A measure is considered linguistically interpretable (a) “when its scale and values represent a real-world language phenomenon that can be understood and explained” (Egbert et al., 2020, p. 24) and (b) when all variables have clear operational definitions. The present chapter covers a discussion between Tove Larsson and Douglas Biber at Northern Arizona University, during which they argue in favor of using linguistically interpretable measures and techniques. Specifically, the chapter covers discussion of four potentially problematic issues in this context, along with a conclusion. The subsections start with a brief description of or background to the issue, followed by answers to the questions posed.