Doing Linguistics with a Corpus

We propose approaches that help place language description back at the center stage of quantitative corpus linguistic research.

This Element explores ways to improve how we approach linguistic research questions with quantitative corpus data. We introduce and illustrate the major steps in the research process, and provide suggestions for improvements that will help researchers make quantitative corpus-based research more language focused and linguistically informative.Paradoxically, doing corpus linguistics is both easier and harder than it has ever been before. On the one hand, it is easier because we have access to more existing corpora, more corpus analysis software tools, and more statistical methods than ever before. On the other hand, reliance on these existing corpora and corpus linguistic methods can potentially create layers of distance between the researcher and the language in a corpus, making it a challenge to do linguistics with a corpus. The goal of this Element is to explore ways for us to improve how we approach linguistic research questions with quantitative corpus data. We introduce and illustrate the major steps in the research process, including how to: select and evaluate corpora, establish linguistically-motivated research questions, observational units and variables, select linguistically interpretable variables, understand and evaluate existing corpus software tools, adopt minimally sufficient statistical methods, and qualitatively interpret quantitative findings.