In this chapter we discuss the results of the experimental component to the study, which is designed to examine the degree to which the probabilistic contrasts uncovered in the corpus analysis are also discernible in language users’ introspective preferences. These experiments were designed to corroborate and interrogate the scope of the patterns observed in our corpus models.

This chapter begins with a general discussion of potential data types in variationist linguistics. Next, we present the two main data sources we use in the study: the International Corpus of English (ICE) and the Global Corpus of Web-Based English (GloWbE). The former comprises a set of parallel, balanced corpora representative of language usage across a wide range of standard national varieties. Each ICE corpus contains 500 texts of 2000 words each, sampled from twelve spoken and written genres/registers, totaling approx. 1 million words. GloWbE contains data collected from 1.8 million English language websites – both blogs and general web pages – from twenty different countries (approx. 1.8 billion words in all). Discussion of the corpora is followed by a detailed description of the data collection, identification, and annotation procedures for our three alternations. Here we carefully define the variable context for each alternation, and outline the methods for coding various linguistic constraints that are included in our analyses.

This chapter interrogates corpus data to analyze the three alternations subject to study in this book one by one using a battery of state-of-the-art analysis techniques in addition to customary descriptive statistics, Conditional Random Forest (CRF) modeling and mixed-effects logistic regression analysis. The goal of the chapter is to uncover qualitative generalizations: for example, we see that while effect directions of constraints on variation are generally stable across varieties of English, effects strengths can be significantly different.

This chapter is inspired by work in comparative sociolinguistics and quantitative dialectometry. We use a corpus-based method (Variation-Based Distance and Similarity Modeling – VADIS for short) to quantify the similarity between, and coherence across, the varieties of English under study as a function of the correspondence of the ways in which language users choose between different ways of saying the same thing. Key findings include the result that probabilistic grammars are remarkably stable across varieties but that coherence across alternations is not perfect.

This chapter surveys the literature on variation in general and on grammatical variables (a.k.a. "alternations'') in particular. Next, we review well-known grammatical variables/alternations in English as well as previous comparative investigations of grammatical alternations in English. Last but not least, we discuss in detail previous variationist work on the three alternations subject to study in this book: the genitive alternation, the dative alternation, and the particle placement alternation.