This study investigates phraseological complexity development in the writing of French learners of English as a Foreign Language from the Longitudinal Database of Learner English (LONGDALE). More particularly, it examines the effects of proficiency and time spent learning English on the phraseological sophistication of verb + object relations, which is measured with mutual information at three data collection points. Results of a mixed-effects model in which we control for potential effects of topic/prompt show that learner proficiency, as assessed by a standardized test, is a better predictor of phraseological complexity in each learner writing sample than the actual data collection point at which the essay was written, thus suggesting that the time dimension does not add anything to, or modify what we already know about, EFL learners’ use of statistical co-occurrences as represented in this longitudinal corpus.

Second Language Acquisition is a complex process, and Learner Corpus Research is increasingly turning to complex statistical methods. While traditional approaches mostly relied on simple frequency counts and monofactorial analyses, some recent studies employ more sophisticated statistics that permit the inclusion of more than one predictor variable as well as more varied kinds of probabilistic/distributional information such as association strengths and dispersion values. One such recent approach is called MuPDAR (Multifactorial Prediction and Deviation Analysis Using Regression; Gries & Adelman 2014, Gries & Deshors 2014). One intriguing aspect of MuPDAR is its extensibility: in its current form, the exploration of the learner data can be L1-specific, but in the present paper we extend this approach towards also including speaker-specific effects. As such, this method should appeal to the growing number of researchers who are interested in investigating individual variation. We present a case study of genitive alternation in the Chinese and German sections of the International Corpus of Learner English alongside English native speaker data obtained from the International Corpus of English, and we illustrate different ways in which the MuPDAR approach could be extended to obtain models that license deeper interpretation at the individual speaker level.

This paper studies the genitive alternation in British English and Sri Lankan English on the basis of more than 4,000 annotated cases of of- and s-genitives from the British and Sri Lankan components of the International Corpus of English. Specifically, we explore the effects of a variety of language-internal and language-external effects, focusing in particular on how these factors affect genitive choices both on their own, but also in interaction with each other and, a first in this kind of variety research, with the gender of the speakers. Our results corroborate previous findings regarding the language-internal factors, but we also obtain a variety of statistical effects representing interactions of those with variety and gender: for instance, animacy effects are stronger in Sri Lankan English, but animacy and length/weight effects are moderated by speaker gender; we discuss these and other findings with regard to processing, language contact and gender (in-)equality. Methodologically, we are developing two innovations for variationist research, namely a principled way to identify and then also visualise the effect of interactions in random forests.

Introduction

This chapter introduces the fundamentals of inferential statistics – that is, methods that help you make inferences or predictions based on your sample data. More specifically, in most empirical studies, researchers cannot study the complete population of a phenomenon of interest – that is, the complete set of objects or speakers of interest – but only a small sample of the phenomenon under investigation. For example, instead of investigating all relative clauses, you investigate a (hopefully carefully constructed) sample of relative clauses in a (part of a) corpus; instead of testing all non-native speakers of a language, you test a (hopefully randomly selected) sample of speakers, and so on. Obviously, you hope that whatever results – percentages, means, correlation coefficients – you obtain from a sample (which you studied) will generalize to the population (which you did not study). However, if researchers draw different samples from the same population and compute point estimates of percentages, means, correlation coefficients, they will just as obviously also get different point estimates; they will encounter variability. The most important application of inferential statistics is to assist researchers in quantifying and studying this variability to (i) arrive at better estimates of population parameters, and (ii) test hypotheses and separate random/accidental from systematic/meaningful variation.

Section 2 will introduce several basic concepts that underlie most inferential statistics. Section 3 presents a set of questions based on Chapter 14 and Section 2 of this chapter that are necessary to identify which statistical test is applicable in a particular research scenario. Sections 4.1 and 4.2 then discuss a small selection of statistical tests involving frequency data of discrete/categorical data and central tendencies (means and medians) respectively.

Introduction

Over the last few decades, corpus-linguistic methods have established themselves as among the most powerful and versatile tools to study language acquisition, processing, variation, and change. This development has been driven in particular by the following considerations:

In this chapter, we will discuss a necessarily small selection of issues regarding (i) the creation, or compilation, of new corpora and (ii) the use of corpora once they have been compiled. Although this chapter encompasses both the creation and use of corpora, there is no expectation that any individual researcher would be engaged in both these kinds of activities. Different skills are called for when it comes to creating and using corpora, a point noted by Sinclair (2005: 1), who draws attention to the potential pitfalls of a corpus analyst building a corpus, specifically, the danger that the corpus will be constructed in a way that can only serve to confirm the analyst’s pre-existing expectations. Some of the issues addressed in this chapter are also dealt with in Wynne (2005), McEnery, Xiao, and Tono (2006), and McEnery and Hardie (2012) in a fairly succinct way, and more thoroughly in Lüdeling and Kytö (2008a, 2008b) and Beal, Corrigan, and Moisl (2007a, 2007b).

Introduction

In Brian Joseph's final editorial as the editor of what many see as the flagship journal of the discipline, Language, he commented on recent developments in the field. One of the recent developments he has seen happening is the following:

For several reasons, this appears to be a development for the better. First, it situates the field of linguistics more firmly in the domains of social sciences and cognitive science to which, I think, it belongs. Other fields in the social sciences and in cognitive science – psychology, sociology, computer science, to name but a few – have long recognized the power of quantitative methods for their respective fields of study, and since linguists deal with phenomena just as multifactorial and interrelated as scholars in these disciplines, it was time we also began to use the tools that have been so useful in neighboring disciplines.

Second, the quantitative study of phenomena affords us with a higher degree of comparability, objectivity, and replicability. Consider the following slightly edited statement: