Research on bilingualism as discovery science

Highlights

• Science is a discovery process that acts on variety.

• Prescriptions about power and method undermine the discovery process.

• Exploration of variety generates incremental progress.

• Research on bilingualism concerns the identification of bilingual phenotypes.

• Effective phenotyping requires sensitive tools informed by rich characterization.

Abstract

An important aim of research on bilingualism is to understand how the brain adapts to the demands of using more than one language. In this paper, we argue that pursuing such an aim entails valuing our research as a discovery process that acts on variety. Prescriptions about sample size and methodology, rightly aimed at establishing a sound basis for generalization, should be understood as being in the service of science as a discovery process. We propose and illustrate by drawing from previous and contemporary examples within brain and cognitive sciences, that this necessitates exploring the neural bases of bilingual phenotypes: the adaptive variety induced through the interplay of biology and culture. We identify the conceptual and methodological prerequisites for such exploration and briefly allude to the publication practices that afford it as a community practice and to the risk of allowing methodological prescriptions, rather than discovery, to dominate the research endeavor.

Introduction

Research on bilingualism generates debate on the neural bases of language that address fundamental questions about language learning (e.g., the role of critical periods), the specificity of language networks (e.g., the nature of any modularity) and their control (e.g., the domain-generality of such control). More recently, specific aspects of the field, namely the putative cognitive and neural consequences (often framed in the form of advantages) of bilingualism, have become a hotspot for controversy tied to the replication crisis in psychology. The critique of this research appears to be broad, addressing issues of power and sample size (e.g., Brysbaert, 2020, Nichols et al., 2020), failures to replicate (e.g., Paap & Greenberg, 2013), noise in samples and methods (e.g., García-Pentón et al., 2016a, García-Pentón et al., 2016b, Valian, 2015), and publication bias (e.g., de Bruin, Treccani, & Della Sala, 2015a; but see Bialystok, Kroll, Green, MacWhinney, & Craik, 2015), suggesting that the effects of bilingualism on cognitive and brain functioning are the result of questionable research practices. Consequently, several prescribed remedies, such as large samples (Brysbaert, 2020) and uniform¹ experimental procedures (García-Pentón et al., 2016a, García-Pentón et al., 2016b), have been marketed as solutions (see also Szucs & Ioannidis, 2020 for an example involving neuroscience more generally). However, such critiques and remedies, though well intended, often fail to place discussions in the broader context of science and its function throughout history. This raises the question of how the implementation of compulsory prescriptions would come to affect research on bilingualism more generally.

In this paper, we argue that the remedies and prescriptions put forward are deceptively simple and place us on a misleading path as they are based on a mischaracterization of the fundamentals of the scientific endeavor. While this paper is geared toward discussing current issues in research on bilingualism, we necessarily draw from the history of science to make the argument self-evident. Our position is that both large samples and conventionalized methods are important, but their role needs to be understood in the context of science as a discovery process, in which research findings are generated through interrelated iterations of exploration and falsification, which in turn lead to new insights and allow for the formulation of new questions. Fundamental to this process is the generation of variety² that permits incremental advance. The generation of variety serves two purposes: to identify reliable signals in the noise of our observations and to allow the formulation of effective theories and constructs about our world. Hypotheses, for instance, that the shape of the head is correlated with psychological traits (Simpson, 2005) or that bilingualism negatively impacts intelligence (Peal & Lambert, 1962), are discarded along the way. Constraints on the exploration of variety, such as those imposed by prescriptive remedies (e.g., keeping experimental designs as simple as possible), hinder the discovery process and so it is imperative in our view to ensure that methodological injunctions and publication practices are understood within the context of science as a discovery process.

The remainder of the paper is organized as follows: in the next section (Why a prescriptive science is problematic), we provide a critique on both practical and conceptual grounds of the rationale for power and uniformity prescriptions. Our line of argument then leads us to consider the implications for research practice in bilingualism (Articulating the research enterprise of bilingualism), where we emphasize the value of rich characterization of the sample, practices that enable the assessment of interactions rather than main effects on their own, and the application of sensitive tools. The implication is that without appropriate characterization, and without research practices and tools that lead to effective signal extraction, replication and large samples may be void of scientific interest. In both sections, we illustrate the manifestation of science as a discovery process with a range of past and contemporary examples drawn from research on bilingualism as well as from other fields. We necessarily draw on a range of examples, including those outside bilingualism, because these points are not unique to research on bilingualism; rather, they reflect a healthy and productive scientific enterprise. We do not argue against the importance of replication, the analytic value of Big Data, nor the application of sensitive and conventionalized research tools. Rather, we suggest that the application of method should be grounded in science as a discovery process.