A Case for Multisite Second Language Acquisition Research: Challenges, Risks, and Rewards

Abstract

Multisite research (MSR) offers the key advantages of greater statistical power and external validity via larger and more diverse participant pools. In second language acquisition (SLA) research, recent developments in meta‐analysis have created a robust foundation for MSR. Although logistical and financial obstacles can complicate expansion beyond a single site, we show that MSR's benefits can justify the investment of resources. We begin by outlining how developments in meta‐analytic research, replication, and access to data and materials have created an especially opportune moment for MSR. Next, we discuss the methodology for a classroom study on metacognitive instruction as an illustrative case, outlining the major elements of its design and implementation. Finally, we review how four issues critical to MSRs (funding, compliance, logistics, and analysis) were addressed in our example study, including a discussion on multilevel modeling. We conclude with a discussion of how the field can build upon existing research to advance multisite work.

Context and Generalizability in Multisite L2 Classroom Research: The Impact of Deductive Versus Guided Inductive Instruction

“…By examining this question across eight intact classrooms within three different schools, the study responds to increasing calls for more ecologically valid instructed SLA (ISLA) research that can help inform evidence‐based grammar pedagogy as implemented by real‐world practitioners (e.g., Kasprowicz & Marsden, 2018; Mitchell, 2000; Spada, 2015). This study also aligns with recent calls for multisite designs in the field of SLA (Moranski & Ziegler, 2021; Morgan‐Short et al., 2018), and more broadly within the social sciences (Simons, Holcombe, & Spellman, 2014), to increase the generalizability of effect‐of‐instruction findings across populations and learning contexts.…”

“…Our second research question focused on examining the variance inherent to the multisite research environment itself. Although discussions of ICCs and nesting effects have become common in larger educational research contexts (see Bickel, 2007), multisite work has traditionally been scarce within SLA and ISLA research (Moranski & Ziegler, 2021). Given that a growing number of recent empirical studies within the field have included multiple sites (e.g., Moranski & Ziegler, 2021; Morgan‐Short et al., 2018; Zalbidea, Issa, Faretta‐Stutenberg, & Sanz, 2021), the goal of our second research question was to examine how our nested data design impacted findings.…”

“…In light of recent calls for increased ecological validity (e.g., Kasprowicz & Marsden, 2018; Marsden & Kasprowicz, 2017; Mitchell, 2000) and expansion of multisite work in SLA (including ISLA) research (Moranski & Ziegler, 2021), this study provided a concrete example of the challenges, considerations, and benefits associated with scaling up single‐site studies. Specifically, it highlighted the need to account for nesting effects in multisite classroom research, particularly for production tasks, which were associated with higher ICCs.…”

Small Samples in Instructed Second Language Acquisition Research

“…The previous suggestions are for individual researchers, but there are also some solutions for small sample sizes that involve multiple researchers. For example, multisite research collaboration is gaining prominence in ISLA (Moranski & Ziegler, 2020) and other fields (Brysbaert, 2020; Button et al., 2013), with one advantage being the ability to obtain a larger sample. Researchers can employ the same research design in multiple contexts, thereby adding to the number of groups and participants in the study.…”

Foreign language speaking competence and self‐regulated speaking motivation

“…However, participants' self‐reports may not be reliable because of issues related to social desirability and self‐deception (Dörnyei, 2002). Second, the study did not benefit from the advantages of multisite research (see Moranski & Ziegler, 2021) as it involved a limited number of participants from one particular university. Therefore, it is difficult to generalize the findings to other contexts.…”