Validating L2 assessments for research purposes: How can insights from language testing inform SLA research

When second language acquisition (SLA) researchers design or adapt an assessment for research purposes, a key concern is to achieve validity, that is, to ensure that the assessment is suitable for measuring what it is intended to measure, and that it is appropriate for the populations or individuals participating in the research. While the issue of validation has been a central concern in the field of language testing since the inception of the field, considerably less attention has been paid to validation work in SLA. Despite repeated calls for attributing greater importance to validation (e.g., Chapelle, 1994, 1999; Douglas, 2001; Norris & Ortega, 2003, 2012; Shohamy, 2000), SLA validation practices often still remain insufficiently aligned with received standards for developing, using and evaluating research instruments (Purpura et al., 2015). In this talk, I will briefly review some key models that can be utilised in SLA research to ensure valid interpretations, including Norris and Ortega’s (2003) framework for designing and implementing research instruments and Kane’s (2006) argument-based approach to validation. Then, I will turn to a description and discussion of three issues in SLA validation work: how to proceed with validating theory-derived research instruments, how to validate research instruments for various uses, and how to plan validation into the design of SLA research studies. In discussing these topics, I will draw on existing work in SLA to showcase how successful validation may proceed. Finally, with the ultimate goal of improving validation practices within the field, I will end the presentation with some practical recommendations that may guide SLA researchers when constructing and validating research instruments.