Reevaluating trials to criterion as a measure in second language research

Abstract

Research on input processing and processing instruction has often employed a scoring method known as trials to criterion to observe the effects of instruction that emerge during training. Despite its common use in this research (see Fernández, 2021) this metric has never been evaluated critically. The present study first discusses several challenges associated with trials to criterion, including issues with its conceptual and methodological implementation. The study then introduces three alternative approaches for analyzing accuracy data collected during training sequences: trials to accuracy threshold, growth curve analysis, and bootstrapped differences of timeseries. For each approach, advantages and disadvantages are discussed and example analyses are presented using data from previous research. This discussion shows how these alternative approaches can supplement current trials-to-criterion-based analyses, expand the methodological choices available to researchers, and permit new and interesting research questions.