The particular need for replication in the quantitative study of SLA: A case study of the mnemonic effect of assonance in collocations

Abstract

Recent surveys of published reports of quasi-experimental studies of second language acquisition (SLA) indicate that low statistical power is pervasive owing in large part to small average sample sizes. The surveys do not indicate a marked trend toward samples that are larger. After illustrating the problem of low power in SLA research, we review arguments that increased replication of original studies can enable small-sample quantitative researchers to make firmer contributions to the field of SLA, especially if estimation of effect sizes and the practice of on-going statistical meta-analysis become routine. As a case study, we describe a series of small-sample quasi-experiments of which the first five found a short term positive mnemonic effect of interword, intra-phrase vowel repetition (or assonance) on learners’ retention of the forms of L2 collocations (e.g. strong bond vs. firm hold), whereas a sixth study newly reported here found negative effects. The case study illustrates the roles of replication and meta-analysis in successive re-adjustments of an original estimate. More specifically, the case study illustrates a meta-analytic approach to making sense of conflicting outcomes. All in all, it illustrates why small-sample researchers need to adopt a more long-term view.