How stable is second language aptitude? Effects of second language learning and language analysis training on second language aptitude test scores

Abstract

While second language (L2) aptitude has traditionally been considered a stable individual factor in SLA, more recent research and theory has questioned this assumption (Singleton, 2017). If L2 aptitude is stable, then the implication is that this set of cognitive abilities are genetic in origin and/or limited (Skehan, 1998). On the other hand, if L2 aptitude is not stable and is sensitive to experience, then it suggests that whatever constitutes L2 aptitude can be taught to make initial L2 learning easier and faster. While more recent studies have found that the L2 learning experience itself seems to have a training effect on L2 aptitude test scores (Ganschow, 1993; Ganschow & Sparks, 1995; Sáfár & Kormos, 2008; Sparks, Ganschow, Pohlman, Skinne, & Artzer, 1992; Sparks, Ganschow, Artzer, & Patton, 1997), both the size of such training effects and the effects of direct training of L2 aptitude abilities (Politzer & Weiss, 1969) remain unclear. The current study contributes to this debate by investigating the stability of L2 aptitude test scores. Conducted over an 8-week period of L2 instruction, the study investigated whether 6 weeks of aptitude training would enhance language analytic abilities, an area less researched in previous studies. Participants were 85 university students taking an introductory Spanish course, which included 6 weeks of instruction and practice on language analysis for learning L2 Spanish. The study operationalised L2 aptitude in terms of the abilities measured by the LLAMA tests: associative memory (LLAMA B), sound discrimination (LLAMA D), sound-symbol association (LLAMA E), and language analytic abilities (LLAMA F). Pre-testing took place at the beginning of L2 instruction and post-testing was conducted after the completion of the language analysis skills instruction. Stability was investigated through changes in L2 aptitude test scores. Results showed that post-test scores were significantly higher for all LLAMA tests except for the LLAMA E (sound-symbol association), which showed a ceiling effect in both pre- and post-tests. At the whole-group level, the size of gain scores differed across all tests with the greatest gains on the LLAMA B (associative memory) and the smallest for the LLAMA F (language analytic abilities). However, at the sub-group level, it was mainly participants with lower than average pre-test scores who achieved significant gains, with comparable effect sizes across all tests (except the LLAMA E). Trainability was investigated by comparing the predictive ability of training variables (e.g. accuracy scores and speed on training tasks) with L2 aptitude pre-test scores. Results for the predictive ability of training language analytic abilities were mixed on post-test scores. A random-forest regression (see Strobl, Malley, & Tutz, 2009) found that accuracy scores on the training were more predictive of post-test scores than LLAMA pre-test scores, but a step-wise linear regression did not confirm this finding. However, training variables were important predictors in both regression models. Overall, the findings suggest that L2 aptitude scores increase over time with L2 learning experience and instruction, in line with previous longitudinal studies (Sparks, Ganschow, Artzer, & Patton, 1997). This may imply that L2 aptitude is trainable, with some training variables being important predictors of post-test scores. The sensitivity of L2 aptitude measures to L2 learning experience and instruction may have implications for the validity of static, one-off measures of L2 aptitude for predicting L2 achievement.